Not supported · Timeframes · before 2026-09-26 (4h audit re-run 2026-09-30); timeframe gap fill 2026-10-11
Do the best 1h instruments also work on 15m and 4h?
What this means
The 15 stocks picked for strong 1h backtests won 60% of tranche trades on 1h, but that is the same data they were picked on. On 15m the same names won 39% and on 4h 44%, with only 10 of 17 and 13 of 19 profitable, which looks like ordinary universe results. The 1h and 15m top-50 lists also share fewer names than chance would predict. Treat this as a weak hint that a stock's good backtest does not carry to other timeframes; the lists were small, hand-picked and the settings were not saved. Re-done properly from the TradingView logs, with the picks made on earlier 1h data and scored later on 15m, 4h and daily, the best 1h stocks were no better than average on any other timeframe, and how a stock did on 1h says nothing about how it does on 15m (correlation -0.13).
The explanation is our best reading; the data shows the result but not the cause directly.
By timeframe
| Timeframe | Verdict | n | Result |
|---|---|---|---|
| 15-minute | Not supported | 1,685 | 1h top-50 (picked pre-2023-08): -0.005R vs 15m universe (-0.059, +0.049), q=0.92; instrument means correlate r=-0.13 gap fill G1, 2026-10-11 (trade logs) |
| 1-hour | Inconclusive | picks 59.7% tranche win in-sample | |
| 4-hour | Not supported | 113 | 1h top-50: +0.003R vs 4h universe after 2023-08 (-0.204, +0.230), q=0.98 gap fill G1, 2026-10-11 (trade logs) |
| Daily | Not supported | 59 | 1h top-50: -0.138R vs daily universe after 2023-08 (-0.473, +0.209), q=0.61 gap fill G1, 2026-10-11 (trade logs) |
- Hypothesis
- Instruments that backtest well on 1h keep working on neighbouring timeframes.
- Method
- TradingView headline audits: 15 hand-picked 1h names re-checked on 1h, then the same names (plus crypto) on 15m (17) and 4h (19). Settings not recorded.
- Result
- 1h picks: 804 tranche trades, 59.7% win, 15/15 profitable (in-sample). Same names on 15m: 1,036 trades, 39.4% win, 10/17 profitable; on 4h: 428 trades, 43.5% win, 13/19 profitable. Separately, the 1h and 15m top-50 lists share 3 names vs ~5 expected by chance. Gap fill (2026-10-11, from the trade logs): the top 50 instruments by 1h result before 2023-08, scored on each other timeframe from 2023-08 on, did no better than all instruments: 15m -0.005R (-0.059, +0.049), 4h +0.003 (-0.204, +0.230), daily -0.138 (-0.473, +0.209); per-instrument means on 1h and 15m correlate at r = -0.13.
- How much to trust it
- Low: small hand-picked lists, headline scrapes, settings not saved
- Did it change the live rule?
- No.
- Notes
- Consistent with the finding that instrument rankings are mostly noise (see ins-persistence).
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).