Buy when a CFO buys at least $100k as part of a cluster, hold for a quarter. Does it work? The Backtest engine answers in seconds, on every qualifying filing of the last two years, with the outcome measured against the S&P 500 over the same weeks.
Three choices define a rule. Who: anyone, an officer or director, a CEO, a CFO, or a CEO/CFO/chair. How big: any size, or a floor of $50k, $100k, $250k or $1M. Which signal: part of a cluster, the insider's first-ever buy, or a buy at least 25% below the recent high. Then how long: 4, 9, 13, 26 or 52 weeks, equal-weighted or weighted by dollars bought.
The number of trades, the win rate, the average and median market-adjusted return, the spread of outcomes, an equity curve of the rule against SPY, and every trade — ticker, insider, filing date, entry and exit price, return, excess. Returns are market-adjusted: each trade's stock return minus SPY over the same weeks. Entry is the Friday close of the filing week, because the tool uses weekly closes and that is the first close you could plausibly have traded.
Two years is a short sample and it contains one market regime. A rule with twelve trades and a 75% win rate has proven nothing. The engine is for finding out whether a pattern you believe in even shows up in the data — not for optimising a strategy, which with this many knobs and this little history is a way to fit noise. The seven-year data archive and the API are there for the longer test.
All seven tools, the full two-year feed, the JSON API and email alerts, refreshed every morning. A free quarterly sample of the dataset needs no account.
See plansFree sampleAny combination of who buys (anyone, officer/director, CEO, CFO, CEO/CFO/chair), a minimum size, a signal (cluster, first buy, dip) and a holding period from 4 to 52 weeks, on every qualifying open-market purchase of the last two years.
Yes. Every trade is the stock's return minus SPY over the same weeks, so a rule is not credited for a rising market.
At the Friday close of the week the Form 4 was filed — the first weekly close after the information was public.
Rarely. Two years, one regime and a few dozen trades is a small sample. Use it to check whether an idea shows up at all, not to tune one.
The members API serves the trailing 24 months. The one-time $49 data archive holds every filing since 2019 as CSV and SQLite, with a build script and starter queries, for backtests on your own machine.