Compare expected annual returns on Russian equities, future prices and dividends. Open the details to assess each forecast’s limitations.
Annual return includes dividends and a sale after three years. It is not the total-period return or a promise. The range limits show calculations under weaker and stronger assumptions.
Price upside excludes dividends. Dividends are per share over all three years; the percentage below is their share of the purchase price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 5.4x, three-year median 16.6x). Re-rating to 7.9x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, market, momentum, research
No flags: Revenue CAGR +23%
Caveat: Year-3 margin 17% is above the best year on record (14%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA -0.5x
No flags: Three-year dividends — 17% of the current price
No flags: IRR +51%
Caveat: The base carries no capex — it is set equal to depreciation
No flags: Base is fresh: report 1 months old
Caveat: Historical check (8 observations): median EBITDA miss 7.5 pp of margin, within ±5 pp in 12% of cases, relative error 68% at bias 0.32x, revenue 19%
No flags: Exit multiple and discounted cash flow agree (12,158 vs 13,567, 0.90x); in the DCF 85% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.12x, EBITDA 0.75x, profit 1.28x
Caveat: The debt build-up beyond cash flow cannot be measured: the group holds a bank or fintech and client inflows inflate operating cash flow
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2022), and 1 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
Caveat: One-year value 3,669 is an outlier: even the nearest independent estimate (eninvs Company_Forecast 9,946) is 0.4x away. Compared with: eninvs Company_Forecast, eninvs DividendPrediction
No flags: Year 1 within -15% of guidance
Year-1 margin from the company's EBITDA guidance: guided 10% vs 14% actual, applied 10% (delivery-record shrink 1.00)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.9x, three-year median 3.3x). Re-rating to 4.2x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +5%
Caveat: Year-3 margin 38% is above the best year on record (37%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 0.9x
No flags: IRR +48%
Caveat: The base carries no capex — it is set equal to depreciation
No flags: Base is fresh: report 4 months old
Caveat: Historical check (4 observations): median EBITDA miss 10.8 pp of margin, within ±5 pp in 25% of cases, relative error 23% at bias 0.77x, revenue 17%
No flags: Exit multiple and discounted cash flow agree (371 vs 513, 0.72x); in the DCF 79% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 0.96x, EBITDA 1.28x, profit 0.74x
No flags: Governance: quality multiplier 0.82 (flags: no_distributions). Penalties measured on distributions cut the model's dividend by 18%
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 0 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
Caveat: Cash conversion is not measured: of 0 reported years only 0 were comparable. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
Caveat: Leverage at entry: the market cap is 26% of enterprise value, so a move in business value is multiplied roughly 3.9x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
No flags: One-year value agrees with an independent estimate: Frontier спот-модель (1.38x). Compared with: eninvs Company_Forecast, Frontier спот-модель
The EBITDA move comes from the out-of-sample-validated branch of the Frontier spot model (×1.182 to the base) rather than my own price/cost split: this name's EI1 revenue map is stale and the branch is more accurate.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.8x, three-year median 3.0x). Re-rating to 4.1x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +9%
No flags: Year-3 margin 38% inside history 34%–39%
No flags: Year-3 net debt/EBITDA 2.0x
No flags: Three-year dividends — 52% of the current price
No flags: IRR +45%
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 35%
No flags: Base is fresh: report 4 months old
No flags: Historical check (13 observations): median EBITDA miss 2.0 pp of margin, within ±5 pp in 92% of cases, relative error 5% at bias 1.00x, revenue 3%
No flags: Exit multiple and discounted cash flow agree (103 vs 167, 0.61x); in the DCF 81% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (5 houses): revenue 1.00x, EBITDA 0.99x, profit 1.17x
Caveat: The name has two share classes: ordinary shares are 94% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.07x
Caveat: Net debt grows for reasons beyond a cash shortfall: over 5 years it added 6.6% of revenue a year on top of cash flow and dividends (range +0.6…+11.2%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Less reaches actual cash than the formula implies: over 3 reported years free cash flow came to 0.45 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
Caveat: Leverage at entry: the market cap is 16% of enterprise value, so a move in business value is multiplied roughly 6.3x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.83x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.53 and the three-year median 0.70. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 0.90 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 27%
No flags: Three-year dividends — 53% of the current price
No flags: IRR +43%
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 4 months old
Caveat: Historical check (8 observations): median profit miss 20.7 pp of revenue, within ±5 pp in 0% of cases, relative error 31% at bias 1.13x
No flags: Versus the analyst forecast for 2027 (7 houses): profit 1.09x
Caveat: The share count is growing: 2592m 3.0 years ago against 12928m now — 70.9% a year, last change 2026-04-29. If that pace holds, the value of a share in year 3 is 49% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: Governance: quality multiplier 0.47 (flags: dividend_decline, curated). Penalties measured on distributions cut the model's dividend by 15%
Caveat: The name has two share classes: ordinary shares are 59% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.70x
Caveat: The reporting-period series needed repair: 3 rows discarded as incompatible with the cumulative convention (years 2022, 2023), and 6 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.17x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 1.17%), not retained earnings; asset growth from смесь взглядов на темп бизнеса. Growth is capped by leverage: equity/assets stays above 4.5%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.98 and the three-year median 1.43. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 1.76 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 29%
No flags: Three-year dividends — 28% of the current price
No flags: IRR +40%
No flags: Minority interests are 13% of group equity; profit is taken net of them (96% of group) and so is equity. Perpetual subordinated instruments are excluded from equity — debt in substance
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 4 months old
Caveat: Historical check (9 observations): median profit miss 4.6 pp of revenue, within ±5 pp in 56% of cases, relative error 21% at bias 0.84x
No flags: Versus the analyst forecast for 2027 (7 houses): profit 1.00x
Caveat: The miss is one-sided: across 9 measurements the median profit forecast came to 0.84 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The share count is growing: 1993m 3.0 years ago against 2683m now — 10.4% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 26% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.08x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 2.86%), not retained earnings; asset growth from смесь взглядов на темп бизнеса. Growth is capped by leverage: equity/assets stays above 10.1%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.57 and the three-year median 0.97. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 1.12 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 20%
No flags: Three-year dividends — 22% of the current price
No flags: IRR +36%
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
Caveat: The reporting perimeter changed in 2022: full-year revenue 137,258 against 130,255 for Q3 of the same year — only 5% left for the rest of the year. That is a discontinued-operations reclassification with restated comparatives. Pre-break history is excluded and analyst forecasts may sit on a different perimeter
No flags: Base is fresh: report 4 months old
Caveat: Historical check (6 observations): median profit miss 10.8 pp of revenue, within ±5 pp in 0% of cases, relative error 38% at bias 1.38x
No flags: Versus the analyst forecast for 2027 (5 houses): profit 0.91x
Caveat: The miss is one-sided and in the dangerous direction: across 6 measurements the median profit forecast came to 1.38 of actual — the model consistently OVERSTATES this name. The IRR here is more a ceiling than a floor
Caveat: The share count is growing: 19044m 2.9 years ago against 22516m now — 6.0% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 16% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
No flags: Governance: quality multiplier 0.89 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 11%
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.26x). Compared with: eninvs Company_Forecast, Frontier див-модель
Profit runs off assets (RoA 1.81%), not retained earnings; asset growth from собственная история активов, медиана 3 темпов (+27%, +12%, +14%). Growth is capped by leverage: equity/assets stays above 7.9%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.8x, three-year median 2.1x). Re-rating to 3.1x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +15%
No flags: Year-3 margin 24% inside history -4%–26%
No flags: Year-3 net debt/EBITDA 1.3x
No flags: IRR +33%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (7 observations): median EBITDA miss 5.8 pp of margin, within ±5 pp in 43% of cases, relative error 32% at bias 0.86x, revenue 9%
Caveat: The two valuations of the same cash flow differ 0.3x: exit multiple 221 vs discounted cash flow 706 — the target rests on the multiple assumption rather than on the flow; in the DCF 78% of value sits in the terminal value
Caveat: Governance: quality multiplier 0.48 (flags: audit_qualification, curated). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 2 years it added 5.3% of revenue a year on top of cash flow and dividends (range +2.8…+7.8%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Cash conversion is not measured: of 3 reported years only 1 were comparable, and in 1 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
Caveat: Leverage at entry: the market cap is 20% of enterprise value, so a move in business value is multiplied roughly 4.9x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.91x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier спот-модель
The price move passes into revenue to the power of 0.98 (своя оценка), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 1.7x, three-year median 2.0x). Re-rating to 2.4x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +11%
Caveat: Year-3 margin 25% is above the best year on record (24%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 0.8x
No flags: Three-year dividends — 70% of the current price
No flags: IRR +33%
No flags: Base is fresh: report 4 months old
No flags: Historical check (17 observations): median EBITDA miss 2.7 pp of margin, within ±5 pp in 88% of cases, relative error 10% at bias 0.94x, revenue 4%
No flags: Exit multiple and discounted cash flow agree (1 vs 1, 0.90x); in the DCF 82% of value sits in the terminal value
Caveat: There was no free cash flow in any of the last 4 years: by our own formula it came out negative, and the cash-flow statement shows a median of 4,006 mln. The model still projects positive cash flow, drawn from EBITDA growth that historically covered only interest and capex. The forecast dividends and debt repayment rest on that assumption
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.85x). Compared with: eninvs Company_Forecast, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.75 and the three-year median 0.91. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 1.22 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 24%
No flags: Three-year dividends — 56% of the current price
No flags: IRR +32%
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 1 months old
No flags: Historical check (13 observations): median profit miss 4.7 pp of revenue, within ±5 pp in 54% of cases, relative error 13% at bias 1.01x
No flags: Versus the analyst forecast for 2027 (7 houses): profit 1.07x
Caveat: The name has two share classes: ordinary shares are 96% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.05x
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.95x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 2.62%), not retained earnings; asset growth from собственная история активов, медиана 3 темпов (+16%, +13%, +14%). Growth is capped by leverage: equity/assets stays above 11.2%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.76 and the three-year median 0.80. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 1.14 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 22%
No flags: Three-year dividends — 57% of the current price
No flags: IRR +31%
Caveat: Asset growth hit the capital limit in 3 of 3 years: the model does not let equity/assets fall below 6.7%, the bank's own historical minimum
No flags: Base is fresh: report 1 months old
Caveat: Historical check (9 observations): median profit miss 8.8 pp of revenue, within ±5 pp in 22% of cases, relative error 19% at bias 0.81x
Caveat: The miss is one-sided: across 9 measurements the median profit forecast came to 0.81 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: Frontier див-модель (1.02x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 1.94%), not retained earnings; asset growth from смесь взглядов на темп бизнеса. Growth is capped by leverage: equity/assets stays above 6.7%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.8x, three-year median 2.8x). Re-rating to 4.1x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +14%
No flags: Year-3 margin 10% inside history 8%–10%
No flags: Year-3 net debt/EBITDA 0.1x
No flags: IRR +31%
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 0.9 pp of margin, within ±5 pp in 100% of cases, relative error 9% at bias 0.99x, revenue 9%
No flags: Exit multiple and discounted cash flow agree (4,164 vs 7,522, 0.55x); in the DCF 80% of value sits in the terminal value
No flags: Governance: quality multiplier 0.82 (flags: no_distributions). Penalties measured on distributions cut the model's dividend by 18%
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 2.7% of revenue a year on top of cash flow and dividends (range +0.7…+13.1%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.17 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.17x — never above 1.3, which would extrapolate a good year
Caveat: One-year value 2,617 is an outlier: even the nearest independent estimate (eninvs Company_Forecast 540) is 4.8x away. Compared with: eninvs Company_Forecast
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 7.1x, three-year median 7.0x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, momentum
No flags: Revenue CAGR +15%
No flags: Year-3 margin 31% inside history 29%–34%
No flags: Year-3 net debt/EBITDA -0.8x
No flags: Three-year dividends — 16% of the current price
No flags: IRR +27%
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 2.3 pp of margin, within ±5 pp in 67% of cases, relative error 7% at bias 0.97x, revenue 10%
No flags: Exit multiple and discounted cash flow agree (2,383 vs 2,034, 1.17x); in the DCF 80% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.03x, EBITDA 1.03x, profit 0.72x
No flags: Governance: quality multiplier 0.86 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 14%
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.24 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.24x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: Frontier див-модель (1.35x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
No flags: Year 1 within -0% of guidance
Year-1 margin from the company's EBITDA margin guidance: guided 30% vs 29% actual, applied 30% (delivery-record shrink 1.00)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 9.1x, three-year median 10.8x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, momentum
No flags: Revenue CAGR +12%
No flags: Year-3 margin 20% inside history 14%–42%
No flags: Year-3 net debt/EBITDA 0.2x
No flags: Three-year dividends — 24% of the current price
No flags: IRR +25%
No flags: Base is fresh: report 7 months old
Caveat: Model year 1 is already 58% over: the base report is 7 months old. We have no reporting for that stretch — the model leans on external series alone
No flags: Historical check (7 observations): median EBITDA miss 4.4 pp of margin, within ±5 pp in 57% of cases, relative error 11% at bias 1.03x, revenue 5%
No flags: Exit multiple and discounted cash flow agree (1,991 vs 1,308, 1.52x); in the DCF 78% of value sits in the terminal value
Caveat: Year-1 EBITDA is 1.64x the last reported figure on revenue growth of 1.15x — so 1.42x of the jump is MARGIN alone (13.7% actual against 19.5% forecast). That is an assumption, not a measurement
Caveat: Net debt grows for reasons beyond a cash shortfall: over 2 years it added 4.5% of revenue a year on top of cash flow and dividends (range +2.2…+6.8%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.83x). Compared with: eninvs Company_Forecast, Frontier див-модель
Caveat: Model year 1 differs from guidance by -15%
Year-1 margin from the company's EBITDA margin guidance: guided 21% vs 14% actual, applied 20% (delivery-record shrink 0.80)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 5.8x, three-year median 12.0x). Re-rating to 8.5x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +18%
No flags: Year-3 margin 40% inside history 40%–56%
No flags: Year-3 net debt/EBITDA 0.6x
No flags: Three-year dividends — 17% of the current price
No flags: IRR +22%
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 3.6 pp of margin, within ±5 pp in 67% of cases, relative error 9% at bias 0.94x, revenue 24%
No flags: Exit multiple and discounted cash flow agree (354 vs 190, 1.86x); in the DCF 80% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 0.94x, EBITDA 0.95x, profit 0.74x
No flags: Governance: quality multiplier 0.90 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 10%
Caveat: Net debt grows for reasons beyond a cash shortfall: over 3 years it added 8.0% of revenue a year on top of cash flow and dividends (range +3.2…+11.1%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2022). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: One-year value 275: nearest independent is eninvs Company_Forecast (0.50x), median 312. Compared with: eninvs Company_Forecast, Frontier див-модель
The company guides net-profit growth of +40% for 2026. It does not drive the model path — profit and EBITDA move differently through interest and tax — it is shown as a cross-check.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 3.2x, three-year median 4.0x). Re-rating to 4.7x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +8%
No flags: Year-3 margin 35% inside history 35%–43%
No flags: Year-3 net debt/EBITDA 2.1x
No flags: Dividends follow the company's announced capital-return policy rather than a share of profit: up to 20% of OIBDA a year with a floor of 70bn, of which 70% is paid out and 30% is spent buying back shares
No flags: Three-year dividends — 40% of the current price
No flags: IRR +20%
No flags: Base is fresh: report 4 months old
No flags: Historical check (10 observations): median EBITDA miss 2.3 pp of margin, within ±5 pp in 80% of cases, relative error 6% at bias 1.04x, revenue 3%
No flags: Exit multiple and discounted cash flow agree (227 vs 158, 1.44x); in the DCF 85% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (8 houses): revenue 1.02x, EBITDA 1.02x, profit 1.47x
Caveat: The model matches the houses on revenue and EBITDA but differs 1.47x on profit. For a leveraged name that is almost always the rate path: the model runs interest down with the key rate, the street is usually more cautious
Caveat: The share count is growing: 1699m 3.0 years ago against 1881m now — 3.5% a year, last change 2026-01-25. If that pace holds, the value of a share in year 3 is 10% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 4.9 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: The debt build-up beyond cash flow cannot be measured: the group holds a bank or fintech and client inflows inflate operating cash flow
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2018, 2022), and 2 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.91x). Compared with: eninvs Company_Forecast, Frontier див-модель
Dividend policy from the primary source: Dividends are no longer a share of profit. Capital return = max(20% of OIBDA; RUB 70bn) a year, 70% in cash and 30% in buybacks. Our EBITDA equals the company's OIBDA on this name (Q2 2026: 84,614mn vs 84.6bn reported). From 17.09.2026 the base case follows the release literally: repurchased shares are NOT retired (buyback_retires=false), so the 30% leaves the company in cash while the share count stays put. Retirement is upside to the base, not the base.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 8.7x, three-year median 8.7x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +11%
No flags: Year-3 margin 30% inside history 21%–105%
No flags: Year-3 net debt/EBITDA -0.9x
No flags: Three-year dividends — 17% of the current price
No flags: IRR +20%
Caveat: The base carries no capex — it is set equal to depreciation
No flags: Base is fresh: report 4 months old
Caveat: Historical check (4 observations): median EBITDA miss 7.0 pp of margin, within ±5 pp in 25% of cases, relative error 27% at bias 0.81x, revenue 3%
No flags: Exit multiple and discounted cash flow agree (1,794 vs 1,260, 1.42x); in the DCF 77% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.19x, EBITDA 0.95x, profit 0.66x
Caveat: The model matches the houses on revenue and EBITDA but differs 0.66x on profit. For a leveraged name that is almost always the rate path: the model runs interest down with the key rate, the street is usually more cautious
Caveat: The formula undercounts cash (conversion 2.23) but the evidence is thin: 2 usable years, excess in 2. No upward correction is applied — the forecast cash flow and dividend are understated
No flags: One-year value agrees with an independent estimate: Frontier спот-модель (1.16x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 667 -> 673 mmboe за 2024-2025 (+0.9% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 71% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.8x, three-year median 1.8x). Re-rating to 2.7x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +7%
Caveat: Year-3 margin 33% is above the best year on record (33%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA -0.6x
No flags: Three-year dividends — 58% of the current price
No flags: IRR +20%
Caveat: The reporting perimeter changed in 2024: full-year revenue 4,420,565 against 4,333,568 for H1 of the same year — only 2% left for the rest of the year. That is a discontinued-operations reclassification with restated comparatives. Pre-break history is excluded and analyst forecasts may sit on a different perimeter
No flags: Base is fresh: report 4 months old
Caveat: Too little history to check (0 observations) — forecast accuracy unmeasured
Caveat: The two valuations of the same cash flow differ 0.4x: exit multiple 6,448 vs discounted cash flow 15,248 — the target rests on the multiple assumption rather than on the flow; in the DCF 77% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.07x, EBITDA 1.29x, profit 1.53x
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 5 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
No flags: The formula undercounts cash: over 3 reported years free cash flow came to 1.72 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 3 years. The forecast is scaled up 1.30x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.97x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 681 -> 666 mmboe за 2024-2025 (-2.2% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Dividend policy from the primary source: LUKOIL pays out of adjusted free cash flow, not profit; the realised payout of the reported base is 82-87%, hence 0.85 in the model.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 4.2x, three-year median 10.7x). Re-rating to 6.1x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, market, momentum
No flags: Revenue CAGR +13%
Caveat: Year-3 margin 23% is above the best year on record (22%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 0.5x
No flags: Three-year dividends — 24% of the current price
No flags: IRR +19%
No flags: Base is fresh: report 1 months old
Caveat: Too little history to check (3 observations) — forecast accuracy unmeasured
No flags: Exit multiple and discounted cash flow agree (4,994 vs 4,117, 1.21x); in the DCF 83% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (9 houses): revenue 0.93x, EBITDA 0.79x, profit 0.70x
Caveat: The share count is growing: 327m 3.0 years ago against 397m now — 6.6% a year, last change 2026-09-16. If that pace holds, the value of a share in year 3 is 18% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: Net debt grows for reasons beyond a cash shortfall: over 2 years it added 3.9% of revenue a year on top of cash flow and dividends (range +2.6…+5.1%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2023, 2024), and 2 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: Cash conversion is not measured: of 3 reported years only 1 were comparable, and in 1 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
No flags: One-year value agrees with an independent estimate: Frontier див-модель (1.02x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
No flags: Year 1 within -2% of guidance
Year-1 margin from the company's EBITDA guidance: guided 18% vs 22% actual, applied 18% (delivery-record shrink 1.00)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 1.6x, three-year median 1.4x). Re-rating to 2.1x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +14%
Caveat: Year-3 margin 19% is above the best year on record (19%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 0.1x
No flags: Three-year dividends — 40% of the current price
No flags: IRR +19%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (18 observations): median EBITDA miss 3.0 pp of margin, within ±5 pp in 72% of cases, relative error 22% at bias 0.82x, revenue 9%
No flags: Exit multiple and discounted cash flow agree (0 vs 0, 1.22x); in the DCF 80% of value sits in the terminal value
Caveat: The miss is one-sided: across 18 measurements the median EBITDA forecast came to 0.82 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: Cash conversion is not measured: of 4 reported years only 1 were comparable, and in 2 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.02x). Compared with: eninvs Company_Forecast, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 3.9x, three-year median 3.8x). Re-rating to 5.5x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +5%
No flags: Year-3 margin 52% inside history 51%–54%
No flags: Year-3 net debt/EBITDA -0.3x
No flags: Three-year dividends — 46% of the current price
No flags: IRR +18%
No flags: Base is fresh: report 4 months old
Caveat: Too little history to check (2 observations) — forecast accuracy unmeasured
No flags: Exit multiple and discounted cash flow agree (140 vs 245, 0.57x); in the DCF 78% of value sits in the terminal value
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 2 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.76x). Compared with: Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 3.6x, three-year median 9.6x). Re-rating to 5.3x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +3%
Caveat: Year-3 margin 60% is below the worst year on record (63%), though inside tolerance
No flags: Year-3 net debt/EBITDA -0.8x
No flags: Three-year dividends — 18% of the current price
No flags: IRR +17%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (5 observations): median EBITDA miss 16.9 pp of margin, within ±5 pp in 20% of cases, relative error 27% at bias 1.12x, revenue 23%
No flags: Exit multiple and discounted cash flow agree (91 vs 142, 0.64x); in the DCF 79% of value sits in the terminal value
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 3 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.08 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.08x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: Frontier див-модель (1.11x). Compared with: eninvs Company_Forecast, Frontier див-модель
The model historically overstated this name's EBITDA by 1.12x (5 observations), so the path is trimmed by 11%.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 1.2x, three-year median 1.4x). Re-rating to 1.8x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +11%
No flags: Year-3 margin 25% inside history 21%–29%
No flags: Year-3 net debt/EBITDA 0.1x
No flags: Three-year dividends — 49% of the current price
No flags: IRR +17%
No flags: Base is fresh: report 4 months old
No flags: Historical check (18 observations): median EBITDA miss 1.8 pp of margin, within ±5 pp in 89% of cases, relative error 7% at bias 0.96x, revenue 4%
No flags: Exit multiple and discounted cash flow agree (1 vs 1, 1.11x); in the DCF 71% of value sits in the terminal value
Caveat: Cash conversion is not measured: of 4 reported years only 1 were comparable, and in 1 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.79x). Compared with: eninvs Company_Forecast, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 2.9x, three-year median 3.2x). Re-rating to 4.2x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +4%
No flags: Year-3 margin 68% inside history 62%–73%
No flags: Year-3 net debt/EBITDA -0.2x
No flags: Three-year dividends — 29% of the current price
No flags: IRR +17%
No flags: Base is fresh: report 1 months old
No flags: Historical check (6 observations): median EBITDA miss 4.5 pp of margin, within ±5 pp in 67% of cases, relative error 7% at bias 1.01x, revenue 8%
No flags: Exit multiple and discounted cash flow agree (9 vs 8, 1.03x); in the DCF 79% of value sits in the terminal value
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 3 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
No flags: One-year value agrees with an independent estimate: eninvs DividendPrediction (0.78x). Compared with: eninvs DividendPrediction, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 4.7x, three-year median 5.9x). Re-rating to 6.9x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +13%
No flags: Year-3 margin 38% inside history 33%–39%
No flags: Year-3 net debt/EBITDA 0.4x
No flags: Three-year dividends — 11% of the current price
No flags: IRR +15%
No flags: Base is fresh: report 4 months old
No flags: Historical check (6 observations): median EBITDA miss 3.6 pp of margin, within ±5 pp in 50% of cases, relative error 9% at bias 0.91x, revenue 6%
Caveat: The two valuations of the same cash flow differ 2.2x: exit multiple 66 vs discounted cash flow 30 — the target rests on the multiple assumption rather than on the flow; in the DCF 79% of value sits in the terminal value
Caveat: The share count is growing: 1000m 2.0 years ago against 1168m now — 7.9% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 20% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: Net debt grows for reasons beyond a cash shortfall: over 3 years it added 3.7% of revenue a year on top of cash flow and dividends (range -8.0…+11.1%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Less reaches actual cash than the formula implies: over 4 reported years free cash flow came to 0.45 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.74x). Compared with: eninvs Company_Forecast, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 3.1x, three-year median 4.3x). Re-rating to 4.5x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +10%
No flags: Year-3 margin 14% inside history 14%–18%
No flags: Year-3 net debt/EBITDA 2.2x
No flags: Three-year dividends — 46% of the current price
No flags: IRR +14%
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 37%
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 2.2 pp of margin, within ±5 pp in 89% of cases, relative error 14% at bias 1.01x, revenue 5%
Caveat: The two valuations of the same cash flow differ 0.2x: exit multiple 265 vs discounted cash flow 1,075 — the target rests on the multiple assumption rather than on the flow; in the DCF 80% of value sits in the terminal value
Caveat: Our gap to the houses is the lease basis, not the forecast. Profit matches (1.01x), EBITDA is 1.31x, and our net debt is 37 912 higher against an EBITDA gap of 5 914 a year — 6.4 years of payments. That is IFRS 16 versus IAS 17: we keep leases in debt and in D&A, the street takes rent out of EBITDA. Below EBITDA the two bases agree, so the EBITDA gap says nothing about forecast quality
No flags: Versus the analyst forecast for 2027 (3 houses): revenue 0.94x, EBITDA 1.31x, profit 1.01x — the EBITDA gap is explained by the lease basis
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 4.4 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
No flags: Governance: quality multiplier 0.91 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 9%
Caveat: Net debt grows for reasons beyond a cash shortfall: over 5 years it added 6.7% of revenue a year on top of cash flow and dividends (range +0.3…+11.3%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
No flags: One-year value agrees with an independent estimate: Frontier див-модель (1.44x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 11.8x, three-year median 16.3x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +15%
Caveat: Year-3 margin 29% is above the best year on record (26%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA -0.2x
No flags: Three-year dividends — 28% of the current price
No flags: IRR +12%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (6 observations): median EBITDA miss 4.8 pp of margin, within ±5 pp in 50% of cases, relative error 23% at bias 0.88x, revenue 10%
No flags: Exit multiple and discounted cash flow agree (759 vs 864, 0.88x); in the DCF 79% of value sits in the terminal value
Caveat: The share count is growing: 70m 3.0 years ago against 78m now — 3.5% a year, last change 2025-11-22. If that pace holds, the value of a share in year 3 is 10% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: Net debt grows for reasons beyond a cash shortfall: over 3 years it added 8.7% of revenue a year on top of cash flow and dividends (range +8.1…+10.6%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.38 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.30x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: eninvs DividendPrediction (0.81x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 4.2x, three-year median 3.9x). Re-rating to 5.6x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +7%
No flags: Year-3 margin 30% inside history 18%–31%
No flags: Year-3 net debt/EBITDA 1.7x
No flags: Three-year dividends — 27% of the current price
No flags: IRR +12%
No flags: Base is fresh: report 4 months old
No flags: Historical check (6 observations): median EBITDA miss 1.5 pp of margin, within ±5 pp in 67% of cases, relative error 6% at bias 0.99x, revenue 4%
No flags: Exit multiple and discounted cash flow agree (622 vs 550, 1.13x); in the DCF 78% of value sits in the terminal value
Caveat: Our gap to the houses is the lease basis, not the forecast. Profit matches (0.97x), EBITDA is 1.20x, and our net debt is 828 272 higher against an EBITDA gap of 217 891 a year — 3.8 years of payments. That is IFRS 16 versus IAS 17: we keep leases in debt and in D&A, the street takes rent out of EBITDA. Below EBITDA the two bases agree, so the EBITDA gap says nothing about forecast quality
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.09x, EBITDA 1.20x, profit 0.97x — the EBITDA gap is explained by the lease basis
No flags: Governance: quality multiplier 0.91 (flags: controlled_by). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 3 years it added 3.2% of revenue a year on top of cash flow and dividends (range -7.3…+4.6%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The formula undercounts cash (conversion 1.12) but the evidence is thin: 2 usable years, excess in 2. No upward correction is applied — the forecast cash flow and dividend are understated
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.97x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
The price move passes into revenue to the power of 0.40 (своя оценка), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Exit valuation basis: No re-rating: min(current 1.1x, three-year median 1.4x). Re-rating to 1.7x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +5%
Caveat: Year-3 margin 28% is below the worst year on record (33%), though inside tolerance
No flags: Year-3 net debt/EBITDA -0.1x
No flags: Three-year dividends — 39% of the current price
No flags: IRR +6%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (10 observations): median EBITDA miss 8.0 pp of margin, within ±5 pp in 40% of cases, relative error 22% at bias 1.22x, revenue 5%
Caveat: The two valuations of the same cash flow differ 4.0x: exit multiple 824 vs discounted cash flow 204 — the target rests on the multiple assumption rather than on the flow; in the DCF 72% of value sits in the terminal value
No flags: Versus the analyst forecast for 2027 (6 houses): revenue 1.00x, EBITDA 0.69x, profit 0.56x
Caveat: The miss is one-sided and in the dangerous direction: across 10 measurements the median EBITDA forecast came to 1.22 of actual — the model consistently OVERSTATES this name. The IRR here is more a ceiling than a floor
Caveat: Less reaches actual cash than the formula implies: over 4 reported years free cash flow came to 0.79 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
Caveat: One-year value 911: nearest independent is eninvs Company_Forecast (0.55x), median 1,821. Compared with: eninvs Company_Forecast, Frontier див-модель
The model historically overstated this name's EBITDA by 1.22x (10 observations), so the path is trimmed by 18%.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (12,866) differs from the last reported (30,630) by a factor of 0.42. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 3.3x, three-year median 3.5x). Re-rating to 4.8x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +7%
No flags: Year-3 margin 21% inside history 9%–52%
No flags: Year-3 net debt/EBITDA 2.1x
No flags: Three-year dividends — 15% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
Caveat: Too little history to check (0 observations) — forecast accuracy unmeasured
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The share count is growing: 2445m 3.0 years ago against 3976m now — 17.6% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 39% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: The reporting-period series needed repair: 8 rows discarded as incompatible with the cumulative convention (years 2018, 2019, 2021, 2022, 2023, 2024, 2025), and 8 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: The formula undercounts cash (conversion 1.43) but the evidence is thin: 2 usable years, excess in 1. No upward correction is applied — the forecast cash flow and dividend are understated
Caveat: Leverage at entry: the market cap is 23% of enterprise value, so a move in business value is multiplied roughly 4.3x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
No flags: One-year value agrees with an independent estimate: eninvs DividendPrediction (1.08x). Compared with: eninvs DividendPrediction, Frontier див-модель
Error: First-year profit (12,866) differs from the last reported (30,630) by a factor of 0.42. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Exit valuation basis: No re-rating: min(current 7.8x, three-year median 6.5x). Re-rating to 9.4x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR -4%
Caveat: Year-3 margin 9% is below the worst year on record (18%), though inside tolerance
Caveat: Year-3 net debt/EBITDA 16.0x — the model does not carry the debt
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Error: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Caveat: The reporting perimeter changed in 2024: full-year revenue 239,067 against 179,468 for H1 of the same year — only 25% left for the rest of the year. That is a discontinued-operations reclassification with restated comparatives. Pre-break history is excluded and analyst forecasts may sit on a different perimeter
No flags: Base is fresh: report 4 months old
Caveat: Too little history to check (0 observations) — forecast accuracy unmeasured
No flags: Governance: quality multiplier 0.90 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 10%
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 4 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2020), and 8 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: There was no free cash flow in any of the last 2 years: by our own formula it came out negative, and the cash-flow statement shows a median of -13,617 mln. The model still projects positive cash flow, drawn from EBITDA growth that historically covered only interest and capex. The forecast dividends and debt repayment rest on that assumption
Error: The model for this name is unvalidated (0 measurements against the four required) and the name carries 2 structural flaws: perimeter change, no free cash flow ever. Either alone would be tolerable; together they are not — the number is withheld
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 33 -> 29.7 Mct за 2024-2025 (-10.0% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 40% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (1,725) differs from the last reported (4,801) by a factor of 0.36. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 6.9x, three-year median 7.6x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +2%
Caveat: Year-3 margin 18% is below the worst year on record (23%), though inside tolerance
No flags: Year-3 net debt/EBITDA 3.5x
No flags: Three-year dividends — 1% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
Caveat: Historical check (9 observations): median EBITDA miss 5.7 pp of margin, within ±5 pp in 44% of cases, relative error 14% at bias 0.95x, revenue 26%
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 2.2% of revenue a year on top of cash flow and dividends (range +0.8…+6.6%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2018). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: Less reaches actual cash than the formula implies: over 2 reported years free cash flow came to 0.54 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.73x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
Error: First-year profit (1,725) differs from the last reported (4,801) by a factor of 0.36. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 30% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Forecast withheldThe first forecast year has elapsed
Price used in model
1 303 ₽
As-of date of this price
not supplied
Model calculated at
2026-10-29T19:37:01
Reporting basis
2025-12-31
Exit valuation multiple
2.20x
EV/EBITDA
One-year EBITDA forecast error / observations
17% / 2
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Model year 1 is already 82% over: the base report is 10 months old and year 1 ends twelve months after it. Forecasting a period that has all but happened, with no interim report, is not defensible
Exit valuation basis: No re-rating: min(current 2.3x, three-year median 2.2x). Re-rating to 3.2x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +7%
No flags: Year-3 margin 16% inside history 6%–24%
No flags: Year-3 net debt/EBITDA 0.5x
No flags: Three-year dividends — 25% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 10 months old
Error: Model year 1 is already 82% over: the base report is 10 months old and year 1 ends twelve months after it. Forecasting a period that has all but happened, with no interim report, is not defensible
Caveat: Too little history to check (2 observations) — forecast accuracy unmeasured
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Governance: quality multiplier 0.70 (flags: curated). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: The name has two share classes: ordinary shares are 89% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.12x
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 4.7% of revenue a year on top of cash flow and dividends (range +4.6…+8.9%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2021), and 8 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.18x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
The price move passes into revenue to the power of 0.44 (своя оценка), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (7 observations): median profit miss 25.0 pp of revenue, within ±5 pp in 0% of cases, relative error 43% at bias 1.23x — poor on both scales, the model's number is withheld
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.52 and the three-year median 0.72. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 0.91 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 23%
No flags: Three-year dividends — 54% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 4 months old
Error: Historical check (7 observations): median profit miss 25.0 pp of revenue, within ±5 pp in 0% of cases, relative error 43% at bias 1.23x — poor on both scales, the model's number is withheld
Caveat: The miss is one-sided and in the dangerous direction: across 7 measurements the median profit forecast came to 1.23 of actual — the model consistently OVERSTATES this name. The IRR here is more a ceiling than a floor
No flags: Governance: quality multiplier 0.90 (flags: no_cash_flow). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.04x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 2.19%), not retained earnings; asset growth from собственная история активов, медиана 3 темпов (+8%, +15%, +8%). Growth is capped by leverage: equity/assets stays above 11.2%, the bank's own historical minimum.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Exit valuation basis: No re-rating: min(current 6.4x, three-year median 7.5x). Re-rating to 9.3x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +1%
Caveat: Year-3 margin 8% is below the worst year on record (12%), though inside tolerance
Caveat: Year-3 net debt/EBITDA 10.9x — the model does not carry the debt
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Error: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
No flags: Base is fresh: report 4 months old
Caveat: Historical check (7 observations): median EBITDA miss 6.7 pp of margin, within ±5 pp in 29% of cases, relative error 44% at bias 0.62x, revenue 5%
Caveat: The miss is one-sided: across 7 measurements the median EBITDA forecast came to 0.62 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 5.1 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
No flags: Governance: quality multiplier 0.82 (flags: no_distributions). Penalties measured on distributions cut the model's dividend by 18%
Caveat: The reporting-period series needed repair: 1 rows discarded as incompatible with the cumulative convention (years 2018). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: There was no free cash flow in any of the last 3 years: by our own formula it came out negative, and the cash-flow statement shows a median of 13,949 mln. The model still projects positive cash flow, drawn from EBITDA growth that historically covered only interest and capex. The forecast dividends and debt repayment rest on that assumption
Caveat: Leverage at entry: the market cap is 13% of enterprise value, so a move in business value is multiplied roughly 7.5x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 30% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Exit valuation basis: No re-rating: min(current 2.0x, three-year median 2.1x). Re-rating to 2.9x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +6%
Caveat: Year-3 margin 9% is below the worst year on record (12%), though inside tolerance
No flags: Year-3 net debt/EBITDA 2.3x
No flags: Three-year dividends — 42% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Error: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
No flags: Base is fresh: report 4 months old
No flags: Historical check (5 observations): median EBITDA miss 2.2 pp of margin, within ±5 pp in 100% of cases, relative error 18% at bias 1.18x, revenue 4%
Caveat: The miss is one-sided and in the dangerous direction: across 5 measurements the median EBITDA forecast came to 1.18 of actual — the model consistently OVERSTATES this name. The IRR here is more a ceiling than a floor
No flags: Governance: quality multiplier 0.80 (flags: dividend_decline). Penalties measured on distributions cut the model's dividend by 20%
Caveat: Net debt grows for reasons beyond a cash shortfall: over 2 years it added 5.9% of revenue a year on top of cash flow and dividends (range +3.0…+8.9%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.36 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.30x — never above 1.3, which would extrapolate a good year
Caveat: One-year value 0 is an outlier: even the nearest independent estimate (Frontier див-модель 0) is 0.3x away. Compared with: eninvs Company_Forecast, Frontier див-модель
The model historically overstated this name's EBITDA by 1.18x (5 observations), so the path is trimmed by 15%.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (7 observations): median EBITDA miss 29.7 pp of margin, within ±5 pp in 0% of cases, relative error 68% at bias 1.68x, revenue 28% — poor on both scales, the model's number is withheld
Exit valuation basis: No re-rating: min(current 4.9x, three-year median 2.6x). Re-rating to 3.7x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +28%
Caveat: Year-3 margin 37% is below the worst year on record (40%), though inside tolerance
No flags: Year-3 net debt/EBITDA 0.2x
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The base carries no capex — it is set equal to depreciation
No flags: Base is fresh: report 4 months old
Error: Historical check (7 observations): median EBITDA miss 29.7 pp of margin, within ±5 pp in 0% of cases, relative error 68% at bias 1.68x, revenue 28% — poor on both scales, the model's number is withheld
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The miss is one-sided and in the dangerous direction: across 7 measurements the median EBITDA forecast came to 1.68 of actual — the model consistently OVERSTATES this name. The IRR here is more a ceiling than a floor
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 4 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2018, 2021), and 3 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
No flags: The formula undercounts cash: over 3 reported years free cash flow came to 1.26 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 3 years. The forecast is scaled up 1.26x — never above 1.3, which would extrapolate a good year
Caveat: One-year value 85 is an outlier: even the nearest independent estimate (eninvs Company_Forecast 298) is 0.3x away. Compared with: eninvs Company_Forecast
Error: Revenue driver map is unconfirmed: revenue elasticity to the TCE benchmark is unidentified: OLS 0.42 with R² 0.17 over 10 quarters, USD revenue per unit of benchmark fell from ~15 (2024) to ~3 (1H2026) — a sanctioned fleet earns an unknown discount
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 30% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
The model historically overstated this name's EBITDA by 1.68x (7 observations), so the path is trimmed by 41%.
The analyst forecast for this name is in USD while the model runs in RUB — the line-by-line comparison is switched off.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Exit valuation basis: The 1.5x exit is the model's FLOOR, not today's valuation: the current multiple is 0.40x, so a 4.3x multiple normalisation is assumed. Enterprise value is near zero — the cash on the balance sheet matches what the shares cost — so the upside comes from re-rating, not the business
Model checks
No flags: Growth sources: consensus, momentum
No flags: Revenue CAGR +9%
Caveat: Year-3 margin 7% is below the worst year on record (7%), though inside tolerance
No flags: Year-3 net debt/EBITDA 2.4x
No flags: Three-year dividends — 32% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Error: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 1.5 pp of margin, within ±5 pp in 100% of cases, relative error 16% at bias 1.12x, revenue 2%
No flags: Versus the analyst forecast for 2027 (4 houses): revenue 1.01x, EBITDA 0.81x, profit 0.72x
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 3.3 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: Net debt grows for reasons beyond a cash shortfall: over 5 years it added 3.5% of revenue a year on top of cash flow and dividends (range -4.6…+7.5%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Less reaches actual cash than the formula implies: over 3 reported years free cash flow came to 0.75 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.89x). Compared with: Frontier див-модель
The model historically overstated this name's EBITDA by 1.12x (9 observations), so the path is trimmed by 10%.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (10 observations): median profit miss 9.4 pp of revenue, within ±5 pp in 30% of cases, relative error 61% at bias 0.93x — poor on both scales, the model's number is withheld
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.34 and the three-year median 0.49. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 0.61 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 17%
No flags: Three-year dividends — 45% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 4 months old
Error: Historical check (10 observations): median profit miss 9.4 pp of revenue, within ±5 pp in 30% of cases, relative error 61% at bias 0.93x — poor on both scales, the model's number is withheld
Caveat: The share count is growing: 35m 2.5 years ago against 38m now — 3.1% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 9% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
No flags: Governance: quality multiplier 0.90 (flags: no_cash_flow). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2019). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.08x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Profit runs off assets (RoA 2.63%), not retained earnings; asset growth from смесь взглядов на темп бизнеса. Growth is capped by leverage: equity/assets stays above 12.4%, the bank's own historical minimum.
Forecast withheldProfit forecasts were historically overstated
Price used in model
1 808 ₽
As-of date of this price
not supplied
Model calculated at
2026-10-29T09:37:03
Reporting basis
2026-06-30
Exit valuation multiple
2.02x
EV/EBITDA
One-year EBITDA forecast error / observations
12% / 9
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Across 6 historical measurements the profit forecast came to a median 1.76 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
Exit valuation basis: No re-rating: min(current 2.0x, three-year median 2.2x). Re-rating to 3.0x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +9%
Caveat: Year-3 margin 8% is below the worst year on record (9%), though inside tolerance
No flags: Year-3 net debt/EBITDA 2.0x
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 354%
No flags: Base is fresh: report 4 months old
No flags: Historical check (9 observations): median EBITDA miss 1.1 pp of margin, within ±5 pp in 100% of cases, relative error 12% at bias 1.09x, revenue 7%
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 5.7 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: Governance: quality multiplier 0.70 (flags: curated). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 5 years it added 3.0% of revenue a year on top of cash flow and dividends (range +1.9…+7.0%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Less reaches actual cash than the formula implies: over 3 reported years free cash flow came to 0.60 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
Caveat: Leverage at entry: the market cap is 20% of enterprise value, so a move in business value is multiplied roughly 5.0x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
Error: Across 6 historical measurements the profit forecast came to a median 1.76 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (18 observations): median profit miss 14.8 pp of revenue, within ±5 pp in 11% of cases, relative error 53% at bias 0.77x — poor on both scales, the model's number is withheld
Exit valuation basis: Partial re-rating: the mean of today's P/B 1.46 and the three-year median 1.71. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 2.32 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 26%
No flags: Three-year dividends — 43% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: No balance sheet to drive the model: profit runs off equity, so growth is hard-capped at ROE x (1 − payout) and understates a growing bank
No flags: Base is fresh: report 4 months old
Error: Historical check (18 observations): median profit miss 14.8 pp of revenue, within ±5 pp in 11% of cases, relative error 53% at bias 0.77x — poor on both scales, the model's number is withheld
No flags: Versus the analyst forecast for 2027 (5 houses): profit 1.03x
Caveat: The miss is one-sided: across 17 measurements the median profit forecast came to 0.77 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: Frontier див-модель (0.90x). Compared with: eninvs Company_Forecast, Frontier див-модель
The balance sheet is not used as a driver here: an exchange earns fees and interest on client balances, not a spread on its own assets. Profit runs off equity.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (80,645) differs from the last reported (38,738) by a factor of 2.08. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 3.1x, three-year median 3.5x). Re-rating to 4.5x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +2%
Caveat: Year-3 margin 6% is below the worst year on record (15%), though inside tolerance
No flags: Year-3 net debt/EBITDA 0.1x
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
Caveat: Historical check (6 observations): median EBITDA miss 5.2 pp of margin, within ±5 pp in 50% of cases, relative error 21% at bias 1.13x, revenue 10%
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2021, 2022). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: Less reaches actual cash than the formula implies: over 3 reported years free cash flow came to 0.79 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.93x). Compared with: eninvs Company_Forecast, Frontier спот-модель
Error: First-year profit (80,645) differs from the last reported (38,738) by a factor of 2.08. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — объёмов нет — рост объёма принят нулевым. A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 30% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
The model historically overstated this name's EBITDA by 1.13x (6 observations), so the path is trimmed by 11%.
The analyst forecast for this name is in USD while the model runs in RUB — the line-by-line comparison is switched off.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (9 observations): median EBITDA miss 22.1 pp of margin, within ±5 pp in 0% of cases, relative error 53% at bias 0.56x, revenue 19% — poor on both scales, the model's number is withheld
Exit valuation basis: No re-rating: min(current 6.4x, three-year median 15.5x). Re-rating to 9.4x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, momentum
No flags: Revenue CAGR +21%
Caveat: Year-3 margin 50% is above the best year on record (50%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 0.0x
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
Error: Historical check (9 observations): median EBITDA miss 22.1 pp of margin, within ±5 pp in 0% of cases, relative error 53% at bias 0.56x, revenue 19% — poor on both scales, the model's number is withheld
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Versus the analyst forecast for 2027 (8 houses): revenue 0.93x, EBITDA 0.98x, profit 0.74x
Caveat: The miss is one-sided: across 9 measurements the median EBITDA forecast came to 0.56 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The share count is growing: 66m 3.0 years ago against 71m now — 2.5% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 7% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 3.7 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: The reporting-period series needed repair: 4 rows discarded as incompatible with the cumulative convention (years 2020, 2023, 2024, 2025). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: Less reaches actual cash than the formula implies: over 2 reported years free cash flow came to 0.57 of 'EBITDA less capex, interest and tax'. The gap is lease principal (IFRS 16 keeps it out of EBITDA but it still has to be paid) and working capital. The forecast cash flow is scaled by this factor — otherwise debt repayment and dividends come out of cash the company never has
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.64x). Compared with: eninvs Company_Forecast
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (19,372) differs from the last reported (8,338) by a factor of 2.32. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 6.4x, three-year median 10.7x). Re-rating to 9.3x on a falling rate is the bull case
Model checks
No flags: Growth sources: guidance, momentum
No flags: Revenue CAGR +36%
No flags: Year-3 margin 45% inside history 37%–50%
No flags: Year-3 net debt/EBITDA 0.2x
No flags: Three-year dividends — 6% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
Caveat: Historical check (6 observations): median EBITDA miss 11.9 pp of margin, within ±5 pp in 17% of cases, relative error 29% at bias 0.74x, revenue 27%
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The miss is one-sided: across 6 measurements the median EBITDA forecast came to 0.74 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The share count is growing: 200m 2.3 years ago against 212m now — 2.7% a year, last change 2026-01-27. If that pace holds, the value of a share in year 3 is 8% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
No flags: Governance: quality multiplier 0.82 (flags: cash_conversion). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 7.4% of revenue a year on top of cash flow and dividends (range -13.6…+10.0%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: There was no free cash flow in any of the last 4 years: by our own formula it came out negative, and the cash-flow statement shows a median of -3,970 mln. The model still projects positive cash flow, drawn from EBITDA growth that historically covered only interest and capex. The forecast dividends and debt repayment rest on that assumption
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.04x). Compared with: eninvs Company_Forecast
No flags: Year 1 within -4% of guidance
Error: First-year profit (19,372) differs from the last reported (8,338) by a factor of 2.32. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Year-1 margin from the company's EBITDA margin guidance: guided 45% vs 40% actual, applied 45% (delivery-record shrink 1.00)
Forecast withheldProfit forecasts were historically overstated
Price used in model
384 ₽
As-of date of this price
not supplied
Model calculated at
2026-10-29T19:37:01
Reporting basis
2026-06-30
Exit valuation multiple
3.12x
EV/EBITDA
One-year EBITDA forecast error / observations
11% / 6
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Across 4 historical measurements the profit forecast came to a median 1.95 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
Exit valuation basis: No re-rating: min(current 3.5x, three-year median 3.1x). Re-rating to 4.6x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +6%
No flags: Year-3 margin 28% inside history 18%–32%
No flags: Year-3 net debt/EBITDA 0.4x
No flags: Three-year dividends — 30% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The base carries no capex — it is set equal to depreciation
No flags: Base is fresh: report 4 months old
No flags: Historical check (6 observations): median EBITDA miss 3.0 pp of margin, within ±5 pp in 67% of cases, relative error 11% at bias 1.07x, revenue 7%
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.07x, EBITDA 1.08x, profit 1.71x
Caveat: The model matches the houses on revenue and EBITDA but differs 1.71x on profit. For a leveraged name that is almost always the rate path: the model runs interest down with the key rate, the street is usually more cautious
Caveat: Governance: quality multiplier 0.63 (flags: no_cash_flow, curated). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 7.9% of revenue a year on top of cash flow and dividends (range -5.9…+14.6%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 4 rows discarded as incompatible with the cumulative convention (years 2019, 2021). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
No flags: The formula undercounts cash: over 3 reported years free cash flow came to 1.64 of 'EBITDA less capex, interest and tax', and the excess held in 2 of 3 years. The forecast is scaled up 1.30x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.95x). Compared with: eninvs Company_Forecast, Frontier див-модель, Frontier спот-модель
Error: Across 4 historical measurements the profit forecast came to a median 1.95 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 255.9 -> 246.6 mln toe за 2024-2025 (-3.6% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: The model's base report is 34 months old; without fresh reporting a three-year forecast has nothing to stand on
Exit valuation basis: No re-rating: min(current 0.9x, three-year median 1.2x). Re-rating to 1.5x on a falling rate is the bull case
Model checks
No flags: Growth sources: market, momentum
No flags: Revenue CAGR +4%
No flags: Year-3 margin 35% inside history 23%–38%
No flags: Year-3 net debt/EBITDA -1.9x
No flags: Three-year dividends — 6% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The base carries no capex — it is set equal to depreciation
Error: The model's base report is 34 months old; without fresh reporting a three-year forecast has nothing to stand on
Error: Model year 1 is already 100% over: the base report is 34 months old and year 1 ends twelve months after it. Forecasting a period that has all but happened, with no interim report, is not defensible
Caveat: Too little history to check (1 observations) — forecast accuracy unmeasured
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Governance: quality multiplier 0.60 (flags: curated). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: The name has two share classes: ordinary shares are 67% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.50x
Caveat: The debt build-up beyond cash flow could not be measured: 0 usable years, 0 discarded (perimeter change, asset sale, revaluation). The forecast runs without it — where it could be measured it adds about 5% of revenue a year to debt
Caveat: Cash conversion is not measured: of 0 reported years only 0 were comparable. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
No flags: One-year value agrees with an independent estimate: Frontier спот-модель (0.62x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
Error: First-year profit (463,026) differs from the last reported (1,322,113) by a factor of 0.35. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
The price move passes into revenue to the power of 0.51 (общая оценка по нефтяникам), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 53.7 -> 54.7 Mt за 2024-2025 (+1.9% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (615) differs from the last reported (2,828) by a factor of 0.22. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 6.9x, three-year median 11.3x). Re-rating to 10.0x on a falling rate is the bull case
Model checks
No flags: Growth sources: guidance, momentum
No flags: Revenue CAGR +13%
Caveat: Year-3 margin 10% is above the best year on record (9%), though inside the +10% tolerance
No flags: Year-3 net debt/EBITDA 4.8x
No flags: Three-year dividends — 8% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 33%
No flags: Base is fresh: report 4 months old
Caveat: Historical check (8 observations): median EBITDA miss 3.0 pp of margin, within ±5 pp in 75% of cases, relative error 39% at bias 0.76x, revenue 40%
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: The miss is one-sided: across 8 measurements the median EBITDA forecast came to 0.76 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The share count is growing: 280m 3.0 years ago against 400m now — 12.6% a year, last change 2024-10-31. If that pace holds, the value of a share in year 3 is 30% below the base figure: the base holds today's count (no issue is announced) and continued dilution sits in the bear case
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 3.9 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
No flags: Governance: quality multiplier 0.80 (flags: cash_conversion). No discount is applied to the exit multiple — it is the name's own market multiple and the market already prices this risk in
Caveat: Net debt grows for reasons beyond a cash shortfall: over 5 years it added 7.5% of revenue a year on top of cash flow and dividends (range -11.1…+7.8%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: The reporting-period series needed repair: 2 rows discarded as incompatible with the cumulative convention (years 2022, 2024). The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
Caveat: Cash conversion is not measured: of 4 reported years only 1 were comparable, and in 2 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
Caveat: One-year value 61 is an outlier: even the nearest independent estimate (eninvs Company_Forecast 241) is 0.3x away. Compared with: eninvs Company_Forecast
No flags: Year 1 within +3% of guidance
Error: First-year profit (615) differs from the last reported (2,828) by a factor of 0.22. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (354,987) differs from the last reported (172,688) by a factor of 2.06. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 3.0x, three-year median 3.6x). Re-rating to 4.5x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, market, momentum
No flags: Revenue CAGR +8%
No flags: Year-3 margin 23% inside history 18%–31%
No flags: Year-3 net debt/EBITDA -0.7x
No flags: Three-year dividends — 39% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Base is fresh: report 4 months old
No flags: Historical check (8 observations): median EBITDA miss 2.0 pp of margin, within ±5 pp in 75% of cases, relative error 9% at bias 0.94x, revenue 8%
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Versus the analyst forecast for 2027 (7 houses): revenue 1.16x, EBITDA 1.25x, profit 1.34x
Caveat: The name has two share classes: ordinary shares are 94% of the market cap. The exit multiple is computed on the whole cap, so the exit value is split between classes in the same proportion — otherwise the ordinary upside would be overstated 1.07x
Caveat: The reporting-period series needed repair: 5 rows discarded as incompatible with the cumulative convention (years 2020, 2021), and 4 cumulative periods rebuilt by summing standalone quarters. The database carries quarters labelled as halves and nine-month periods; LTM and year-on-year growth computed on those are wrong, so they are dropped
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (1.04x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель, Frontier спот-модель
Error: First-year profit (354,987) differs from the last reported (172,688) by a factor of 2.06. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Error: Across 7 historical measurements the profit forecast came to a median 1.69 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
The price move passes into revenue to the power of 0.41 (своя оценка), not one for one. The tax part of costs still follows the full price move — mineral tax is levied on price, not revenue.
Volume is taken separately, not from revenue momentum — фактическая добыча: 27.3 -> 27.83 mln t за 2024-2025 (+1.9% в год, в модель взята половина). A commodity producer's revenue is mostly price, so using revenue growth as volume would count the price twice.
Commodity prices held at spot; 90% of cash costs follow the output price (mineral tax, energy, freight), the rest follows inflation. FX drifts with the inflation differential (PPP), so USD revenue still grows in roubles at a flat commodity price.
Dividend policy from the primary source: Applied literally off RAS net profit of the parent; the model works from IFRS profit, so years with a wide RAS/IFRS gap will differ.
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
Exit valuation basis: No re-rating: min(current 4.7x, three-year median 4.9x). Re-rating to 6.8x on a falling rate is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Revenue CAGR +10%
No flags: Year-3 margin 17% inside history 11%–20%
No flags: Year-3 net debt/EBITDA 4.4x
Caveat: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Error: Holding prices flat with current cost inflation wipes out modelled equity — read it as a risk signal, not a valuation
No flags: Base is fresh: report 4 months old
Caveat: Historical check (9 observations): median EBITDA miss 2.5 pp of margin, within ±5 pp in 78% of cases, relative error 20% at bias 0.95x, revenue 6%
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 8.5 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: Net debt grows for reasons beyond a cash shortfall: over 4 years it added 3.8% of revenue a year on top of cash flow and dividends (range +0.0…+14.3%) — new leases, acquisitions, revaluation. That addition is carried in the forecast; without it the model would show deleveraging where debt keeps rising
Caveat: Cash conversion is not measured: of 4 reported years only 0 were comparable, and in 2 of them the formula cash flow was negative. The forecast runs on the uncorrected formula — where the correction could be measured it cuts roughly a third
Caveat: Leverage at entry: the market cap is 28% of enterprise value, so a move in business value is multiplied roughly 3.6x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
Forecast withheldLimited validation and model flaws
Price used in model
36.28 ₽
As-of date of this price
not supplied
Model calculated at
2026-10-29T09:37:02
Reporting basis
2026-06-30
Exit valuation multiple
5.01x
EV/EBITDA
One-year EBITDA forecast error / observations
43% / 3
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: The model for this name is unvalidated (3 measurements against the four required) and the name carries 5 structural flaws: thin equity, perimeter change, growth bought with debt, no free cash flow ever, leverage at entry. Either alone would be tolerable; together they are not — the number is withheld
Exit valuation basis: No re-rating: min(current 5.0x, three-year median 6.4x). Re-rating to 7.3x on a falling rate is the bull case
Model checks
No flags: Growth sources: guidance, momentum
No flags: Revenue CAGR +7%
No flags: Year-3 margin 30% inside history 28%–42%
No flags: Year-3 net debt/EBITDA 3.8x
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 41%
Caveat: The reporting perimeter changed in 2022: full-year revenue 6,961 against 6,295 for H1 of the same year — only 10% left for the rest of the year. That is a discontinued-operations reclassification with restated comparatives. Pre-break history is excluded and analyst forecasts may sit on a different perimeter
No flags: Base is fresh: report 4 months old
Caveat: Too little history to check (3 observations) — forecast accuracy unmeasured
Caveat: The growth is bought, not earned: for every rouble of EBITDA added over three years, net debt grew by 9.4 roubles. The model carries the historical EBITDA pace forward but only funds maintenance capex — acquisitions are not in the outflows. If growth keeps costing the same, free cash flow and dividends here are overstated
Caveat: There was no free cash flow in any of the last 4 years: by our own formula it came out negative, and the cash-flow statement shows a median of -2,905 mln. The model still projects positive cash flow, drawn from EBITDA growth that historically covered only interest and capex. The forecast dividends and debt repayment rest on that assumption
Caveat: Leverage at entry: the market cap is 23% of enterprise value, so a move in business value is multiplied roughly 4.4x in the shares. It cuts both ways: the IRR here is bought with debt, not with business quality
Caveat: One-year value 46 is an outlier: even the nearest independent estimate (eninvs Company_Forecast 126) is 0.4x away. Compared with: eninvs Company_Forecast
No flags: Year 1 within -3% of guidance
Error: The model for this name is unvalidated (3 measurements against the four required) and the name carries 5 structural flaws: thin equity, perimeter change, growth bought with debt, no free cash flow ever, leverage at entry. Either alone would be tolerable; together they are not — the number is withheld
Year-1 margin from the company's EBITDA margin guidance: guided 30% vs 28% actual, applied 30% (delivery-record shrink 1.00)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: First-year profit (217,206) differs from the last reported (82,201) by a factor of 2.64. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Exit valuation basis: No re-rating: min(current 3.0x, three-year median 3.5x). Re-rating to 4.4x on a falling rate is the bull case
Model checks
No flags: Growth sources: consensus, guidance, momentum
No flags: Revenue CAGR +11%
Caveat: Year-3 margin 6% is below the worst year on record (6%), though inside tolerance
No flags: Year-3 net debt/EBITDA 2.0x
No flags: Three-year dividends — 36% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
Caveat: Thin equity: it is a small slice of EV, a 10% EBITDA move shifts the valuation by roughly 31%
No flags: Base is fresh: report 1 months old
No flags: Historical check (7 observations): median EBITDA miss 0.4 pp of margin, within ±5 pp in 100% of cases, relative error 6% at bias 1.06x, revenue 8%
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Versus the analyst forecast for 2027 (8 houses): revenue 0.98x, EBITDA 0.88x, profit 1.74x
Caveat: The model matches the houses on revenue and EBITDA but differs 1.74x on profit. For a leveraged name that is almost always the rate path: the model runs interest down with the key rate, the street is usually more cautious
No flags: The formula undercounts cash: over 4 reported years free cash flow came to 1.12 of 'EBITDA less capex, interest and tax', and the excess held in 3 of 4 years. The forecast is scaled up 1.12x — never above 1.3, which would extrapolate a good year
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.67x). Compared with: eninvs Company_Forecast, Frontier див-модель
No flags: Year 1 within -1% of guidance
Error: First-year profit (217,206) differs from the last reported (82,201) by a factor of 2.64. A one-year step of that size means the base and the forecast rest on different assumptions (interest, one-offs), so dividends and IRR built on it are not anchored in fact — the number is withheld
Error: Across 6 historical measurements the profit forecast came to a median 1.69 of actual — the model consistently overstates this name's profit. Dividends and IRR run off profit; the number is withheld
Year-1 margin from the company's EBITDA margin guidance: guided 6% vs 6% actual, applied 6% (delivery-record shrink 1.00)
Error is the median absolute percentage miss in a one-year operating forecast, not an error estimate for three-year investment returns. The observation count follows the slash; a small sample limits conclusions.
Why the forecast is withheld: Historical check (10 observations): median profit miss 10.1 pp of revenue, within ±5 pp in 20% of cases, relative error 52% at bias 0.48x — poor on both scales, the model's number is withheld
Exit valuation basis: Partial re-rating: the mean of today's P/B 0.94 and the three-year median 1.09. For banks this base is measurably better — on our daily series (14 names, 5,163 observations) the median three-year error is 0.42 against 0.45 for today's multiple and 0.50 for the median alone, and the odds of ending above today's are 48%. Full re-rating to 1.48 is the bull case
Model checks
Caveat: Growth rests on a single source (momentum)
No flags: Year-3 ROE 15%
No flags: Three-year dividends — 36% of the current price
No flags: This check’s numerical outcome is withheld with the forecast. The reason is explained above.
No flags: Profit runs off assets rather than retained earnings; the capital cushion covers the whole horizon
No flags: Base is fresh: report 1 months old
Error: Historical check (10 observations): median profit miss 10.1 pp of revenue, within ±5 pp in 20% of cases, relative error 52% at bias 0.48x — poor on both scales, the model's number is withheld
Caveat: The miss is one-sided: across 10 measurements the median profit forecast came to 0.48 of actual — the model consistently FAILS TO KEEP UP with this business. So the IRR here is more a floor than a ceiling; erring this way is safer, but the precision cannot be relied on
Caveat: The forecast cash flow cannot be checked against the statement: the group holds a bank or fintech, so client money inflows land in operating cash flow — Ozon's 'actual' comes to +216bn where the formula gives minus 42. No conversion correction is applied here
No flags: One-year value agrees with an independent estimate: eninvs Company_Forecast (0.77x). Compared with: eninvs Company_Forecast, eninvs DividendPrediction, Frontier див-модель
Error: Across 10 historical measurements the profit forecast came to a median 0.48 of actual — the model consistently understates this name's profit. Dividends and IRR run off profit; the number is withheld
Profit runs off assets (RoA 14.70%), not retained earnings; asset growth from собственная история активов, медиана 3 темпов (+5%, +8%, +2%). Growth is capped by leverage: equity/assets stays above 69.3%, the bank's own historical minimum.
Each name is modelled from its latest LTM (last twelve months) figures three years forward. Corporate revenue uses available sources (leading related-market prices, company guidance adjusted for its own delivery record, independent market research and the realised pace of acceleration or deceleration), then EBITDA, profit, dividends per policy, debt and an exit value. For banks, profit is derived from assets and their return; growth is constrained by capital. IRR is the cash flow: price today → three years of dividends → price in three years. Payments and the sale are placed at year ends: this is equal-period annual IRR, not calendar-date XIRR. Bear and bull cases are not a confidence interval.
EBITDA means earnings before interest, income tax, depreciation and amortisation. The EBITDA error in the model details describes forecast testing: the model was anchored at past reporting dates and its one-year-ahead forecast compared with the actual. Where the error is wide, the number is withheld.
Backtest accuracy and caveats
Accuracy across the book: revenue median error 8%, bias 0.98x, within ±20% in 79% of cases; EBITDA 15%, bias 0.99x, ±20% in 60%; net profit 30%. Observations: 341 / 312 / 346.
Backtest caveats: The raw-period financial statements have been revised; the initial figures available in real time may have differed.; The EI1 commodity-price series were completed retrospectively.; The guidance and research directory was assembled today; its selection of companies incorporates future information.; Profit is calculated using realised interest rates; without a historical policy-rate series, the profit metric is unreliable.