Mean reversion · issue of Sep 17, 2026 · A few trades a year
RSI(2) Mean Reversion
Buy deep dips inside an uptrend.
Wins about 8 trades in 10 with a small drawdown, but sits in cash 90% of the time.
In cash — RSI(2) is 81, the rule needs below 10
Re-read at every close.
| Ticker | Name | Position |
|---|---|---|
| SPY | S&P 500 | Flat |
Levels
| Last close | 759.81 | |
| 200-day average below it, dips are not bought | 728.85 | -4.1% |
| Buys on a close below RSI(2) under 10 — but that is under the 200-day average, so no buy | 694.30 | -8.6% |
The closes at which the rule, as published, would act. Read from the last close; they move with the market.
Track record
| Last 5 years | Since 2000 | Buy & hold SPY, since 2000 | |
|---|---|---|---|
| Annual return | 4.5% | 3.7% | 8.5% |
| Sharpe ratio | 0.81 | 0.64 | 0.52 |
| Worst drawdown | -6.8% | -13.9% | -55.2% |
A backtest with the issue's own costs, extended to the last close. Past performance does not guarantee future results.
Stance history
In the market 11 of the last 63 sessions, returning +3.0% across them. Last change Nov 5, 2026: Long → Flat.
Research Lab
Written down before the results existed: Every 21 sessions after the issue returns between -1.5% and +2.3%: the range of 90% of the 6,508 such stretches in the backtest up to Sep 17, 2026.
| Window | Return | Result | |
|---|---|---|---|
| Check 1 | Sep 18, 2026 – Oct 16, 2026 | -1.1% | Inside the range |
Check 2 is running: 18 of 21 sessions in, judged on Nov 16, 2026.
Standing check: the deepest fall since the issue is -1.1%, against a worst of -13.9% in the backtest — within it.
Checked by the program when each window completes and sealed into the record. Misses stay public; the record counts both ways.
The issue's numbers, reproduced
Every figure the issue printed, recomputed on the same window from the platform's own price history.
| Figure | In the issue | Recomputed | |
|---|---|---|---|
| Annual return | 3.6% | 3.6% | matches |
| Sharpe | 0.62 | 0.62 | matches |
| Worst drawdown | -13.8% | -13.9% | matches |
| Winning trades | 81.8% | 81.4% | matches |
More from the newsletter, live
- The Closing-Range TradeBuy the country ETF that closed weakest
- Catching Crypto TrendsNine breakout models, one position
- The Rebalancing TradeFront-run month-end pension flows
- Momentum 2.0Two fixes for the oldest strategy in quant
- The Factor SwitchTrend-following the factor zoo
- The 'Do Less' StrategyHold less stock when markets get wild
- The Turn-of-the-Month EffectSix trading days around each month end
- The Overnight EffectOwn the close-to-open, skip the day
- The Turtle Breakout on Bitcoin20-day breakout with a trailing ATR stop
Get the next stance change
in your inbox.
Follow any issue and you get one email when its strategy changes position. Free, no card.