Track record

Past performance is not indicative of future results — this is not investment advice. Every stat below shows its sample size (n); a median from a handful of events is not as meaningful as one from hundreds. Figures marked "vs S&P 500" compare each event's own forward return against the S&P 500's return over the same window, starting the same day — without a benchmark, a raw return is close to meaningless.

Benchmark (S&P 500) comparison data is not yet available — raw returns below are still shown.

Methodology — how these numbers are calculated

We don't ask you to trust our math — here's exactly how every number below is computed, so you can check it yourself.

• Price data source: daily closing prices from Tiingo, a third-party market data provider — not computed or estimated by us.
• Baseline price: the closing price on the transaction date itself, or the next available trading day's close if the transaction date falls on a weekend/holiday or has no price within 5 calendar days.
• Return at horizon N (1d/3d/7d/30d): the closing price nearest N calendar days after the baseline date (same 5-day lookahead tolerance), compared to the baseline price: (price at N days − baseline price) / baseline price.
• Excess return vs. S&P 500: the S&P 500 ETF (SPY)'s own return over the identical baseline date and horizon, subtracted from the transaction's own return — the same formula, applied to the same dates, for both.
• Congressional trades (STOCK Act PTRs): returns measured from disclosure date, not transaction date. Unlike Form 4 (filed within ~2 business days of a trade), members of Congress have up to 45 days to disclose a trade. Measuring returns from the transaction date would overstate how actionable the signal was for anyone reading the disclosure — the stock may have already moved during the lag. We measure forward returns starting from the disclosure/notification date instead, and we show the lag itself (median days between trade and disclosure) as its own disclosed stat below.
• Missing data is excluded, not hidden or zero-filled. A transaction with no price available within tolerance for a given horizon simply isn't counted in that horizon's sample — this is why n can differ across horizons and between the raw and excess-return columns for the same segment.
• Median, not mean. A single large outlier can't dominate the headline number; win rate is the fraction of transactions with a strictly positive return.

Performance by signal type

A transaction that's part of a detected cluster is counted only under its cluster segment, never also under the plain buy/sale segment, so each row reflects an independent population. See what's a cluster alert? for the exact rule behind cluster detection.

Congressional trading

Every purchase/sale disclosed under the STOCK Act (Senate + House Periodic Transaction Reports), tracked the same disclosed, benchmarked way as everything else on this page — sample sizes shown, no undisclosed scoring. Median gap between trade and public disclosure across 3552 tracked trades: 16 days.

Insider & Congress overlap

A discovery stat, not a performance claim: how often an SEC-reporting insider and a member of Congress independently traded the same stock, same direction, within a few days of each other — matched purely on ticker, never on identity. We do not claim, imply, or attempt to determine whether these are the same person or are connected in any way; each entry links to both original source filings so you can verify it yourself.

99 congressional trades detected trading the same stock, same direction, as an SEC-reporting insider within 5 days.

Full event log

Showing 100 of 55134 events, most recent first.