61 measures across five dimensions, each computed from a company's own regulatory filings and nothing else.
No opinion, no research and no news goes into any of them. Each page below gives the formula, the figures it needs, and the XBRL tags those figures are read from.
How much of the revenue is really kept — and what did it take to earn it?
What it covers. Four separate things, and the inputs you have switched on decide which of them your score is actually about: how much of the revenue survives into profit (the margins), whether that profit arrives as cash (cash conversion, accruals), what capital it took to produce (return on capital, return on assets and equity, asset turnover, capital expenditure), and how much of it goes to staff rather than owners (stock compensation). Built from margins alone this answers a narrower question than it does with all fourteen inputs in. A high score means every measure you asked for came out well — not that the company is good in some general sense.Why GAAP. Stock compensation counts as the payroll expense it is. That single choice separates companies whose profit is real from those whose profit is paid in dilution, and it is why the same cost appears twice: once on its own and once inside owner earnings. The cash operating margin is the one concession, at a tenth of the dimension, because GAAP alone cannot tell a loss-maker with a real business underneath it from one with nothing underneath it at all.
What does the price ask for what the company earns, owns and pays out?
What it covers. The price, held against whatever the company can be measured by, and the inputs you have switched on decide which: operating profit and revenue, owner earnings and free cash flow, gross profit, earnings past and forecast, the assets and cash on the books, and the dividends and buybacks handed back. Cheap on one and dear on another is the normal case, and is the reason for having more than one. A high score says the price is low against the measures you chose; it says nothing about whether the business deserves a low price — that is the other four dimensions.How the numbers are treated. Enterprise value already nets out cash, so leverage is not counted twice here; it sits under Resilience. Multiples are scored logarithmically, because doubling a multiple halves the earnings yield whether it goes from 20x to 40x or from 40x to 80x, and a linear scale spent most of its range on the difference between expensive and very expensive. EV/EBITDA is deliberately not in the catalogue at all: it is the first measure above, except where depreciation is large, and there it forgives precisely the cost this model counts on purpose. The free cash flow yield that is in the catalogue forgives stock compensation in the same way — add it to see how much of a company's valuation rests on not counting it, not as a second opinion on the price.
How predictable is the revenue, and how much of it is already committed?
A high multiple buys growth duration, not growth rate. The worst year is used rather than the variability of growth, because variability punishes a strong year exactly as hard as a weak one, and a company that has never shrunk is not erratic. Deferred revenue is not a second reading of contracted revenue: under the revenue standard it is the billed part of the same obligations, so it stands in only where a filer discloses no backlog. Weights rebalance over whatever a filer tags, and the detail panel says how many inputs were available.
Can the balance sheet absorb a bad year?
Deferred revenue is taken out of current liabilities because it is cash already collected, not a claim on cash; left in, a business that is paid in advance scores its strength as distress. Margin stability used to sit here as a fourth input and was removed for measuring the same thing as the two stability inputs under Durability. Expect this dimension to cluster near the top: a balance sheet is a hurdle rather than a scoreboard, and its job is to catch the exceptions. It does not capture customer concentration, regulatory dependency or management turnover, which are usually what actually breaks a company.
Is the business getting bigger, and is it getting more profitable?
Per share, because dilution is the difference between the business growing and your claim on it growing, and that difference is invisible in total revenue. Over one year and four, because a single year cannot tell a compounder from a rebound off a collapse. Margin is measured in percentage points rather than percent, because a percentage cannot cross zero, so a narrowing loss would register as no improvement, and because percentage growth flatters a small starting base. The forward estimate is the only figure in the model that does not come from a filing: analyst consensus is not in XBRL, so it is entered by hand, carries a date, and the dimension rebalances without it.