The Thesis
Mean-reversion runs one of the oldest ideas in technical trading: buy weakness, sell strength. Concretely, it buys names whose 14-period RSI drops below 30 (oversold) and sells when RSI climbs above 70 (overbought). The universe is a clean slate of 24 US large-caps — the megacap tech complex (AAPL, MSFT, GOOGL, NVDA), financials (JPM, BAC, V, MA), healthcare, staples, and energy. These are liquid, well-behaved names, which is exactly where a mean-reversion premium tends to be modest but real.
Backtest Performance
On paper, the numbers look inviting. Over 451 trading days the strategy returned 14.73% (final equity $11,473 on a $10k base), a 7.98% CAGR, with a 70.59% win rate across 38 trades. That hit rate is the headline strength — better than seven trades in ten closed green.
But the risk-adjusted picture is more sober. The Sharpe ratio is just 0.58, and the maximum drawdown reached 15.64% — larger than the annualised return. Turnover of 879% is also aggressive for a book this size; the strategy churns its capital nearly nine times over, and while total fees were a contained $38, that level of trading leaves little margin for slippage in live conditions.
Validation: The Warning Light
This is where the story turns. The walk-forward validation did not pass. Across four folds, three were positive — folds one through three returned +2.06%, +11.10%, and +2.21% — but the most recent fold (Dec 2025 to May 2026) posted -2.84% with a -0.33 Sharpe and a 14.96% drawdown. Out-of-sample performance mirrored that weak tail: an OOS return of -2.84% and OOS Sharpe of -0.33.
The deflated statistics tell the same story. With 6 trials in the search, the Probabilistic Sharpe Ratio sits at 0.785, but the Deflated Sharpe Ratio falls to 0.304 — below the confidence threshold you'd want before trusting the edge is real rather than a product of selection. In plain terms: the backtest may be flattered by the favourable middle folds, and the strategy's edge decays badly in the newest data.
Recent Activity
Live behaviour reinforces the caution. The last executed trade was a buy of 21 WMT shares at $115.75 on 31 May. Since then, every scheduled run — 31 July through 7 August — reports 0 executed, 0 rejected, with cash pinned at $7,569.25 and total account value drifting between roughly $9,894 and $9,914. The strategy is effectively parked: no RSI extremes are triggering entries, and the paper book sits slightly below its starting line.
The Balanced Read
Mean-reversion's strengths are genuine — a high win rate and a coherent, well-understood thesis on a quality universe. But the failed validation gate is doing its job here. The negative most-recent fold, sub-threshold DSR, and a drawdown exceeding the return all argue against promoting this to serious capital in its current form. The idle streak also raises a practical question: an RSI-30/70 filter may simply be too restrictive to stay invested in a low-volatility regime.
The honest verdict: an interesting candidate, not a validated edge. It belongs in continued observation — not auto-deployment.