How to read the condition and timing summary

Before you start: the words this guide uses¶

A company that is listed on a stock exchange divides its ownership into shares. Anyone can own a few of them, and their price moves every trading day. Two quite different questions can be asked about a listed company, and this product keeps them apart on purpose:

• What condition is the business in? Is it profitable, how much does it owe, does its profit turn into cash? The answers come from the company's financial statements — the accounts it must publish every year.
• Where does the share price sit against its own past? Is it above or below its long-run average, near its high for the year, moving more or less than usual? The answers come from the history of the share price and of the number of shares traded each day (the volume).

A company publishes three financial statements for each fiscal year (its accounting year, which does not always end on 31 December):

• the income statement — what the company sold (its revenue) and what was left as profit once its costs were paid, over the year;
• the balance sheet — what the company owns (its assets) and what it owes (its liabilities) on the last day of the year; what is left for the owners is the equity;
• the cash-flow statement — the cash that actually came in and went out during the year, split into operating (running the business), investing (buying equipment, buying other companies) and financing (borrowing, repaying, paying the owners).

A ratio divides one figure by another so that companies of different sizes can be compared: 50 million of profit on 1,000 million of assets is a 5% return. A threshold is the line a ratio is compared with. A check is one question with a yes-or-no answer — "is the return on assets above zero?" — together with its threshold and the place the threshold comes from. The summary asks about thirty such questions, grouped into six areas.

Every figure in the summary is computed by the product from data it has already stored, at the moment an administrator orders a recompute. Nothing is computed in your browser, and nothing is estimated where data is missing.

How to read the band¶

On a stock's Workspace page, directly under the price chart, the Company condition band answers the first question. Read it from the top:

1. The overall line — for example 70% · Good · mean of 6 areas. The percent is the average of the six area percents below it, each area weighing the same. Beneath it, labelled as such, is the pooled count — 22 of 31 checks passed across areas — and beside it the single worst failing check with its figure, so the weakest point is never hidden by a good average.
2. The company type and the sector benchmark — one line says which kind of company the rules treated this as (an ordinary operating company, a bank, a utility, a company not yet making a profit…) and where that came from; another names the sector benchmark: the group of similar companies whose typical figures some checks are compared with.
3. Six area tiles, always in the same order. Each carries a verdict word, its glyph and its percent, the literal n of m checks passed, one reason sentence naming the check that decided the tile with its figure in brackets, and a five-year history of small marks.
4. The evidence panel. Activate a tile — click it, tap it, or press Enter on its Show checks button — and one panel opens beneath the row. It lists every check of that area with its figure, unit, threshold, basis and state (passed, failed, not evaluable, not applicable), a bar showing where the figure sits between the thresholds, and a How this check works link to the check's own page. Only one panel is open at a time. Its last line names the rule set — the version of the check list and thresholds the area was judged by.
5. The Valuation strip — what investors currently pay for the company against its own past and its sector, in words only (see the Valuation strip).
6. The Why paragraph — what moved the company's cash last year (see the Why paragraph).

Every tile also has an (i) link. It opens the area's own page in a new tab — the six areas are listed in the next section.

Colour is never the only carrier of meaning. The verdict word and the glyph are always printed beside the colour; colour appears only on a tile's thin left rule, a chip's rule and the zones of a bar. If the green–red scale is hard to tell apart, the band's palette switch replaces it with a blue–orange scale; the choice is kept in this browser only.

The same verdicts appear in two other places, read from the same stored summary:

• The ticker bar at the top of every page of a stock shows two counts — Condition: 4 of 6 areas good and Timing: 2 of 4 good — followed by · 1 n/a when an area or tile has no verdict. "Good" here means Good or Strong. A small chip per area and per voting timing tile follows; each is a link to its tile.
• The Fundamentals and Technical pages open each panel with the verdict of the area that panel belongs to, and every indicator row carries a bar and an (i) link to its page. The figures those panels compute are unchanged.

The six areas¶

Each area asks one plain question about the business. Each has its own page with every check listed.

Valuation — what investors currently pay for the company — is not an area and never enters a percent. It is described separately, in words.

The percent, its verdict and the cap¶

An area's percent is checks passed ÷ checks that could run, rounded to a whole number (a half rounds up, so 79.5 reads 80). A check that could not run is left out of both numbers — it is never counted as a failure, and never as 0%. The percent is turned into a verdict word by one fixed table, the same for the six areas, the overall figure and the timing strip:

The word is always printed with its number: 67% · Good, never Good alone.

The overall figure is the plain average of the area percents that have one, printed as {percent}% · {word} · mean of {k} areas. Averaging areas rather than pooling every check keeps an area with more checks from weighing more than another.

The cap. Three findings are serious enough that a good average must not hide them. When any of them is present, the overall word is held at Mixed at most, and the band prints Mixed (capped — {flag}) beside the unchanged percent:

• the company's Altman Z-score is in the distress zone;
• its Beneish M-score is above −1.78;
• its sources of cash cover its uses less than 1.0 times.

A cap limits the word and the colour; it does not change the percent and it does not withhold the figure. Each of these models raises false alarms on a meaningful share of healthy companies (about one in seven, for the Beneish line), which is why they cap rather than veto.

When the overall figure is withheld. The band prints No verdict — {reason} instead of an overall figure when:

• fewer than two consecutive fiscal years carry all three statements;
• any of the four core areas — Profitability, Debt & solvency, Cash & liquidity, Earnings quality — has fewer than three checks that could run;
• the company is a bank, an insurer or a holding company (their accounts need checks of their own, which are not written yet), or its type could not be resolved;
• its latest fiscal year ended more than 18 months before the computation;
• no cash-flow statement is stored.

The area tiles that could reach a verdict still show it when the overall figure is withheld.

What a verdict is and is not¶

A verdict is a description of published figures against stated thresholds. Good on Debt & solvency says that most of that area's checks that could run were met by the figures in the company's latest published statements. Every threshold has a named source and was fitted on a named group of companies, which each check's page states.

A verdict is not:

• an instruction to trade the shares, in either direction, or at any price;
• a forecast of the share price, of the company's results, or of anything else;
• a judgement of whether the shares are priced too high or too low — valuation is described, never scored;
• a rank against other companies — the summary never compares one company's percent with another's;
• a complete picture: it covers what the statements and the price history can show, and nothing about the products, the management or the industry.

Published screens like these were fitted on particular groups of companies in particular decades, and they misclassify some companies even there. Read a verdict as a pointer to what deserves a closer look, then open the evidence panel and the check's page to see the figure behind it.

The Price & timing strip¶

Beneath the band, the Price & timing strip answers the second question: where the share price sits against its own past. It is captioned checks passed (price & timing) and carries a fixed sentence: Describes where the price sits against its own past; it forecasts nothing. It is never combined with the condition figure — a sound company in a falling market and a weak company in a rally read differently at a glance.

The strip has six tiles. Four of them vote; a vote is one yes-or-no question about the price history:

• Trend — three votes: is the price above a rising 30-week average, above a rising 200-day average, and is the 50-day average above the 200-day? The tile is labelled with the Weinstein stage.
• Momentum — two votes: did the price rise over the twelve months to a month ago, and is it within 25% of its high of the past year?
• Relative strength — one vote: is the share's strength against the market's benchmark above its own average of the past year?
• Participation — one vote: has trading volume over the past week been heavy on a rise?
• Volatility and Key levels describe and never vote: how much the price moves day to day, and how far it is from nearby price levels.

The strip's percent is votes passed ÷ votes that apply, printed as n of m votes passed, and mapped to a verdict word by the same table. Trend and Momentum carry a verdict word of their own; Relative strength and Participation, with one vote each, carry a plain noun instead (lagging · in line · leading; quiet · normal · heavy), while their vote still counts in the strip.

The strip needs enough history: at least 50 daily price bars for the strip at all, 200 daily or 34 weekly bars for Trend, 252 daily bars (about a year) for Momentum, and a stored benchmark series for Relative strength. A tile without enough says No verdict — {what is missing} and leaves its votes out of the count. When the newest stored daily price is older than the previous weekday, the whole strip is marked stale with the date of that price; the verdicts stay and are marked, never presented as current.

The Why paragraph¶

The band ends with one sentence about the company's cash in its latest fiscal year, built from the cash-flow statement alone: how much cash rose or fell, then the largest movements that explain it, in order of size, until four fifths of the change is explained. For example: Cash rose 2.6 billion in fiscal 2025: operating cash flow (6.4 billion) alongside investing cash flow (−2.7 billion) and financing cash flow (−2.0 billion).

The movements are joined by alongside, never by to fund or because: the statement records that the amounts moved in the same year, not that one paid for the other. The paragraph never states a business reason — why a factory was built or a company bought — because the statements do not say.

Some areas carry sentences of the same kind in their evidence panel, each with a fixed band:

• free cash flow against profit — between 0.8 and 1.2 times reads in line;
• operating cash flow against profit — below 0.8 times in two consecutive years is flagged;
• capital spending against the depreciation of existing equipment — below 0.9 times reads harvesting, above 1.5 times building;
• money paid to the owners against three years of free cash flow — 1.0 times or less reads funded;
• the change in earnings per share, split into the part from profit and the part from the number of shares;
• the change in the operating margin, named by its largest cost line;
• the change in net debt.

The Valuation strip¶

Under the tiles, four valuation multiples — what investors pay for a unit of the company — are described against the company's own past and its sector: the price against a year's profit (P/E), the whole company's value against its operating profit (EV/EBIT), the price against the owners' equity on the balance sheet (P/B), and the company's value against its revenue. Each is placed above, within or below its own five-year band — the same multiple at each of the last five fiscal year-ends, within meaning between the 10th and the 90th percentile of those values — and against its sector: within the middle half of similar companies where the benchmark gives quartiles, otherwise within 10% of their median. The reference lines Benjamin Graham published (15 times profit, 1.5 times equity, and the two multiplied not above 22.5) are drawn as labelled lines. The strip uses only those positional words and never enters any percent. A range needs at least three stored year-ends; with fewer, the strip says why — how many year-ends had what the multiple needs, or a share split the older year-ends cannot be priced across.

Each figure names what it divides by and the year that comes from — P/E 45.31× on FY2025 net income. A multiple without a figure says why in its place, and no figure is worked out in the browser instead: the share count has moved beyond its split, so today's market value and the statements describe different numbers of shares; the profit, the operating profit or the equity is not positive or not stored; or, for a bank or an insurer, the value of the whole company means nothing.

No verdict, stale and not computed¶

No verdict on a tile. An area needs at least three checks that could run before it gets a percent. Below that the tile shows the hollow glyph ○, No verdict — {the missing input} and n of m checks could run, with no percent and no coloured rule. There are two kinds:

• not applicable — the check does not make sense for this company: a debt ratio for a company with no debt, a volume reading for a market that publishes no volume. That is a fact about the company, not a gap in the data;
• missing — the check makes sense but an input it needs is not stored, and the input is named: No verdict — net property, plant and equipment FY2025.

A bank, an insurer and a holding company read No verdict — bank checks not yet available (or insurer, or holding company) on every area: the ordinary checks do not describe their accounts, and their own checks are a later piece of work. Their timing still shows, and so do their P/E and P/B. A bank and an insurer show no EV/EBIT and no EV/sales: their debt and cash are the business itself, so the value of the whole company means nothing for them, and the strip says so in their place. None of the three gets a Why paragraph — a bank's deposits, loans and trading assets run through its cash-flow statement, so the sentences would describe how it is funded, not its condition — and the band says so in one line.

Stale. The stored summary was computed from data that has since moved on: a newer set of statements is stored, the condition half is more than 190 days old, a sector benchmark it used is more than 400 days old, or a newer daily price exists — or the checks themselves have changed since it was computed (a newer rule set). The band keeps every verdict and shows a badge — stale — computed 12 days ago — naming the input that moved. The age is that half's own: the condition half and the price & timing half are computed at different times. The condition half's age is counted from the oldest of the inputs it used — the date the band's as of prints — and the price & timing half's from its newest daily price.

Not computed yet. No summary is stored for the company, or it is stored in a format this page does not read. The band says so in one sentence and shows no verdict at all; nothing is estimated in its place. An administrator also sees how to order a recompute.

Unsupported market. A company listed only on a market for which no data provider has been chosen shows a statement saying so, and no summary at all.

The five-year history¶

Every tile ends with five small marks, the oldest on the left: the verdict the same rules give each of the five most recent fiscal years, computed from that year's statements and the years before it. Each mark carries the year's colour and its glyph inside a darker border; a year without a verdict is a hollow, dashed mark. The marks are not buttons: hover over one or tap it to read the year and its result, and the evidence panel lists all five in a table (year · verdict · n of m checks passed). The timing strip's history works the same way by calendar year.

The history is recomputed from the stored statements every time; it is not a record of earlier verdicts, so a corrected statement changes the history too.

“n of m checks passed” and “n of m checks could run”¶

The two literals are deliberately different:

• n of m checks passed — printed on a tile that has a verdict: 3 of 4 checks passed means four checks could run and three of them were met. The percent beside it is the same fraction, 75%.
• n of m checks could run — printed on a tile without one: 2 of 5 checks could run means only two of the area's five checks had their inputs, which is below the three an area needs.

On the timing strip the same idea reads n of m votes passed. A Trend tile whose weekly and daily readings disagree scores those two votes as half a vote each — one vote between them, whichever clock passed — and reads Mixed whatever its count: Linde's Trend tile on 23 October 2026 read Mixed with 1 of 3 votes passed.

Where the thresholds come from¶

Each check's page gives its threshold, the study or convention it comes from, the group of companies it was fitted on, and an evidence grade:

• peer-reviewed and replicated — published in a reviewed journal and found again by other researchers;
• peer-reviewed once — published, not yet independently repeated;
• practitioner convention — the line rating agencies, analysts or data platforms use, with no published test;
• folklore — a widely repeated rule of thumb with no known source testing it.

A check judges its figure on one of three bases:

• sector band — against the typical figures of similar companies (the sector benchmark), for checks of a level, such as debt against operating profit, where what is normal differs greatly between industries;
• own history — against the company's own previous years, for checks of a change, such as "did the return on assets improve?";
• absolute — against a fixed line, for checks where the line does not depend on the industry, such as "is the current ratio at least 1.0?".

For a level check, the band shows the sector-band state as the verdict and the own-history state beside it. The sector benchmark is the narrowest group of similar companies that has at least ten members, climbing from the industry to the sector to wider groups. While no benchmark is stored for a company's group, its level checks read not evaluable — no sector benchmark and are left out of the count.

Every page of the documentation¶

Profitability & returns — the area: ROA positive · CFO positive · ROA improving · Gross profitability · Value creation

Debt & solvency — the area: Leverage · Coverage · Leverage not rising · Altman zone · Debt issuance

Cash & liquidity — the area: Liquidity improving · Current ratio floor · Free cash flow · Sources and uses · Cash conversion

Earnings quality & red flags — the area: Accruals · Manipulation screen · Misstatement screen · Cooking-the-books count

Growth & reinvestment quality — the area: Gross profitability trend · Asset growth discipline · Growth paid for · Efficiency improving · Rule of 40 · Burn multiple · Runway

Capital allocation — the area: No dilution · Payout funded · Acquisition discipline · Incremental returns

The Price & timing tiles: Trend · Momentum · Relative strength · Participation · Volatility · Key levels

Indicators on the Fundamentals page — valuation: P/E · P/B · P/S · P/FCF · EV/EBITDA · EV/EBIT; returns and margins: ROE · ROA · ROIC · ROCE · Gross margin · Operating margin · Net margin · EBITDA margin · Free-cash-flow margin · Effective tax rate · DuPont decomposition; debt and liquidity: Debt to equity · Net debt to EBITDA · Interest coverage · Current ratio · Quick ratio · Cash ratio; efficiency: Asset turnover · DSO · DIO · DPO · Cash conversion cycle; earnings quality: Sloan accruals ratio · Operating cash flow against net income · Beneish M-Score · Altman Z-Score · Piotroski F-Score; capital and growth: Capital expenditure to depreciation · Stock-based compensation to revenue · Observed revenue growth · Reinvestment rate · RONIC · CAGR · ROIC against WACC · Moat evidence

Indicators on the Technical page: RSI · MACD · ATR · ATR % · ADX · ROC · ADR % · RVOL · Weinstein stage analysis · Minervini Trend Template · Mansfield relative strength · RSraw · 1R and stop levels