How we research, source and correct.

Yuriy Matso designs the measures, writes the research and reviews what is published. His background is on his author page. The site is independent: it is not owned by a broker, fund or data vendor, and it publishes no investment advice.

Each page starts from a question a reader actually asks, such as “is the rally broad?” or “what happened the last time credit spreads did this?”, and answers it with data, a date and a sample size.

• Tested before it ships. Every “what happened next” study is run against the unconditional baseline, every day in the same window, before the page exists.
• Results published whichever way they point. When a famous indicator turns out to predict nothing, or a worrying setup was followed by better-than-average returns, the page says so. Rejected ideas are documented, not dropped.
• History, not forecasts. Base rates describe what followed similar conditions. Small samples are labelled as small, and overlapping episodes are counted as such.
• Survivorship stated. A history built from today’s index members makes the past look better than it was. Those series are shown as labelled exhibits and never used for a forward study.
• Opinion is labelled. The tools measure. Where an essay or letter takes a view, it says it is the author’s view and names what would change it.

• Computed, never typed. Every number in a reading, caption, FAQ or table is calculated from the data when the page is built.
• Dated. Each reading names the session or observation it comes from. Economic data enter on the day they were released, so no study uses a figure before the market could have seen it.
• Freshness-gated. The daily pipeline refuses to compute on incomplete data, and a page that falls behind shows a notice.
• Deviations disclosed. Where a calculation departs from a textbook specification, the page says how and why.

The calculations themselves are on each tool’s page and summarised in the methodology. When a method changes in a way that moves a published reading, the change is recorded on Product & Research Updates.

Every page names the series it is built from. The main sources are:

Prices and market internals

Our own database of daily bars for about 5,500 US-listed symbols, licensed from TradeStation and Massive, plus intraday futures bars from TradeStation.

Company filings

SEC EDGAR: 13F holdings for the fund pages, and 10-K and 10-Q financial statements for the valuation pages.

Economic data

FRED and ALFRED at the Federal Reserve Bank of St. Louis, which carry the BLS, BEA, Census and Federal Reserve series, with the release date of each print.

Government finance

U.S. Treasury Fiscal Data, the Monthly Treasury Statement and OMB historical tables.

Derivatives and positioning

Cboe volatility indexes and put/call data, OCC volume, CFTC Commitments of Traders and swaps reports, FINRA short interest and margin statistics.

Energy and global data

The U.S. Energy Information Administration, OECD, IMF and World Bank.

Long history

The Kenneth R. French Data Library (CRSP portfolios back to 1926) and Robert Shiller's market data back to 1871.

Where a source restricts redistribution, we publish what its terms allow and say what is missing. The datasets behind the pages are listed in the data catalog.

No company, fund or advertiser pays for coverage, a reading or a ranking. Sponsorship, when sold, is a fixed monthly fee for a clearly labelled unit, kept visibly apart from answer cards, rankings and methodology, and it never buys editorial content. Terms are on the advertise page.

Tool pages carry one placement for System Trader, a separate site run by the same founder. It is labelled Sponsored and marked as such for search engines, because a reader cannot otherwise see that relationship.

AI helps build the tools and draft the writing. It never supplies a number: every figure comes from the data pipeline, and a person reviews each page before it is published. The details are in how we use AI.

• On the page. A factual error is corrected on the page where it appeared, and the page’s modified date moves with it. Nothing is changed silently.
• On the record. A correction that changes a published reading, a study’s result or a historical series is logged on Product & Research Updates with what changed and why.
• In the letter. An email cannot be edited after it is sent, so an error in a letter is corrected plainly in the next one.
• Data revisions are not errors. Agencies revise economic data. Pages show the latest values and say when history uses revised figures.

Found an error? Email info@thetrading.tools with the page and the figure.

• Insights archive: every research essay, dated.
• Market Questions: dated answers built from several tools at once.
• Product & Research Updates: launches, method changes and corrections.
• The Research Letter, published on Substack.