Not supported · Filters · 2026-09-27
Does an 'RSI <= 40 at Point D' filter help on 1h?
What this means
Requiring RSI ≤ 40 at Point D changed nothing on 1h: +0.191R per setup with the filter and +0.193R without, a difference of −0.002R with a tight range around zero. Inside the unfiltered log, setups with RSI ≤ 40 earned +0.200R and the other 36% earned +0.180R, a gap well within noise. So the filter mainly removes about a third of the setups without improving the ones that remain.
Explanation checked against the data.
By timeframe
| Timeframe | Verdict | n | Result |
|---|---|---|---|
| 15-minute | Not supported | 16,496 | RSI at D <= 40: within +/-0.01R |
| 1-hour | Not supported | 8,470 | -0.002R (-0.019, +0.014) vs unfiltered |
| 4-hour | Not supported | 3,990 | CI spans 0 (filter-rsi-ma-recheck) |
| Daily | Not supported | 2,098 | CI spans 0 (filter-rsi-ma-recheck) |
- Hypothesis
- Only taking setups where RSI(14) at D is 40 or lower (oversold) improves results.
- Method
- Two full TradingView 1h sweeps of the same 503 symbols, 2016-01 to 2026-09: RSI <= 40 at D enabled vs all filters off. Setup R recomputed; cluster bootstrap.
- Result
- RSI-filtered: 8,470 setups, 46.6% win, +0.191R (+0.163, +0.223). Unfiltered: 13,208 setups, 46.6% win, +0.193R (+0.170, +0.216). Difference -0.002, 95% CI (-0.019, +0.014). Same split inside the unfiltered log: D<=40 +0.200 vs D>40 +0.180, diff CI (-0.025, +0.068).
- How much to trust it
- High: two real TV sweeps of the same universe
- Did it change the live rule?
- the 1h basket was first built from this RSI<=40 sweep; rebuilt on 2026-09-30 from the filter-off sweep so alerts match the live (filter-off) settings
- Notes
- File nasdaq_sp_1h_sweep_results.json is labelled timeframe '15m' internally but is the 1h RSI<=40 sweep (per create_1h_top50_alerts.py).
Scripts and data for this study will be linked here when the research code is published on GitHub.
Education and research only. Not financial advice. Backtests use today's index members (survivorship bias).