The thesis
Mean-reversion is one of the oldest ideas in systematic trading: extremes correct. This strategy buys names when their RSI drops below 30 (oversold) and sells when RSI climbs above 70 (overbought), ranging across a 24-stock large-cap US universe spanning tech (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and energy. It is currently running live. The appeal is intuitive — fade the panic, trim the euphoria — and the historical hit rate backs it up. The question, as always, is whether that edge survives out of sample.
Recent activity
November has been quiet. Of the six most recent scheduled runs, five executed zero trades — the strategy simply had no oversold signals to act on. The exception was 6 November, which fired a single buy. The active positioning happened in late October: BAC (48 shares @ $51.15), GOOGL (8 @ $280.28), and CAT (4 @ $622.54) on 29–30 October, followed by NVDA (13 @ $188.24) on 6 November. Those purchases drew cash down from $2,742 to just $271.
Less encouraging is the equity line. The portfolio total slipped from roughly $9,847 to about $9,511 over the window, leaving it below the $10,000 base the backtest grows from. For a live strategy, that gap between promise and reality is the thing to watch.
Backtest performance
On paper the numbers look healthy. Over 451 days the strategy returned 14.73% (final equity $11,473), a 7.98% CAGR, with a 70.59% win rate across 38 trades — exactly the high-hit-rate profile mean-reversion is known for. Fees were negligible ($38 total, no FX cost). The blemishes: a Sharpe of 0.58 is mediocre, max drawdown reached 15.64%, and turnover of 879% is heavy — this strategy trades its book over roughly nine times a year.
Validation — where it breaks down
This is the part that should temper any enthusiasm. The walk-forward analysis did not pass. Across four folds, three were positive, but the fourth and most recent (Dec 2025 → May 2026) lost 2.84% with a negative Sharpe of −0.33 — the aggregate out-of-sample result. The deflated Sharpe ratio (DSR) sits at just 0.304; after accounting for six trials, that says the edge is not statistically distinguishable from luck. PSR of 0.785 is better but not reassuring given the DSR.
Verdict
Strengths: a genuinely high win rate, tightly controlled costs, and a diversified, liquid universe. Risks: negative out-of-sample returns, a deteriorating most-recent fold, a failed validation gate, punishing turnover, and a live book already underwater. The 14.73% headline is real, but the overfitting signals are loud. I would treat this strategy as a research candidate that needs a sturdier signal before it earns more capital — not one to lean on live today.