How performance is scored

Every verdict on TraderPunk is scored automatically against real market prices, using the fixed rule on this page. Results are never picked by hand — one rule applies to every report, every trading day, and the prices and bands each result was judged against are stored and never edited. If the rule itself ever changes, the change is dated in the revisions list at the bottom of this page and applied to the whole history at once, never selectively.

The scoring window

A report's verdict is scored across exactly one trading session: from the closing price of its entry session — the last completed NYSE session when the report was generated — to the closing price of the next trading session. Prices are split- and dividend-adjusted. Each report is scored once, and scores land nightly after market close, Monday through Friday.

Win, loss, and push

• BUY wins when the next-session return is above +0.5%, and loses when it is below −0.5%.
• SELL wins when the return is below −0.5%, and loses when it is above +0.5%.
• A BUY or SELL whose move stays inside that ±0.5% band is a push — the market did not move enough to prove the call right or wrong. A push counts as neither a win nor a loss.
• HOLD wins when the move stays within the stock's own typical daily range: one standard deviation of its last 20 daily returns, never tighter than ±0.5%; otherwise it loses. A stock with fewer than 10 days of price history uses a fixed ±1.5% band instead. HOLD never pushes — staying inside the band is exactly what a HOLD predicts.
• OVERWEIGHT wins when the stock beats the market — its next-session return minus the universe median return is above +0.5% — and loses below −0.5%; in between it is a push.
• UNDERWEIGHT wins when the stock lags the market — its next-session return minus the universe median return is below −0.5% — and loses above +0.5%; in between it is a push, because outperform-the-market and lag-the-market are the claims those words make.

The hit rate is wins divided by decisive calls (wins plus losses), and it is always published beside the full win–loss–push record, so the pushes it excludes are never hidden. Every outcome stores the exact band it was judged against, so a future change to the rule can never rewrite a past result.

Simulated following

Each stock page can show what a reader would hold today had they followed every call on that stock from a chosen date, with a notional $10,000. This is a different rule from scoring, on purpose: a report is generated after the close it priced, so the simulation fills at the next session's opening price, the first price a reader could actually trade. BUY and OVERWEIGHT hold the stock, UNDERWEIGHT and SELL hold cash, and HOLD keeps whatever is already held. Nothing is shorted, no commissions or slippage are applied, and buy-and-hold over the same window is always shown beside the result. The simulation runs only over calls the viewer can read: it ends at the first call that is still locked, and says how many follow it. Below three completed round trips the block says the sample is too thin to judge.

Methodology revisions

• 2026-08-25 — Dead-band results on BUY and SELL calls are now recorded as pushes. They previously counted as losses, which scored a +0.3% move after a BUY as a failed call while the identical move under HOLD counted as a win. The stored returns and bands of past outcomes are unchanged; every historical verdict was re-stated under the same single rule.
• 2026-08-26 — Ratings now published on the research pipeline's own five-point scale (Buy / Overweight / Hold / Underweight / Sell). Previously Overweight and Underweight reports were displayed and scored as BUY, HOLD or SELL — an artifact of a data-extraction step, not a judgment of the research — so 506 historical reports were re-labelled to Underweight (498 previously shown as SELL, 8 as HOLD) and 97 to Overweight (previously shown as BUY), from the stored original reports. Overweight and Underweight are scored against the market (the median return of every stock we cover) from this date and retroactively, with the benchmark stored on every result.
• 2026-09-06 — Added the simulated-following block to each stock page, with the fill and position rules stated above. It introduces no change to how verdicts are scored.

Who sees results when

The site-wide totals on the performance page count every verdict we have ever scored. Individual report results follow the same visibility as the report itself: subscribers see them same-day, free accounts see a report once it has been scored — usually the next trading session — and everyone can read any report once it is 7 days old.

Every score covers a single trading day. A hit rate is not a return forecast, and past performance does not predict future results — see the full disclaimer.