@robinhoodalphas

low sample (3 verified calls) — treat the score and win rate as indicative, not proven. the 95% confidence range shows how wide the uncertainty still is.

source: twitterapi.io · dexscreener · blockscout · win = +50% still held at the 24h mark (or latest check) — a single peak wick is not a win · calls at −50% or worse count against the score · repeat promotion of the same contract counts once (promos column) · handles can change, the underlying twitter id is what's tracked