Every score has
a paper trail.

What the scores measure

These are research heuristics for disclosed purchases, not actual portfolio returns. Disclosures usually supply amount ranges and dates, not exact execution prices, share counts, complete holdings, or who made the decision. Spouse and dependent transactions remain identified in the evidence. No trade execution is offered.

Political trading performance

We group purchases by verified filer, security, and transaction date. Repeated same-day rows count as one equal-weight cohort. The outcome is the security's transaction-session closing price return to the close 63 trading sessions later, minus SPY's price return over the same sessions. A nontrading transaction date uses the next session. Only completed horizons enter the score; unresolved identities, corporate actions, missing prices, and amendments remain omissions.

At least ten valid cohorts and 80% valid coverage of matured cohorts are required. For n valid cohorts, A = clamp(50 + 250 × mean excess return, 0, 100); H = 100 × fraction with positive excess return; P = 50 + n/(n+20) × (0.70A + 0.30H − 50). Each page shows the weighted contributions, sample, omissions, and shrink factor. Price series must use the same verified split convention; dividends are excluded.

Separately labeled post-publication outcomes use the next session's open after the information date to the close 63 sessions later. They never enter the political trader score. Unknown historical public-availability times do not become invented backtests.

Weekly trade ideas

The weekly cohort uses disclosures available within the seven days before the actual report cutoff. Backfilled old filings do not become new purchases. Each eligible political trader contributes once per security. Q is their mean trader score; B = 100 × min(distinct eligible buyers / 5, 1); R = 100 × 2^(−median transaction age in days / 14). The security score is S = 0.60Q + 0.25B + 0.15R. A score of at least 60 and verified inputs is required. Ties use buyer count, newest information, and permanent security ID. Recency uses the trade date, not discovery time.

Options filters

We research standard long calls with 45–90 days to expiration, delta 0.40–0.60, open interest at least 100, same-session volume at least 10, positive bid, ask at least bid, spread/mid at most 10%, and a $500 one-contract premium cap. Only a standard 100-share deliverable is supported. Underlying and contract quotes must be no older than 15 minutes, Greeks no older than 90 minutes, and all timestamps must be usable. Unknown timezones, stale data, market closures, or missing fields withhold the contract.

O = 0.50S + 0.30L + 0.20D, where L = 100 × (1 − spread ratio / 0.10) and D = 100 × (1 − |delta − 0.50| / 0.10), each clamped to 0–100. Ties use spread, open interest, premium, and OCC symbol. At most three contracts are published, with one per security. No qualifying contracts is a valid result. Delta is not a probability of profit. One long call can lose its entire premium; stated breakeven and maximum contractual loss exclude fees.

Coverage and corrections

House, Senate, and already-published White House reports are collected independently. Annual holdings never become invented purchases. Original documents, parser versions, source references, calculation inputs, and run history are retained in one database. Unknown layouts and unresolved amendments are quarantined. Published coverage gaps remain visible; a missing report does not show an absence of activity. Historical editions are frozen; a correction creates a new edition.

The service collects on Monday morning and prepares the weekly edition for 3 p.m. New York time using the actual available cutoff. The email contains the same frozen report as the website. Delayed reports can change the evidence later; no score predicts future returns or establishes wrongdoing.