Methodology

The Tenet Score

Not the same as a stock's tenets

The 0-100 Tenet Score and the named tenets on a stock's Tenets tab are two separate computations that share a word. This section describes the score: the percentile work below, reported as one of four verdict words (Strong, Solid, Mixed or Weak). Each scored tenet runs its own criteria and its own pass rule, shown criterion by criterion on the tab itself; the screens listed under The screens further down this page also publish theirs here. No tenet verdict enters the score. The one path in the other direction is the Quality Floor tenet, which reads the score's Quality sub-score as one of its criteria and in turn feeds the synthesized Qualified Dip verdict.

Four sub-scores

Value (EBIT/EV yield, P/E, P/B, P/FCF, EV/EBITDA), Quality (ROIC, ROE, margins, income quality, Piotroski, Altman-Z), Health (current & quick ratio, net-debt/EBITDA, interest coverage, debt/equity), Growth (revenue & EPS CAGR, FCF margin).

Percentile basis

Each metric is ranked as a percentile against the stock's sector peers, direction-adjusted (low P/E ranks high). Sub-score = mean percentile × 100; composite = equal-weighted mean of available sub-scores. Beside every metric, the Statistics tab also shows where the current value (the trailing twelve months or the latest fiscal year, as its row names) sits in the company's own ten fiscal years, as context; it is never blended into the score.

Trailing twelve months

Since October 2, 2026, the score can read a company's trailing twelve months (TTM) in place of its latest fiscal year. Where a company files quarterly statements and its trailing figure passes our check against the filings, the score uses the trailing twelve months; where it does not, or where a sector's trailing figure is still under review, the latest fiscal year. The Statistics row names which, metric by metric. Where the trailing figure is used, earnings, EBIT, EBITDA, cash flow, revenue and interest are summed over the last four reported quarters; book value, working capital, debt and cash come from the most recent quarter's balance sheet; prices are the latest close. EV/EBITDA, net debt to EBITDA and interest coverage are on the latest fiscal year in every sector for now, pending the trailing check. Piotroski and the five-year revenue and EPS growth rates stay on fiscal years, because their methods are annual. A company whose latest fiscal year already ends at its newest quarter uses that year, which is its trailing twelve months. A company stays on its latest fiscal year when it files no quarterly statements (most foreign filers), when a quarter is missing, when its quarters do not add up to its fiscal year, or when its quarterly and annual statements are in different currencies, and on any night its trailing figures were not computed alongside its fiscal-year ones; a single figure stays on the fiscal year when a quarter lacks one of its inputs, when it rests on a correction made only to the annual filing, when a quarter's earnings to common shareholders cannot be worked out or the four quarters' share counts are too far apart (both for P/E), or when its sector's trailing figure for that metric is still under review. Peers on either basis are ranked together, and every figure on the Statistics and Tenets tabs names its period. Scores before October 2, 2026 were computed on fiscal-year figures and are kept as they were; score changes across that date are not reported as moves.

Your own record, shown not scored

A business compared only to its peers hides whether it is above or below its own normal, so the Statistics tab shows each metric against the company's own ten-year record as context. We tested blending it into the score against a pre-registered evaluation on our own ten-year dataset; it did not clear our bar, so the score stays peer-based and the comparison is published as context. We re-test as our data deepens.

Guards

Negative earnings/equity/EBIT null the affected ratio rather than producing a misleading number; metrics with fewer than 3 peer values are excluded and shown as such; thin sector groups fall back to a labeled broad-market basis. Metrics with fewer than 5 annual observations show no own-record marker for that metric. Missing coverage is displayed, never silently averaged. Banks and insurers are classed by their SEC industry code. Banks are not scored on EBIT/EV yield, P/FCF, EV/EBITDA, gross margin, income quality, Piotroski, Altman-Z, current ratio, quick ratio, net-debt/EBITDA, interest coverage, debt/equity or FCF margin, because those ratios do not describe a deposit-funded balance sheet; with no health metric left, a bank's composite averages the other three sub-scores. Insurers are not scored on EBIT/EV yield, EV/EBITDA, gross margin, Piotroski, Altman-Z, current ratio, quick ratio or net-debt/EBITDA, nor on interest coverage when only the data vendor's ratio is on file. Income quality is shown and scored only when net income is above zero; at or below zero, cash flow over net income has no meaning, so the cell is withheld and says why. Each night the newest fiscal year's revenue, net income and operating cash flow are checked against the sum of its four quarters: a gap above 0.5 percent is flagged, and where the annual figure is less than half or more than double the four-quarter sum, the figures built on it are not scored. Price-based figures (P/E, P/B, P/FCF, the EV multiples and EBIT/EV yield) are not shown when the statements behind them are out of date (the newest quarter more than 200 days old for trailing figures, the newest annual statements more than 18 months old for fiscal-year figures), or when the market value is not verified to be on the statements' share basis and currency.

Corrected inputs

Greenblatt earnings yield (EBIT ÷ EV; vendors' "earnings yield" is often net-income ÷ market cap) and interest coverage from raw statement fields (vendor ratios can return 0 where interest expense nets out). Where the vendor's standardized statement nets interest into other income, the interest line is read from the company's own filing, and where a filer reports interest inside its operating expenses, EBIT adds it back. P/E is the share price over earnings per share to common shareholders, from the last four reported quarters where the trailing figure passes our check against the filings (net income to common summed over the quarters, divided by their average diluted share count), else from the latest fiscal year, as in each sector whose trailing P/E is still under review; the data vendor's quoted P/E is used only as a check. EPS growth uses diluted EPS as filed, restated only for stock splits our own price history confirms. A company that has never carried debt scores a debt/equity of 0.

One score, many styles

The score rates the business

Not the moment. It reads value, quality, health and growth, each a sector-peer percentile, folded into one calibrated number, recomputed nightly.

Styles reorder what leads

An income or value style leads with what that school of investing weights most across the screener, the home dashboard and the playbook order; a style reorders and highlights, and it never blends into the score or recomputes it. Same numbers, same rules, weighted the way a value investor weights them. Nothing here is chosen for you. A chosen style is stored in this browser like the theme, and it shapes no list that is not available under every other style.

Presentation never changes the number

Grades, styles and section order change only what surfaces first; the stored composite is untouched by how it is shown.

Timing signals stand apart

They publish as their own dated, live-forward receipts (the per-criterion inputs behind a number, each with its source and date) carrying their own track record, never folded into the business score.

Factor grades & the radar

Five factor axes

Each an A+…F letter grade: Value, Quality, Health and Growth reuse the sector-calibrated sub-scores; Momentum is the 12-1 price return (the price change over the trailing 12 months excluding the most recent month; dividends not reinvested) as a percentile across the whole equity universe, never folded into the Tenet Score.

Honest basis

The four fundamental grades rank against sector peers (or the broad market for names with no sector); Momentum ranks against the equity universe, and the chip says so. An axis with no data shows "—", never a failing F.

The bands

Over the 0–100 percentile shown on each axis: A+ ≥ 97, A ≥ 90, B ≥ 70, C ≥ 50, D ≥ 30, else F. A grade is a label over a receipt. Recompute it yourself from the shown score.

What's free, what's Premium

The radar and every stock's grades are free on its report. The per-axis grade column on the screener (grades across the whole result set at once) is a Premium convenience.

Independence & conflicts

The method is version-controlled, the recompute unattended

Every formula, threshold and screen rule lives in version control. A change is committed and timestamped before it can affect a published number, and the nightly recompute that applies it runs unattended, on a schedule, with no one selecting outcomes. That commit history is the audit trail: what changed, and when, is a matter of record rather than of recollection.

No issuer payments, ever

Tenet takes no payment from any company, fund, issuer or their agents, for coverage, for inclusion in a screen, for ranking, for placement, or for a favourable reading. There are no sponsored names, no paid research, no investor-relations arrangements. Subscriptions are the revenue today.

No compensation tied to coverage

No one at Tenet is compensated, by anyone, on the basis of which securities are covered or how a security scores. There is no arrangement under which a rating, a screen membership or a written line moves anyone's pay.

No per-name human discretion

Scores and screen membership are computed by a systematic model from published criteria. No human discretion over any name's score or ranking; names are removed from publication only by published, mechanical evidence and hygiene rules that apply identically to every security. Those rules exist because data can be wrong or stale: a security whose identity evidence fails our cross-checks is quarantined, and one the exchange record shows is no longer trading is delisted from the universe. Both are evidence tests, applied to every security alike, never a judgment about a particular company's merits.

Founder holdings, and the methodology-change rule

Tenet's founder may hold securities Tenet covers; a research publisher whose author owns nothing is not more independent, only less invested. Two rules bound it. Holdings are never a reason a name is added to, removed from, or ranked differently within any screen, because no human sets those outputs. And no one at Tenet may trade ahead of a methodology change: a change that will alter published output is committed to version control before it runs, and personal trading in any name whose published status that change would move is prohibited from the time the change is drafted until the output is public.

What Tenet does not claim

Tenet is not a registered investment adviser and provides no personalised advice. The ledgers on this site are model-portfolio records computed by Tenet from its own published lists; they are not audited, not verified by any third party, not investor returns, and not a composite presentation under any performance-reporting standard. No one has examined our returns. Nothing here is a recommendation to buy or sell any security.

The performance ledger

Equal-weight, no look-ahead

Positions open when a name joins a screen's published list, priced at the next trading day's close after publication; a position is then held only while the name keeps qualifying, closing when it drops off the list. You can't trade a list you haven't seen.

No survivorship laundering

Delisted names close at their last price; the loss stays on the record. Win rate counts the losers: closed losses and negative open marks included.

Two lanes, indexed to 100

Price return and total return with every dividend reinvested at its ex-date close; benchmark is SPY from each screen's inception, same close basis.

No backfilled performance

Each screen's live record begins at its own first published list; the earliest began July 2, 2026, the newest July 16. Any future backtest will be shown in its own clearly-labeled lane, never merged. Each reinvestment is recorded as its own auditable line, an explicit record that accrues forward, never rewriting a published number.

No guarantee of future results

Past performance never guarantees future results.

The screens

Each applies its published criteria mechanically to the eligible US stock universe, nightly. Screens whose ratios break for banks and REITs exclude Financials and Real Estate, so those sectors are not eligible for every list. Each screen reports the pool it actually screened: the names left after its own structural exclusions, counted before its pass/fail criteria are applied, so the number that qualified and the pool printed beside it describe the same set. Momentum Leaders is the one exception. Its figure counts every name that cleared the filters, of which the strongest ranks are listed, which is what its “names ranked” label reports and is not an eligible pool. Every name on a list publishes the math behind its place on it; a Free plan shows the top 5 per screen, Premium the full list. Naming & attribution: the classic screens are Tenet's independent, mechanical implementations of criteria the named investor or academic published; the timing and income screens are Tenet's own. The names are citations, not partnerships.

The Dividend Safety grade