Methodology

How the model values and picks companies

One fixed rule, run after every US market close on the 499 companies of the S&P 500. It reads nothing but each company’s own SEC filings and its share price: no analyst forecasts, no news, no opinions. Here is every step, and what every number on the site means. New to value investing? Start with the idea in plain English, then come back for the details.

01 · Data

Where the numbers come from

Financials come from the XBRL data in each company’s 10-K and 10-Q filings on SEC EDGAR. A watcher checks EDGAR every hour and refreshes a company’s numbers when it files.
Trailing twelve months (TTM) means the last four reported quarters: the last full year, plus this year to date, minus the same period last year. When a company restates a number, the restated figure wins.
Prices are daily closes. Everything is recomputed after the close, around 22:30 UTC, on the latest data: today that is 2026-10-28.
The universe is the S&P 500, 499 companies once share classes are merged. Each keeps its GICS sector and sub-industry, which decide how it’s valued.

02 · Value

How a company is valued

The model value is a ten-year discounted cash flow. In plain terms: how much cash will this business hand its owners, and what is that worth today if you want about 9% a year for waiting?

Start from today’s cash earnings. For most companies, the average of free cash flow and after-tax operating profit (operating profit × 0.79), so a single lumpy year of investment doesn’t swing the answer.
Pick a growth rate from the company’s own record. The average of its three-year and one-year revenue growth, held between −5% and +15% (commodity cyclicals at most 5%), then nudged by its AI exposure.
Grow for five years, then slow down. Years one to five grow at that rate; years six to ten fade it in a straight line to 3%, roughly the long-run growth of the economy.
Add a terminal value for everything after year ten, growing at 3% forever.
Discount it all back at the required return below, and subtract net debt: what’s left belongs to shareholders.

Kind of company	Cash earnings it starts from	Required return
Most companies	Average of free cash flow and after-tax operating profit	9%
Banks, insurers, asset managers	Net income less what must be retained to grow, capped at 15% of book equity	9.5%
Utilities	Net income less what must be retained to grow	8%
REITs	Operating cash flow less what must be retained to grow	8.5%

Cyclicals at high margins are valued on normal margins. When a commodity company earns over 1.25× its 2021–2025 average margin (1.5× for chips and hardware), its cash earnings are scaled back toward that average before valuing, so a boom isn’t capitalised as if it lasts forever.

Upside is model value divided by market cap, minus one. Growth the price implies runs the same model backwards: the five-year growth rate at which model value would equal today’s price. Put it next to the company’s three-year revenue growth: a price that implies less than the company has been delivering is undemanding.

03 · Quality

The quality score

A cheap price is only a bargain if the business is good. The quality score ranks every company against the rest of the universe on three things and blends the rankings into one number from 0 to 1:

50%Returns on capitalROIC, or ROE for banks, insurers, utilities and REITs. Does every dollar in the business earn a lot?

25%GrowthThree-year revenue growth. Is the business getting bigger?

25%Low debtNet debt / EBIT, lower is better. Can it survive a bad year?

Each part is a percentile (0.8 means better than 80% of companies), so the score says how a company compares, not how it does in absolute terms. The rule needs 0.5 or more to buy and sells below 0.4.

04 · The rule

The six tests

A company qualifies only when it passes all six on the same day. 55 do today.

Upside of 20% or more. A margin of safety: the model can be wrong by a lot and the price still be fair.
Quality score of 0.5 or more. Only good businesses.
Profitable over the last twelve months: positive net income.
Positive cash earnings, the figure the valuation starts from. Nothing to value otherwise.
Net debt no more than 3.5× operating profit. A company with no operating profit has no meaningful ratio and isn’t failed on this test; it fails the profit tests instead.
Not a cyclical at peak margins. Commodity, chip and hardware businesses earning over 1.5× their 2021–2025 average margin look cheapest just before profits fall.

Qualifying companies are ranked by upside, with anything above 200% counted as 200% so one extreme estimate can’t dominate.

05 · Trading

How the portfolio trades

The model portfolio is a paper portfolio that started with $200,000 on 2026-10-15. After each close the rule runs four steps, in this order:

1 · Sell: A holding is sold when upside falls under 5% (the price has reached the value), quality drops below 0.4, it makes a loss or negative cash earnings, net debt goes above 4.5× operating profit, or it becomes a cyclical at peak margins.
2 · Swap: A holding that has been owned at least 28 trading days and has slipped below rank 30 (or stopped qualifying) is swapped for the best candidate when that candidate has 25 points more upside. At most 2 swaps a day.
3 · Buy: Empty slots are filled from the top of the ranking until the portfolio holds 15 companies, each with an equal 6.6% of the portfolio (1% stays in cash).
4 · Rebalance: A position that drifts more than 25% away from its target weight is trimmed or topped back up.

Why the buy and sell lines differ. The rule buys at 20% upside but only sells under 5%; it buys at quality 0.5 but sells under 0.4. The gap stops it trading back and forth on small daily moves. Every trade pays 5 basis points of cost, fills at the closing price, and is compared with the S&P 500 (SPY) and its equal-weight version (RSP). See the portfolio and every signal.

06 · Categories

Company categories

Each company page labels the business in plain words so you can tell at a glance what kind of company you’re looking at. The labels come straight from the numbers, with fixed thresholds.

Size

Mega cap. Market cap above $200B: one of the largest listed companies in the world.
Large cap. Market cap $10–200B: an established, widely owned company.
Mid cap. Market cap $2–10B: big enough to be established, small enough to still grow fast or stumble.
Small cap. Market cap under $2B: smaller, more volatile, less covered by analysts.

AI exposure

Set by sub-industry. It moves the model’s growth assumption for the first five years.

Group	Who	Growth nudge	Companies
Infrastructure	Chips, chip equipment, hardware, networking, data-center REITs, electrical equipment, construction & engineering	+4 pts	59
Platform	Internet platforms, systems software, cloud, broadline retail	+3 pts	16
Power	Utilities, pipelines and storage, copper	+1.5 pts	33
Adopter	Everyone else: should get somewhat more efficient	+0.5 pts	346
Disrupted	IT consulting, staffing, advertising, application software, publishing, data vendors, insurance brokers	−3 pts	45

Traits

High returns on capital: ROIC of 20% or more.
Fat margins: Operating margin of 25% or more.
Growing fast: Revenue up 15% or more over the last year.
Shrinking sales: Revenue down over the last year.
Cash machine: Free cash flow yield of 6% or more.
Net cash: More cash than debt.
Heavy debt: Net debt above 3.5× operating profit.
Cyclical at a peak: Cyclical business with margins well above its own history.
Deep discount: 50% or more upside to model value.
Above model value: Share price above model value.
Funds piling in: 13F holders up 5% or more last quarter.
Funds leaving: 13F holders down 5% or more last quarter.

07 · Who else owns it

Funds and Congress

Two kinds of public disclosure show who else is buying. Neither feeds the valuation or the six tests: they are context, not signals the rule acts on.

13F filings. Every manager with over $100M in US stocks reports its holdings each quarter, within 45 days. We read every filing and show how many funds hold a company, how that changed, and what about 40 well-known stock pickers did. More in what a 13F filing is and the latest filings.
Congress trades. Members of the US House must disclose stock trades within 45 days. We count purchases, sales and members trading over the latest 180 days of reports. House only, and counts rather than amounts.

08 · Limits

What the model can't see

The future. Growth is projected from the company’s own past. A business about to turn up or fall apart looks like its history until the filings say otherwise.
One-offs. A big write-down or a one-time gain moves twelve-month profits, and with them P/E and sometimes the tests.
Management, moats and brands, except as they show up in returns, margins and growth.
Banks and insurers fit any cash-flow model awkwardly: their debt is raw material, not a burden. Treat their leverage figures loosely.
Timeliness of disclosures. 13Fs and Congress reports arrive weeks after the trades.

That is why the rule asks for a 20% margin of safety, holds 15 companies rather than a few, and publishes everything. Research only, not investment advice.

09 · Glossary

Every term, A to Z

The basics

Market cap: Share price times shares outstanding: what you would pay for every share at today's price. It also sets the size category (mega cap above $200B, large cap $10–200B, mid cap $2–10B).
1-year return: The change in the share price over the last 52 weeks, measured from weekly closing prices. The sparkline beside it is the same series.
Sector & industry: Every company carries its GICS classification: one of 11 sectors (Information Technology, Financials…) and a narrower sub-industry (Broadline Retail, Regional Banks…). The sub-industry decides which valuation model a company gets, whether it counts as a cyclical, and its AI exposure.

Valuation: is it cheap?

Model value: A 10-year discounted cash flow built only from the company's own filings. It starts from today's cash earnings, grows them for five years at the company's own recent growth, fades that to 3% by year ten, adds a terminal value, discounts it all back at about 9% a year and subtracts net debt. The full recipe is in the valuation section above.
Upside to model value: Model value divided by market cap, minus one. The rule only buys with 20% or more upside, a margin of safety for the model being wrong, and sells once upside falls under 5%: the price has caught up with the value.; How to read it: Under 0% the price is above the model's value; 20–50% is a solid discount; above 100% usually means the market expects something the filings don't show yet, so read the rest of the page before trusting it.
Growth the price implies: A reverse DCF: the five-year growth rate at which the model's value would exactly equal today's market cap, with everything else unchanged. If the price implies 9% a year and the company has grown revenue 12% a year, the market is asking for less than its track record. Shown only for companies valued on free cash flow, not banks, insurers, utilities or REITs.; How to read it: Below the company's own growth record: the price looks undemanding. Far above it: the price needs the business to speed up.
P/E: Market cap divided by net income over the trailing twelve months. A P/E of 20 means paying 20 years of today's profit. Blank when the company lost money. A quick sanity check rather than the model: one-off gains and losses swing it a lot.; How to read it: Under 15 is cheap for most businesses, 15–25 typical, above 30 needs fast growth to justify it.
FCF yield: Operating cash flow minus capital expenditure over the trailing twelve months, divided by market cap. Think of it as the cash interest rate the business pays on today's price, before any growth. Blank for banks and insurers, whose cash flow statements don't work this way.; How to read it: Above 5% is generous; near zero or negative means the price is mostly about the future.
EV / EBIT: Enterprise value (market cap plus debt minus cash) divided by EBIT, operating profit before interest and tax, over the last twelve months. Because it counts debt, it compares a debt-free company and a heavily borrowed one on equal terms. Shown only for companies valued on free cash flow with positive EBIT.; How to read it: Under 12 is cheap, 12–20 typical, above 25 expensive.
P/B: Market cap divided by shareholders' equity on the latest balance sheet. Useful for banks and insurers, whose assets are mostly financial. For software or brands, book value misses most of what the business is worth, so a high P/B there means little.
AI exposure: Each sub-industry is placed in one of five groups, and the model adds or subtracts growth for the first five years accordingly. Infrastructure (chips, hardware, networking, data centers, electrical equipment): +4 points. Platform (internet platforms, systems software, cloud): +3. Power (utilities, pipelines, copper): +1.5. Adopter (everyone else, who should get somewhat more efficient): +0.5. Disrupted (IT consulting, staffing, advertising, application software, publishing, data vendors): −3.

Quality: is it a good business?

Quality score: A blend of three rankings across the whole universe: 50% return on capital (ROIC, or ROE for banks, insurers, utilities and REITs), 25% three-year revenue growth and 25% low leverage (net debt / EBIT, lower is better). Each is a percentile from 0 to 1, so 0.62 means better than most on balance. The rule buys at 0.5 or more and sells below 0.4. Blank when returns or growth can't be measured.; How to read it: Above 0.75 is a top-quartile business; 0.5–0.75 good; under 0.4 the rule won't hold it.
ROIC: Operating profit after a standard 21% tax, divided by invested capital (equity plus debt minus cash) on the latest balance sheet. For every dollar tied up in the business, how much does it earn? Businesses that earn well above their cost of capital (roughly 8–10%) for years usually have a moat. Shown for companies valued on free cash flow.; How to read it: Above 15% excellent, 10–15% good, under 8% barely covers the cost of the money.
ROE: Net income over the trailing twelve months divided by shareholders' equity. High ROE can come from a great business or from heavy borrowing, which is why ROIC and leverage sit next to it. For banks and insurers it's the main measure of quality.
Operating margin: EBIT divided by revenue, both over the trailing twelve months. High, steady margins usually mean pricing power. Margins far above a company's own history in a cyclical industry mean the opposite: a peak.
Revenue growth (1y): Revenue in the latest full calendar year compared with the year before, from the 10-K filings.
Revenue growth (3y): Compound annual revenue growth over the last three calendar years, from annual filings (the one-year figure when three years aren't available). It feeds the quality score and, averaged with the one-year figure, the growth the model assumes.
Net income (TTM): Bottom-line profit over the trailing twelve months, built from the latest 10-K and 10-Qs (last full year plus this year to date minus last year to date), using restated figures when a company revises them. The rule only owns companies that are profitable today.
Net debt / EBIT: Borrowings on the latest balance sheet (long- and short-term debt, notes and commercial paper, including a finance arm's debt but not operating leases) minus cash and short-term investments, divided by EBIT over the trailing twelve months. Negative means net cash. Blank when operating profit is zero or negative, because the ratio means nothing then; the rule doesn't count a blank against a company. For banks and insurers debt is part of the business, so read it loosely there.; How to read it: Under 1× very safe, 1–3× normal, above 3.5× the rule won't buy, above 4.5× it sells.

The rule's verdict

Qualifies: A company qualifies when it passes all six tests at once: upside of 20% or more, quality of 0.5 or more, positive cash earnings, positive net income, net debt no more than 3.5× EBIT, and not a cyclical at peak margins. The list is recomputed after every close.
Rank: Qualifying companies are sorted by upside, with anything above 200% counted as 200% so one extreme number can't dominate. The portfolio fills empty slots from the top of this list, and a holding that falls below rank 30 can be swapped out.
Positive cash earnings: The valuation starts from a measure of cash earnings: the average of free cash flow and after-tax operating profit for most companies, operating cash flow for REITs, and net income for banks, insurers and utilities. If that is zero or negative there is nothing to value, so the company can't qualify.
Cyclical at peak margins: Cyclicals are the commodity industries (oil and gas, chemicals, metals and mining, steel, paper, airlines, homebuilders and similar) plus semiconductors, semiconductor equipment and tech hardware. One is flagged when its operating margin today is more than 1.5× its 2021–2025 average and above 10%, or, with no margin history, when its share price has more than doubled in a year. The rule won't buy a flagged company and sells one it holds.
Held by the model: The company is one of the 15 positions in the paper model portfolio as of the latest close.

13F: what the funds are doing

13F filing: Form 13F lists a manager's US stock positions at the end of each quarter and is due within 45 days. It only shows long positions, no shorts, and is weeks old when it lands. We count shares only (no options) across every filer in the SEC's data.
13F holders: How many 13F filers reported holding the stock at the end of the last complete quarter. Changes are shown against the quarter before and the same quarter a year earlier. A rising count means more managers are buying in; for the largest companies the count is mostly index funds and moves slowly.
Shares held, q/q: Total shares reported by all 13F filers at the end of the last full quarter, compared with the quarter before. Not adjusted for stock splits, so a split shows as a jump.
This quarter so far: 13Fs arrive over six weeks after a quarter ends; we parse each one within the hour. Among funds that have filed for the new quarter (and whose previous holdings we know), we count how many hold the stock and how many opened a position minus how many closed one. It firms up as more funds file.
Largest holders: The five largest positions reported in the last full quarter, in billions of dollars. These are usually Vanguard, BlackRock, State Street and their peers, who hold the index; the notable managers are the more telling signal.
Notable managers: A fixed list of about 40 concentrated, long-term managers, among them Berkshire Hathaway, Baupost, Pershing Square, Akre, Fundsmith, Dodge & Cox, Oakmark, Markel, Li Lu's Himalaya, Egerton and Wedgewood. For each we compare shares with the quarter before: new (from zero), added (over 10% more), trimmed (over 10% fewer) or exited. When some have already filed for the current quarter, their moves are shown first.

Congress trades

Congress trades: Members of the House must report stock trades within 45 days in Periodic Transaction Reports. We count purchases and sales of the stock (and options on it) over the latest 180 days of reports, and how many members traded. Counts, not amounts, and House only.