Not supported · Exits · 2026-09-22; timeframe gap fill 2026-10-11
Is a tight stop just under D (D x 0.998) better than min(D, X) x 0.99?
What this means
This Python sim showed a tight stop just under D (+0.23R) beating the wider min(D, X) stop (−0.14R). But the code books any trade still open after 100 bars as a full −1R loss, which hurts the wider stop more, and it checks targets before stops within each bar. The same wider stop earns +0.19R per setup on TradingView's 1h fills. These flaws are the likely reason for the result, and no TradingView comparison of stop types was ever saved, so the question remains open. Re-tested on TradingView's setups, the tight stop does not raise the win rate or the average: on 1h it loses about 0.05R per setup, because stops just under D are hit by ordinary noise before the bounce develops, and on 4h and daily it makes no difference. An ATR-based stop looks slightly better on 4h and daily and slightly worse on 1h, so it is a mixed lead rather than a finding. These are simulations, not TradingView fills.
Explanation checked against the data.
By timeframe
| Timeframe | Verdict | n | Result |
|---|---|---|---|
| 15-minute | Not tested: not tested yet: no 15-minute price data; needs 15m prices or a TradingView run; the stop variants in exit_config_sweep were never saved (see tradingview_runs.md) | ||
| 1-hour | Not supported | 3,669 | simulated D x 0.998: -0.045R vs min(D,X) x 0.99 (q=0.02), win 44.2% vs 46.1% gap fill G12, 2026-10-11 (price simulation; 1h/4h prices cover Nov 2023-Sep 2026 only) |
| 4-hour | Not supported | 1,057 | D x 0.998: +0.008R (q=0.84) gap fill G12, 2026-10-11 (price simulation; 1h/4h prices cover Nov 2023-Sep 2026 only) |
| Daily | Not supported | 2,031 | D x 0.998: -0.007R (q=0.80) gap fill G12, 2026-10-11 (price data) |
- Hypothesis
- A tighter stop raises win rate and total R.
- Method
- Own Python sim, yfinance 1h 730 days, 119 symbols, 682 setups; entry at close of D+12; dual tranche; max hold 100 bars.
- Result
- Baseline min(D,X)x0.99: -0.14R/setup (PF 0.79). D x 0.998: +0.23R (PF 1.44); '+157R vs -97R'. Gap fill (2026-10-11): price simulation on TradingView setups, stop at D x 0.998 (D read from the bars at the tagged D) with targets re-based on the new risk, paired with the simulated baseline: 1h -0.045R (-0.078, -0.012; q = 0.02), 4h +0.008 (q = 0.84), daily -0.007 (q = 0.80); win rate 44.2% vs 46.1% on 1h. Other Pine stop modes (exploratory): D x 0.99 -0.013 / +0.010 / +0.003; D - 1 x ATR(14) -0.036 (1h, q = 0.01), +0.038 (4h, q = 0.12), +0.036 (daily, q = 0.04). Median stop: 1h 2.4% tight vs 3.4% logged; daily 6.6% vs 8.2%.
- How much to trust it
- Medium: price simulation on TradingView setups (baseline matches TradingView outcomes on 95-96%), paired comparison; 1h/4h only 2023-26; not yet confirmed with TradingView fills
- Did it change the live rule?
- made the default 2026-09-22 (eb04447), changed to D x 0.99 the same day, reverted to min(D,X) x 0.99 on 2026-09-30; no TV comparison of stop modes was ever saved
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).