mean-reversion-scanlisted

# mean-reversion-scan Find US equities that are **short-term oversold inside a confirmed long-term uptrend** — Larry Connors's canonical RSI(2) setup, augmented with persistence tracking and per-name outcome resolution so each subsequent run shows you the **running win rate** of past picks. The core bet: in a healthy uptrend, a stock whose 2-day RSI dives below 5 is likely panicking on a short-term overreaction (margin calls, ETF rebalancing, headline noise) rather than starting a real breakdown. The mean-reversion edge is **the bounce back to the 5-day average within 1-5 trading days**. Connors's published win rates on this exact setup are 70-75% on liquid US large-caps in RISK-ON regimes; the 25-30% losing trades are usually small-to-moderate but include occasional gap-down disasters when the trend was actually breaking. The natural complement to `momentum-scan` and `base-breakout-scan`: - `momentum-scan` finds **what's already running** (trailing return + low drawdown) - `base-breakout-scan` finds **what's about to run** (compressed pre-breakout bases) - `mean-reversion-scan` finds **what just got punched in the face but is structurally fine** (oversold inside an uptrend) The three skills filter for non-overlapping price patterns by design — you wouldn't want any of them to surface the same name on the same day. By default each run surfaces: 1. A **regime gate** (SPY > 200DMA + rising — Connors's hard filter; mean-reversion longs in a confirmed downtrend is the classi