The thesis
Mean-reversion is the most classical of technical plays: buy what's oversold, sell what's overbought. Here that rule is made concrete with RSI — accumulate names trading below RSI 30, trim them above RSI 70. The universe is 24 large-cap US blue chips (AAPL, MSFT, NVDA, JPM, V, JNJ, WMT, XOM and the like), which is a sensible playground for the idea: liquid, well-covered stocks whose short-term dips tend to attract buyers rather than signal terminal decline.
Recent activity
The live book has been mostly dormant. Scheduled runs from 19 to 22 October all report 0 executed, 0 rejected — RSI simply isn't hitting extremes across the universe, so the strategy sits on its hands. That is the design working as intended, not a malfunction: no signal, no trade.
The one recent move of note was a sell of 21 WMT at $118.83 on 16 October, closing a position opened at $115.75 back on 31 May — a modest gross gain of roughly $3.08 per share. That sale lifted cash from $329.71 to $2,819.82, where it has since stayed. Earlier buys (BAC, HD, MCD in September) round out the current holdings. Total portfolio value has drifted between about $9,858 and $10,090 over the past week — broadly flat, with the strategy carrying a meaningful cash cushion of roughly 28%.
Backtest and validation
On paper the numbers look inviting: +14.73% total return over 451 days (≈7.98% CAGR), a 70.59% win rate across 38 trades, and a max drawdown of 15.64%. Fees were negligible at $38, though turnover of 879% is high and worth watching if frictions rise.
The honest story is in the walk-forward validation, which did not pass. Three of four folds were positive — fold 2 was excellent (+11.1%, Sharpe 1.32) — but the fourth and most recent fold (Dec 2025–May 2026) turned negative: −2.84% return, Sharpe −0.33, with a 14.96% drawdown. The overall Sharpe of 0.58 is unremarkable, and the deflated Sharpe ratio of 0.304 — after accounting for 6 trials — is well short of what we'd want to trust the edge as real rather than mined. The probabilistic Sharpe of 0.785 is better but not decisive.
The balance
Strengths: a transparent, well-understood thesis; a high win rate; and discipline to stay flat when there's no setup. Risks: the out-of-sample weakness is exactly where it matters — the most recent regime — and the low deflated Sharpe warns of overfitting. A 70% win rate paired with a sub-0.6 Sharpe also hints that the losers, when they come, are large.
Verdict: a respectable classical strategy that reads well historically but hasn't yet proven it generalises. It stays live and on a short leash — we want to see it clear the out-of-sample bar before leaning on it.