How we compute our numbers
Every backtest figure, screen and ranking on this site is computed by our own engine over real market data — the same engine that powers the tools, so anything an article claims, you can re-run yourself and get the same answer. This page is the contract: what the engine does, where the data comes from, and the things we deliberately refuse to do.
The backtest engine
When an article quotes what a strategy returned, that number came from the backtester — not from a spreadsheet, another site, or memory. Every run we publish includes a trading fee and slippage on each trade (in the tool they are yours to set, including to zero — the caveat list on every result says exactly what was and wasn’t counted). Results are always shown next to buy-and-hold over the same window, and the trade count is shown with the return, because a result built on a handful of trades is an anecdote whatever it says. Where an article embeds a result, it is a live run that refreshes as new bars land — we learned early that a table typed into prose disagrees with the live tool within weeks, so we stopped typing them.
Congress copy-returns
Congress trades are read from the official House Clerk and Senate eFD filings, daily, back to 2012. Every copy-return starts at the disclosure date— the first opening price after a filing became public — because that is the earliest any real person could have acted, and it is usually a very different price from the member’s own. Position sizes are equal across trades: a filing only discloses an amount band, so any site quoting precise position sizes is filling in a blank the law left empty. Filings that exist only as paper scans are counted and shown as a coverage number rather than silently skipped.
Screens and fundamentals
The value screen reads as-reported figures from SEC filings — the numbers a company actually filed, stamped with the date they became public. Its governing rule is abstain, never substitute: when a company doesn’t report a figure the screen needs, it is listed as unscored with the missing figure named, because a fabricated zero would quietly rank it worst-in-class. The same rule holds everywhere on the site — a missing number is shown as missing, never invented.
Where the data comes from
Daily prices from major market-data feeds, checked for staleness against each series’ own latest bar. Currency conversion at ECB reference rates. Fundamentals from SEC EDGAR as filed. Congressional disclosures from the primary government sources, with a small community-transcribed backfill for old paper filings — every such row tagged, and our own read of a document always wins where we have one.
What we refuse to do
No present-tense performance promises — we report what a strategy would havereturned, past tense, with the costs stated. No cherry-picked windows: a result states its own start and end dates, and they are set by the data, not chosen to flatter the strategy. No pay-to-rank: nothing on any ranking or review is placed there for money. Where an article carries affiliate links, it says so at the top, and the disclosure is generated from the article’s own links rather than a flag someone has to remember to set — and we say where a product is the wrong choice.
Who runs this
Investing Paths and Pulse are built and run by one person, who tracks his own portfolio in Pulse and eats the same cooking — the engine that scores these pages is the one his own money is measured with. Found a number that looks wrong? Tell us — a methodology page is only worth anything if it can be held to.