The Thesis
RSI Snap-Back is a mean-reversion play on the Mag-7 complex — AAPL, MSFT, NVDA, GOOGL, AMZN, META, and TSLA. The premise is that large-cap tech names overshoot on short-term momentum and then snap back. Mechanically, it buys the most oversold names when RSI drops below 35 and exits when RSI climbs above 70, rotating to hold a tight four-name book. That hard 4-slot cap is the strategy's discipline mechanism: it bounds how many concurrent positions can draw down at once, trading breadth for containment.
It's a coherent, well-worn idea. Mean reversion on liquid mega-caps is a legitimate edge in choppy regimes — and a liability in trending ones, where "oversold" names keep falling.
Recent Activity
Here the picture is quiet to the point of inert. Across every scheduled run on record — spanning 15 through 21 October — the strategy executed zero trades and rejected zero. Cash and total equity both sit flat at $10,000. No RSI reading in the universe has crossed the entry threshold, so the book has stayed empty.
That isn't a malfunction; it's the design working as intended. A strategy that only acts on momentum extremes should do nothing most of the time, and sitting in cash means it isn't bleeding fees or absorbing drawdown while it waits. The flip side: a signal this selective can idle through long stretches, and an idle book earns nothing. There are no realized trades in the live record yet to confirm the backtested behaviour carries over.
Backtest and Validation
Over 451 days the backtest returns 20.95% (final equity $12,095), an 11.21% CAGR, with a 66.67% win rate across 37 trades. Fees were a trivial $37 and FX cost nothing. On the surface, a two-thirds hit rate and double-digit annualised return read well.
The risk side is more sobering. The Sharpe of 0.61 is modest — this return came with meaningful volatility, not smooth compounding. More pointedly, the max drawdown of 23.73% nearly matches the entire period's return: an investor would have had to stomach a loss almost as large as the eventual gain. Turnover of 773% is high for a 37-trade sample, implying concentrated, hard-swinging positions within that four-slot book.
The most important caveat is what's missing: validation is null. There is no out-of-sample or walk-forward check here, only the in-sample backtest. A strong in-sample curve on a parameter-tuned RSI rule is exactly the kind of result that overfits. Until the live record accumulates real trades, these numbers should be read as a hypothesis, not a track record.
Verdict
RSI Snap-Back is a disciplined, low-cost design with a plausible edge and a genuinely good win rate — but one whose headline return is shadowed by a near-equal drawdown and, crucially, no validation beyond a single backtest. Its current all-cash stance is defensible patience. The honest position: promising on paper, unproven in the wild, and in need of out-of-sample evidence before the backtest earns any trust.