The thesis
RSI Snap-Back bets that large-cap tech names overshoot in the short term and then revert. The rule set is deliberately simple: buy the most oversold names when RSI falls below 35, exit when RSI pushes above 70, and hold no more than four names at once. That hard 4-slot book is the strategy's risk governor — it caps how much concurrent drawdown exposure the book can accumulate and forces rotation into the genuinely oversold rather than sprawling across the whole universe (AAPL, MSFT, NVDA, GOOGL, AMZN, META, TSLA).
It's a coherent idea. Mean reversion in liquid mega-caps is one of the better-documented short-horizon effects, and the tight book keeps the concept honest.
Recent activity
Here's the notable part: over its last six scheduled runs (19–26 August), the strategy executed nothing. Each run reports 0 executed, 0 rejected, cash and total equity both pinned at $10,000. No entry conditions triggered — evidently none of the Mag-7 dipped under the RSI-35 threshold in that window.
That is not a bug; it is the strategy behaving as specified. A mean-reversion system that only fires on genuine momentum extremes should be idle when the tape is calm or grinding higher. The flip side is opportunity cost: a strategy sitting flat in cash earns nothing while the underlying names may be drifting up out of reach. Patience is the feature and the tax at the same time.
Backtest and validation
The backtest, over 451 days, is respectable but not spectacular:
- Total return: 20.95% (final equity $12,095)
- CAGR: 11.21%
- Sharpe: 0.61
- Max drawdown: 23.73%
- Win rate: 66.67% across 37 trades
- Turnover: 773%
Two numbers deserve scrutiny. First, a Sharpe of 0.61 against a 23.73% max drawdown is a modest risk-adjusted profile — you're accepting nearly a quarter of the book in peak-to-trough pain for low-double-digit annualised returns. A two-thirds win rate is encouraging, but the drawdown says the losers, when they come, are not small.
Second, 773% turnover means the book churns roughly eight times over. Total modelled fees were only $37, so cost drag looks benign here — but that assumes the fee and slippage model is realistic. In live trading, high turnover is where optimistic backtests quietly bleed.
Most importantly: the validation block is null. There is no out-of-sample or walk-forward confirmation attached to these figures. A 20.95% in-sample return with no holdout is exactly the kind of result that warrants caution before treating it as predictive. On this platform, robust validation is the first line of defence against overfitting, and it hasn't been run.
Verdict
RSI Snap-Back is a clean, disciplined concept executing exactly as designed — including its current, entirely reasonable stretch of inactivity. The strengths are a sensible thesis, a strong win rate, and a hard risk cap. The open questions are the deep drawdown, the churn, and the absence of any validation. It earns cautious interest, not conviction, until an out-of-sample test backs the story.