The Whistle
Trades whose timing was statistically unusual for that security — and which also sat against committee work the member was part of, a vote they cast, or a shutdown or lame-duck window.
What "unusual" means here, exactly. For each trade we ask a testable question: if this member had traded this same security on a random day instead, how often would they have done at least this well against the market over the next 90 trading days? Two hundred random dates are drawn from that security's own price history. The p-value is the share of them that did as well or better. A p-value of 0.02 means only 2% of random timings in that stock beat this one.
What it does not mean. It is not a probability of guilt, or of innocence. It does not measure intent, knowledge, or wrongdoing. Someone has to be in the top few percent — across tens of thousands of trades, unusual things happen constantly by chance alone. That is precisely why an unusual p-value on its own is not flagged here: a Whistle also requires a documented connection, and a size floor of $15,000 so token amounts do not generate noise.
Across 29,227 trades tested, 2,361 (8.1%) were unusual at the 0.05 level. Chance alone produces about 5%. The gap between those two numbers is the whole finding — and it is modest.
The published rules
| Rule | What it requires | Flags |
|---|---|---|
| huddle_timing | Unusually timed trade inside a committee window, p ≤ 0.05, at least $15,000 | 75 |
| window_timing | Unusually timed trade during a shutdown or lame-duck window, p ≤ 0.05, at least $15,000 | 48 |
Every threshold is fixed and published — nothing here is tuned per member or per trade. The full definitions are on Methodology.
0 flagged trades.
Nothing flagged yet. Statistical scoring runs across the whole trade database and the rules are applied afterwards — an empty page here means the analysis has not finished, not that nothing qualifies.
A flag is a prompt to look, not a conclusion. Definitions: The Whistle and The Line. Method and every threshold: Methodology.