The Thesis
RSI Snap-Back is a mean-reversion strategy built on a simple, well-worn observation: large-cap tech names tend to overshoot in the short term and then snap back. It buys the most oversold names by RSI (entering below 35) and exits once they run hot (above 70), rotating capital to keep a tight four-name book at all times. The universe is the Mag-7 — AAPL, MSFT, NVDA, GOOGL, AMZN, META, and TSLA — a deliberately narrow, high-liquidity pool where mean-reversion signals are cleaner and slippage is low.
The four-slot cap is the disciplinary heart of the design. By hard-limiting concurrent positions, it forces the strategy to prioritise its strongest signals and mechanically bounds how much drawdown it can accumulate at once. That's a sensible guardrail for a style that, by definition, buys names precisely when they look worst.
Backtest Performance
Over 451 days, the backtest returned 20.95% (final equity $12,095 on a $10k base), an 11.21% CAGR. The win rate is a healthy 66.67% across 37 trades, which fits the mean-reversion profile — many small, reliable wins. Trading costs were negligible: $37 in fees, no FX drag.
The numbers to watch are risk-adjusted. A Sharpe of 0.61 is modest — the returns came with real volatility rather than a smooth ride. More pointed is the 23.73% max drawdown, which is large relative to the total return. In plain terms, an investor would have had to stomach a peak-to-trough loss nearly as big as the entire multi-month gain. Turnover of 773% also signals a lot of churn; the four-name rotation keeps the book moving constantly, and while fees stayed low here, that pace leaves the strategy more exposed to execution quality in a live, less forgiving environment.
Recent Live Activity
Here's the honest part: live, the strategy has done nothing this past week. Every scheduled run from 21 through 28 August executed zero trades and rejected zero — cash and total equity have sat flat at $10,000. No RSI reading in the Mag-7 crossed the entry threshold, so the model correctly stayed on its hands.
That's not a failure; a mean-reversion strategy that refuses to force trades when no signal exists is behaving exactly as designed. But it's a reminder that this style is episodic. It waits for extremes, and extremes don't arrive on a schedule. A flat week tells us discipline is intact; it tells us nothing yet about live edge.
The Gap: No Validation
The most important caveat is what's missing. The validation field is null — there's no out-of-sample or walk-forward result on record. Everything above rests on a single backtest, and a 66.67% win rate over just 37 trades is a small sample that could flatter a curve-fit. Until validation lands, treat the headline return as a promising hypothesis, not proven edge.
Verdict
RSI Snap-Back is a clean, well-disciplined design with an intuitive thesis and encouraging backtest wins. The risks are equally clear: a heavy drawdown, a middling Sharpe, high turnover, and — critically — no validation yet. It has earned a live slot to prove itself. It hasn't yet had the chance to.