Methodology

How a model gets built.

Every report Intrinsic produces follows the same playbook a junior analyst would. What we model, how we discount, where the numbers come from, and where the model knowingly bends.

Two models, one router

Different businesses need different lenses. A discounted cash flow reads cleanly on operating companies — places where free cash flow actually means something. It falls apart on banks, where leverage and regulatory capital make “cash flow” a misleading metric. So we route financials to a Residual Income model instead.

  • DCF. Technology, Consumer, Healthcare, Industrials, Energy, Utilities, Materials, Communication Services, Real Estate.
  • Residual Income. Financial Services — banks, insurers, asset managers.

The Results page tells you which model ran and why. Routing keys off the company’s sector classification as reported by Yahoo Finance.

The DCF model

Unlevered free cash flow

For each year in the projection horizon, the engine computes UFCF from your assumptions:

UFCF — per projection year
UFCF = EBIT × (1 − tax rate) + D&A − CapEx − ΔNWC

EBIT comes from your revenue growth and margin trajectory inputs. D&A, CapEx, and working capital movements project as a percentage of revenue, with defaults seeded from the company’s three-year historical average — not the latest year, which can be noisy.

Discount rate (WACC)

The weighted average cost of capital weighs equity and debt by market value, taxes interest as a shield, and falls out of CAPM:

WACC
WACC = (E/V) × Ke + (D/V) × Kd × (1 − t)
  • Cost of equity (Ke). CAPM. Risk-free rate from the current 10-year US Treasury, equity risk premium fixed at 5.5%, beta from Yahoo Finance.
  • Cost of debt (Kd). Interest expense over average total debt, floored at the risk-free rate to handle names with negligible borrowings.
  • Capital weights. Market value of equity (market cap) and book value of debt from the most recent balance sheet.
Terminal value

At the end of the explicit forecast, Gordon Growth takes over:

Terminal value
TV = UFCF(final) × (1 + g) / (WACC − g)

Terminal growth g defaults to 2.5% — roughly long-run US nominal GDP — and you can move it. The Results page reports terminal value as a percentage of enterprise value, which is worth watching.

Rule of thumbIf terminal value exceeds 75% of enterprise value, the model is leaning on perpetuity assumptions rather than the explicit forecast. Tighten the forecast, lower g, or extend the projection horizon.
Equity bridge
Equity value
Equity value = EV − Net debt = EV − (Total debt − Cash)

Divide by diluted shares outstanding. That’s the per-share intrinsic value the Results page shows. The difference against current market price is the upside, or the warning.

Residual Income (financials)

For banks and insurers, free cash flow is the wrong yardstick. What matters is whether the bank earns more than the cost of the equity capital it’s using. Residual Income measures exactly that:

Residual income — per period
RI(t) = Net income(t) − Ke × Book value(t−1)

Equity value equals current book value plus the discounted value of all future residual income, with Gordon Growth applied to the final year. For comparables, financials use P/B and P/TBV — not EV/EBITDA, which doesn’t translate to leveraged balance sheets.

Comparable companies

The Comps sheet pulls live multiples for the peer set you choose. Multiple selection is sector-aware:

  • Operating businesses. EV/EBITDA, EV/Revenue, P/E.
  • Financials. P/B, P/TBV, P/E, dividend yield.

Peer median and mean compute for each multiple and apply to the target’s own financials, producing a comp-implied range. Treat it as a cross-check on the DCF — when the two disagree sharply, the assumptions are usually the place to look.

Data sources

  • Yahoo Finance (yfinance). Price, market cap, beta, full financial statements, sector, 52-week range, shares outstanding.
  • SEC EDGAR.Filings and structured financial data for S&P 500 constituents.

Both are free and public. No paid aggregators, no cached historicals — every report pulls fresh data at the moment you run it.

What the model can’t do (yet)

  • S&P 500 only. Coverage is US large-cap during beta. FTSE 100 and HSI are on the roadmap.
  • Single-segment modelling.Conglomerates with materially different segments (Berkshire, GE) model in aggregate. Sum-of-the-parts isn’t supported yet.
  • No options or convertibles bridge. Diluted shares are used as reported. Treasury stock method modelling is planned for v2.
  • yfinance latency.Yahoo’s data is reliable for S&P 500 names but can lag filings by a few days around earnings. For anything consequential, cross-check the 10-K or 10-Q.
Outputs are for educational and research purposes only. They do not constitute investment advice, a recommendation, or a solicitation to transact. Always conduct your own due diligence.