The thesis
RSI Snap-Back is a classic mean-reversion play narrowed to a hard target: large-cap tech. The idea is that Mag-7 names — AAPL, MSFT, NVDA, GOOGL, AMZN, META and TSLA — overshoot on short-term momentum and then revert. The rules are deliberately mechanical: buy the most oversold names when RSI drops below 35, exit when RSI pushes above 70, and never hold more than four at once. That 4-slot cap is the risk control doing the heavy lifting — it bounds how much concurrent drawdown the book can take on, and it forces the strategy to rank conviction rather than spray capital across the whole universe.
Recent activity
The live record so far reads as patient to the point of quiet. Across the scheduled runs from 20–27 October, five of six executed zero trades; the lone action was a single buy on 23 October — 6 shares of GOOGL at $321.67. As of the 27 October run the book held $8,064.65 in cash against a total of $9,997.91, meaning only one of four slots is filled and roughly 80% of capital is sitting idle.
That is the signature of a strategy waiting for its signal rather than manufacturing activity. With an entry trigger of RSI < 35, a flat-to-rising tape simply doesn't offer setups, and the strategy correctly declines to force them. The flip side: live total equity has drifted from a $10,000 start to just under it — essentially flat, marginally negative — so there is no live edge to point to yet.
Backtest and validation
The backtest is the strongest part of the story: +20.95% total return over 451 days, a 66.67% win rate across 37 trades, and a final equity of $12,095.14 for an 11.21% CAGR. Fees were trivial ($37 total, no FX cost).
But balance demands the caveats. The Sharpe of 0.61 is modest — this return came with meaningful volatility, confirmed by a 23.73% max drawdown. An investor would have had to stomach losing nearly a quarter of the book at the worst point. Turnover of 773% is high for a four-name strategy, implying frequent rotation that live fees and slippage could erode. Most importantly, the validation field is null — there is no reported out-of-sample or walk-forward test. A 20% backtest on a seven-name universe over a single 451-day window is encouraging but not yet evidence the rules generalise rather than fit the period.
Verdict
RSI Snap-Back is a clean, disciplined concept with a respectable in-sample result and, so far, well-behaved live conduct — it isn't over-trading, and it's honouring its cash discipline. The open questions are the ones that matter most: can the edge survive out-of-sample, and will the reversion signal fire often enough in this tape to put the idle capital to work? Until the validation gap closes and live P&L moves off the flat line, this one stays firmly in the 'promising, unproven' column.