The thesis
RSI Snap-Back bets on a well-worn idea: large-cap tech names tend to overshoot in the short term and snap back once momentum reaches an extreme. The rules are refreshingly simple — buy the most oversold names when their 14-period RSI drops below 35, exit when RSI climbs above 70, and rotate holdings to keep a tight four-name book. The universe is the Magnificent Seven: AAPL, MSFT, NVDA, GOOGL, AMZN, META, and TSLA.
The four-slot cap is the strategy's defining risk control. By refusing to hold more than four positions at once, it forces the book to concentrate on the strongest reversion candidates and limits how much concurrent drawdown the portfolio can absorb. That discipline is a feature, not an afterthought.
Backtest and validation
Over 451 days the backtest turned $10,000 into $12,095.14 — a 20.95% total return, or roughly 11.21% CAGR. The win rate is a healthy 66.67% across 37 trades, and trading costs were negligible ($37 in fees, no FX drag thanks to the all-US universe).
The less flattering numbers sit on the risk side. The Sharpe ratio is 0.61 — positive, but modest for a concentrated equity strategy — and the maximum drawdown reached 23.73%. In other words, the path to that 21% was bumpy: an investor would have had to stomach losing nearly a quarter of peak equity along the way. Turnover of 773% also confirms this is an active rotator, not a buy-and-hold proxy.
The most important caveat is what is missing: the validation field is null. There is no out-of-sample or walk-forward result on record. A clean in-sample backtest on a seven-name universe is exactly where mean-reversion rules are most prone to curve-fitting, so the 20.95% should be read as encouraging rather than proven.
Review and recent activity
The code itself has cleared our review gate. A reviewer approved it at risk level 0.35, citing "sound RSI mean-reversion logic with correct position-sizing and cash tracking" and no look-ahead or unsafe code — flagging only three minor defensive-coding gaps not exploitable under normal inputs. Notably, an earlier backtest failed the gate outright (0% return, 0% drawdown) before the working version was deployed, a reminder that the gate does bite.
Live activity is quiet. The scheduled run on 2026-10-29 executed zero trades and rejected zero, leaving cash and total equity both at exactly $10,000 — consistent with a fresh book awaiting its first RSI < 35 signal. With no names currently oversold enough to trigger entry, the strategy is doing what it should: waiting.
Verdict
RSI Snap-Back is a clean, well-governed implementation of a sensible idea, with a strong win rate and tight risk framing. But the 0.61 Sharpe against a 23.73% drawdown is a mediocre risk-adjusted profile, and the absence of out-of-sample validation is the real open question. Until it logs live trades and we see walk-forward numbers, treat the backtest as a hypothesis under test — not a track record.