How the Sharp Score works
Raw profit crowns lucky whales. The Sharp Score exists to separate repeatable skill from one trade that happened to hit. Everything below is computed from Polymarket's public, on-chain-backed records, you can verify any wallet's numbers yourself via the profile links we show on every row.
The inputs (Sharp Score v1)
- Profit (PnL).The wallet's profit over the selected window (weekly / monthly / all-time), from Polymarket's official leaderboard data. Scaled with a saturating curve so the score rewards consistent scale without letting a single mega-win dominate.
- Capital efficiency (ROI). Profit divided by volume traded. A wallet making $100K on $500K of volume is sharper than one making $100K on $50M. It saturates faster than the profit term, so a thin edge on enormous volume cannot buy a high score.
- Hit rate vs price. For wallets where we load full position history (the top of each board), the share of settled positions that won. Winning 60% of trades entered around 45¢ is skill; winning 80% of trades entered at 90¢ is just paying for favorites, the ROI term keeps that honest.
The three terms are bounded, not free-floating: profit can move the score by at most ±24 points, capital efficiency by ±18, and hit rate by ±8 — and hit rate only counts once a wallet has enough settled positions to mean anything. No single input can carry a wallet on its own, which is the whole point of blending them.
The blend maps to a 1–99 scale centered at 50. Wallets with too little settled history are penalized rather than excluded, and boards labeled "Weekly / Monthly" re-rank on that window's results, so a stale legend drifts down as their edge fades.
What's deliberately excluded
- Long shots under 3¢ in the Tail Machine simulation — settled positions at those prices produce absurd hypothetical multiples nobody could actually replicate.
- Dust trades (tiny positions), they say nothing about conviction.
Kalshi: flow, never identities
Kalshi publishes no per-user data, every trade on its public tape is anonymous. So there are no Kalshi wallets on the leaderboard, ever. Instead we read the tape itself: block prints, one-sided aggressive taker runs, and volume anomalies. That signal is real, but it is flow, not an identified trader, and we always label it that way.
The Flow Score on Kalshi rows scores the print, not the (hidden) person behind it: where the trade's size sits in that market's recent tape, whether it was a negotiated block, and whether it rides a one-sided run of same-side aggressive takers. A Flow Score is identity-free by construction, two prints from the same (unknown) trader are scored independently, so it can never be a track record, and we never present it as one.
Known limits (v1)
- Market-maker-shaped wallets (constant two-sided flow) are not yet automatically filtered; MM detection ships with our historical database.
- Hit-rate enrichment covers the top of each board, not all 50 rows, rows showing "-" are ranked on PnL and ROI alone.
- Score v2 will add category-specific sub-scores, entry-timing quality (did they beat the close?), and multi-wallet cluster detection.
The only honest disclaimer
Past performance does not guarantee future results. A high Sharp Score means a wallet has been good; it is not a promise it will stay good. The Tail Machine is a hypothetical, retroactive simulation that ignores slippage, timing, and liquidity. Use WhaleTracks as intelligence, not as a reason to skip your own judgment. Never risk money you can't afford to lose — and if your trading on event-contract markets is causing you harm, call 1-800-GAMBLER.