The thesis
RSI Snap-Back is a straightforward mean-reversion play on large-cap tech. The premise: the Mag-7 names — AAPL, MSFT, NVDA, GOOGL, AMZN, META, and TSLA — tend to overshoot on short-term momentum and then revert. The strategy buys the most oversold names when RSI dips below 35, exits when RSI pushes above 70, and enforces a hard four-slot book to cap how much concurrent drawdown exposure it can take on. It's a clean, legible rule set, and the discipline of a fixed book is a genuine strength: it structurally limits how badly a single crowded reversal can hurt.
Recent activity: quiet, deliberately
Here's the headline for the past week — nothing happened. The last six scheduled runs (July 10 through July 17) each report 0 executed, 0 rejected, with cash and total equity both parked at $10,000. No trades appear in the recent-trade log at all.
That isn't necessarily a fault. A rule that only fires on RSI < 35 will simply stay in cash when nothing in the universe is oversold enough — a market where the Mag-7 aren't hitting momentum extremes gives this strategy nothing to do. Sitting flat rather than forcing marginal entries is arguably the discipline working as intended. But it's worth being honest that a strategy in cash is a strategy earning nothing, and prolonged dormancy is a real opportunity cost.
Backtest and validation
Over 451 days of backtest, RSI Snap-Back returned 20.95% (final equity $12,095.14), an 11.21% CAGR, across 37 trades with a 66.67% win rate. Fees were negligible at $37 and there was no FX cost.
The strengths are visible: a two-thirds win rate and a double-digit annualized return from a simple, transparent rule. The risks are equally visible. The Sharpe of 0.61 is below 1, meaning the returns came with meaningful volatility relative to their size. The max drawdown of 23.73% is deep — nearly a quarter of capital at the trough — which is a lot to stomach for an 11% CAGR. And turnover of 773% signals a lot of rotation; in a live book with wider spreads or slippage than the backtest assumed, that churn could quietly erode the edge.
Most importantly: the validation field is null. We have an in-sample backtest and no out-of-sample or walk-forward confirmation. With only 37 trades over 451 days, the sample is thin, and a 66.67% win rate on that count carries real estimation error.
Verdict
RSI Snap-Back is an honest, well-constructed idea with a respectable backtest and sensible risk controls. But an unvalidated edge, a sub-1 Sharpe, a 24% drawdown, and a week of idleness mean it should be watched, not trusted. The next thing this strategy needs isn't a trade — it's out-of-sample validation.