The thesis
Mean-reversion is one of the oldest ideas in systematic trading, and this strategy states it plainly: buy when a name is oversold (RSI below 30), sell when it's overbought (RSI above 70). It runs on a 24-name large-cap universe spanning tech, financials, healthcare, staples and energy — AAPL, NVDA, JPM, JNJ, XOM and the like. The bet is simple: extreme short-term moves tend to snap back, and a disciplined RSI trigger harvests that snap.
Recent activity
The strategy is live and trading on a daily schedule, though most sessions do nothing — a feature, not a bug, for a signal that only fires at extremes. Across the last week of runs (19–23 October), four of six sessions executed zero trades. The exceptions tell the story well: on 16 October it sold 21 shares of WMT at $118.83, shares it had bought at $115.75 back on 31 May — a textbook buy-low, sell-into-strength round trip. On 23 October it sold 8 shares of HD at $318.76 against a $299.12 entry from 21 September, another clean mean-reversion win, and rotated into 3 shares of CAT at $688.79 as a fresh oversold entry.
The live account total sat at $9,966.13 after the 23 October run, drifting in a tight $9,858–$10,090 band over the week. In plain terms: the live book is roughly flat on its $10,000 start, with realized gains on HD and WMT offsetting mark-to-market chop.
Backtest and validation
The full-sample backtest looks attractive at first glance: 14.73% total return over 451 days (7.98% CAGR), a 70.59% win rate across 38 trades, and only $38 in fees. A seven-in-ten hit rate is genuinely rare and speaks to the edge in fading extremes.
But the headline number hides the problem. The Sharpe ratio is a modest 0.58, max drawdown reached 15.64%, and turnover ran to 879% — this strategy trades a lot relative to its returns. More importantly, the strategy failed walk-forward validation. Three of four folds were positive, but the most recent fold (16 December 2025 to 29 May 2026) lost 2.84% with a negative Sharpe of −0.33 and a 14.96% drawdown. The out-of-sample Sharpe is negative, and the deflated Sharpe ratio — which penalizes for the 6 trials run — collapses to 0.30. A PSR of 0.79 is decent, but the DSR tells the honest story.
Verdict
The high win rate and clean recent exits are real strengths, and nothing here is broken. But a strategy that works in three historical windows and breaks in the fourth — the most recent one — is exactly the overfitting risk our validation gate exists to catch. For now, mean-reversion earns a cautious hold: keep it live at small size, watch whether the live book can pull decisively above $10k, but don't scale capital into it until out-of-sample performance recovers.