Methodology

How we get our data

By DeltaScreener Research · Updated · Coverage figures computed 2026-10-14 across all 4,896 screenable tickers

A stock screener is only as good as the field behind each filter. This page documents where every number on DeltaScreener comes from, how it is calculated, how often it changes, how complete it is, and the specific cases where it is deliberately stored as not available rather than as a misleading number.

Nothing here is an opinion about any company. It is a description of a dataset and its limits, so you can judge for yourself whether a given filter means what you want it to mean.

1. The screenable universe

Every screen on this site runs against the same set of rows. A ticker is included only when all of the following are true, which is why the universe count is 4,896 rather than the larger number of symbols that exist on US exchanges:

  • it is flagged active in the symbol table, which is rebuilt once a day from the Nasdaq Trader symbol directory (its Nasdaq-listed and other-listed files), leaving out ETFs, test issues and names that are not common-stock-like;
  • it has a company name and a listed exchange;
  • it has a last price above zero;
  • it has a market cap above zero.

Delisted shells, symbols with no price and symbols with no market cap are therefore absent from every count on this site. The universe includes ADRs and foreign issuers listed on US exchanges, so it is not a US-domiciled-only list.

2. Sources, in order

Financial Modeling Prep is the primary provider. Each field is resolved independently: if the primary response is missing that one field, the record falls back to the next provider that has it — Yahoo Finance summary and quote data, then Finnhub, Alpha Vantage and Polygon. One row can therefore combine a price from one provider with a balance-sheet ratio from another.

Three other sources each cover one job: the Nasdaq Trader symbol directory decides which companies are listed, Yahoo Finance supplies the long-run daily price history behind the charts and the price-history pages, and SEC EDGAR supplies the list of recent filings linked from each stock page.

What that means in practice: two fields on the same row can be measured over different periods. The clearest case is P/E and EPS. EPS is a trailing figure; the P/E chain accepts a forward P/E as a fallback when no trailing P/E is available. When that happens the two are not two views of the same number, and the earnings yield shown alongside them is recomputed from the P/E so that at least those two always agree.

3. Update schedule

There is no single nightly batch. Several jobs run on their own clocks (times in UTC):

WhatWhenDetail
Quotes and market capevery 3 minutesThe 500 largest companies become due after 10 minutes, every other company after 2 hours. Up to 280 quotes are refreshed per run, most overdue first.
Fundamentals and ratiosevery 3 minutesThe same run works through the stalest companies: the 500 largest become due after 3 days, the rest after 7 days, and a company with incomplete statements is retried after 1 day.
Statements after a reportevery 2 hoursCompanies that reported in the last 1–10 days and whose stored quarters lack the reported quarter are rebuilt, up to 40 per run. When the provider has not published the new quarter yet, the company is retried after 6, 12, 24, 24 and 48 hours.
Price history22:30, Mon–FriAfter the US close, so the daily series ends on that day's close.
Symbol listonce a dayNew listings enter the universe; symbols no longer listed leave it.

A run is skipped if the previous one is still going. At any moment different tickers therefore carry different timestamps. Every stock page prints a "Data as of" line with the trading session of its quote and the latest quarter in its statements. If a figure here disagrees with a figure somewhere else, compare those dates first.

How key figures are computed

Trailing twelve months (TTM) and fiscal years. Annual tables show each fiscal year as the company reported it; a company whose year ends in September shows its September year. A TTM figure is the sum of the last four reported quarters, shown only when all four quarters are on file. EPS in the screener is a trailing 12-month figure.

P/E for loss-making companies is blank. A P/E computed from negative earnings is a negative number that sorts as though it were a low multiple, so it is stored as not available and shown as a dash. The same applies to a P/E that fails the limits in section 5.

Market cap is share price times shares outstanding, in US dollars. In the screener's query language it is entered in millions of dollars: Market Cap > 2000 means above $2 billion.

Currency for foreign filers. Prices and market cap are in US dollars. Financial statements are kept in the company's reporting currency — Taiwan Semiconductor reports in New Taiwan dollars — and every statement table names its currency. When the provider does not state the currency, it is inferred from the company's country and size; when it cannot be determined, the table says "reporting currency" instead of printing a dollar sign. Ratios that divide a US-dollar price by a per-share figure in another currency (common for ADRs) can be off by the exchange rate; the limits in section 5 drop the clearest of these cases.

Share classes. Each listed class — for example Alphabet's GOOGL and GOOG, or Berkshire Hathaway's BRK.B and BRK.A — has its own row in the screener, its own price and its own page. The rankings list a company once, using the class with the larger market cap, and the market heatmap shows one tile per company, using the more heavily traded class.

4. Field coverage

Coverage is not uniform, and a filter silently drops every row where its field is missing. Below is how many of the 4,896 companies actually reported each field on 2026-10-14. A screen on payout ratio is working with 7% of the market; a screen on market cap is working with all of it.

FieldRowsCoverageHow it is computed here
Market cap 4,896 100.0% Share price x shares outstanding, in USD.
Price 4,896 100.0% Last traded price in USD.
Beta 4,895 100.0% Provider-supplied 5-year beta against a broad US index.
Dividend yield 4,844 98.9% Trailing 12-month dividends per share divided by price, as a percentage.
EPS 4,820 98.4% Trailing 12-month earnings per share, in USD.
Profit after tax 4,801 98.1% Trailing 12-month net income, in USD millions.
Revenue growth 4,751 97.0% Latest full-year revenue against the prior full year, as a percentage.
ROA 4,716 96.3% Net income divided by total assets, as a percentage.
ROCE 4,686 95.7% Operating profit divided by capital employed, as a percentage.
Current ratio 4,535 92.6% Current assets divided by current liabilities.
Debt to equity 4,527 92.5% Total debt divided by shareholders equity.
Interest coverage 4,472 91.3% Operating profit divided by interest expense.
P/B 4,442 90.7% Price divided by book value per share.
P/S 4,427 90.4% Price divided by trailing 12-month revenue per share.
5-year average ROE 4,415 90.2% Mean of the last five reported annual ROE figures.
ROE 3,962 80.9% Net income divided by shareholders equity, as a percentage.
FCF yield 3,959 80.9% Free cash flow divided by market cap, as a percentage.
Operating margin 3,883 79.3% Operating profit divided by revenue, as a percentage.
Net margin 3,778 77.2% Net income divided by revenue, as a percentage.
PEG 3,572 73.0% P/E divided by an earnings growth rate.
Gross margin 3,541 72.3% Gross profit divided by revenue, as a percentage.
EV/EBITDA 2,824 57.7% Enterprise value divided by trailing EBITDA.
P/E 2,574 52.6% Price divided by trailing 12-month EPS. Null for companies reporting a net loss.
Payout ratio 364 7.4% Dividends paid divided by net income, as a percentage.

P/E coverage is 52.6% because 2,322 companies either reported a net loss or had no usable earnings figure. That is a property of the market, not a gap in the data: roughly 46% of the companies that reported net income reported a negative one.

5. When a ratio is stored as "not available"

Raw provider ratios contain values that are arithmetically real but economically meaningless: a 680% ROE on negative equity, a 125,000x price-to-sales on a zero-revenue shell, a 118% operating margin at a closed-end fund. Left in, they dominate every sort. Each stored row therefore passes through a fixed set of limits before it is saved. These are the rules, in full:

FieldRule
P/E Dropped when 0 or below, above 2,000, or below 0.5. A $1B+ company below 0.3x, or a $50B+ company below 2x, is treated as a currency mismatch between a USD price and local-currency EPS (common on ADRs) and both P/E and EPS are dropped.
ROE, 3y and 5y average ROE Dropped entirely when debt-to-equity is negative, because negative shareholders equity makes return on equity arithmetic rather than performance. Also dropped above 300% in absolute terms, and when its sign disagrees with reported net income.
P/B Dropped when negative or above 500.
EV/EBITDA Dropped when negative, above 1,000 in absolute terms, or below 0.5x on a company above $50B (a currency artefact).
Gross and operating margin Dropped above 100.5%. Operating profit cannot exceed revenue; closed-end funds that book investment gains as negative expenses otherwise report 115%+.
All three margins Dropped when trailing revenue is under $10M. Any margin on a pre-revenue company is a tiny-denominator artefact.
ROA / ROCE / net margin Dropped above 200% / 300% / 200% in absolute terms.
Dividend yield Dropped when negative or above 50%.
P/S and EV/Sales Dropped when negative or above 10,000. Zero-revenue shells otherwise report six-figure multiples.
FCF yield Dropped beyond plus or minus 100% of market cap.
PEG Dropped beyond plus or minus 100.
Revenue growth Dropped below -100% (revenue cannot shrink by more than all of it) or above 10,000%.
Payout ratio Dropped when negative or above 1,000%.
Earnings yield Recomputed as 100 divided by the sanitized P/E whenever the provider figure disagrees with it, so the two can never contradict each other on the same page.

A dropped field shows as an em dash, never as a zero. Zero and "not available" are different statements, and a filter treats them differently.

6. Known limits

  • Quotes are not real-time. Between refreshes a price can be up to about 10 minutes old for the 500 largest companies and up to about 2 hours old for the rest, on top of any delay at the provider.
  • Statements lag a new report. A quarter appears once the provider publishes it and the post-report job picks it up, which can take several days after the earnings release.
  • The coverage table above is a snapshot from 2026-10-14; coverage moves as companies report.
  • Analyst estimates on forecast pages are the provider's consensus figures, shown with the date they were retrieved; DeltaScreener produces no estimates of its own.
  • Provider errors pass through when they stay inside the limits in section 5. A figure that looks wrong is worth checking against the company's own filing — and worth reporting to us.

Three things to know before you screen

Negative debt-to-equity passes a "less than" filter. Debt-to-equity turns negative when shareholders equity is negative. A filter of Debt to Equity < 0.5 is satisfied by a company at -9.2, which carries more leverage, not less. On 2026-10-14 the query Dividend Yield > 4 AND Debt to Equity < 0.5 AND Market Cap > 2000 returned 51 companies, of which 10 had negative shareholders equity. Adding Debt to Equity > 0 returned 41. Run the guarded version.

A missing field is an exclusion, not a pass. Adding a filter on a sparse field shrinks the result set by far more than the threshold implies. Check the coverage table above before adding a payout-ratio, EV/EBITDA or free-cash-flow condition.

A screen name is a label, not a judgement. The prebuilt screens on this site are named after the query most people search for. The filters are printed on every screen page; where the name implies data the screen does not use — analyst ratings, insider transactions, short interest, dividend-increase streaks — the page says so explicitly.

Frequently Asked Questions

Where does DeltaScreener get its stock data?

Financial Modeling Prep is the primary provider for prices, statements and ratios. When a single field is missing from that response, the record is filled from a secondary provider — Yahoo Finance summary and quote data, Finnhub, Alpha Vantage or Polygon — field by field, in a fixed order. Long-run price history comes from Yahoo Finance, the list of listed companies from the Nasdaq Trader symbol directory, and filing links from SEC EDGAR. Every stock page prints the date of its quote and of its latest statements.

How often is the data refreshed?

A maintenance job runs every 3 minutes. Quotes for the 500 largest companies become due after 10 minutes and the rest after 2 hours, up to 280 quotes per run. Fundamentals for the 500 largest become due after 3 days and the rest after 7. A separate job every 2 hours rebuilds the statements of companies that reported in the last 10 days. Price history is updated after the US close on weekdays, and the symbol list once a day. Because this is a rolling queue rather than a single nightly batch, different tickers carry different timestamps at any given moment.

Why does a stock show no P/E ratio?

Because it reported a net loss. A P/E computed from negative earnings is a negative number that sorts as though it were a low multiple, so it is stored as not available instead. On 2026-10-14, 2,322 of the 4,896 companies in the database had no P/E for this reason or because earnings data was missing.

Why does a screen for low debt-to-equity return companies with heavy debt?

Because debt-to-equity goes negative when shareholders equity is negative, and a negative number satisfies any "less than" test. A filter of debt-to-equity below 0.5 therefore also admits a company at -9.2. Adding "Debt to Equity > 0" to the same query excludes them. On 2026-10-14, the query "Dividend Yield > 4 AND Debt to Equity < 0.5 AND Market Cap > 2000" returned 51 companies; adding the positive-equity guard returned 41.

See the data for yourself

Aggregate statistics for the whole universe, broken down by sector, are published with the same as-of date on the US market statistics page. To write a query in a sentence, see the plain-English screener guide. Or open the screener and set the filters yourself.

Open the screener →

This page describes a dataset. It is not investment advice and contains no view on any security. See the disclaimer and terms. Found a figure that looks wrong? Tell us — corrections are the fastest way this page improves.