Not supported · Outside indicators & papers · 2026-09-30
Is the 'Harmonic Auto-Validator [GBB]' indicator's 89-98% win rate real?
What this means
The indicator's 89-98% win rate only holds if you could buy exactly at Point D, but D is only labelled after price has already bounced. Entering when its alert actually fires gave a 44% win rate on 1h and results close to random entries; daily looked better (+0.21R over random) but was not significant. The study code was deleted, so these numbers survive only in a session log and cannot be checked. They should not be relied on, but nothing in them supports the headline win rate.
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 tested: outside indicator replication (rejected); not part of Harmonic Master's timeframe coverage | ||
| 1-hour | Not tested: outside indicator replication (rejected); not part of Harmonic Master's timeframe coverage | ||
| 4-hour | Not tested: outside indicator replication (rejected); not part of Harmonic Master's timeframe coverage | ||
| Daily | Not tested: outside indicator replication (rejected); not part of Harmonic Master's timeframe coverage | ||
- Hypothesis
- Strict Carney harmonics (HAV) give a tradeable edge, possibly as the basis for a new Harmonic Master.
- Method
- Own Python rebuild of HAV (checked swing-for-swing against the real indicator on one ETH chart); US large caps 1h/4h/1d; entries at D (chart view), at the alert, limit at D, and at D-confirmation (Harmonic Master timing); controls: random entries and any confirmed swing low.
- Result
- Win rate if you could buy at D: 89% (1h), 94% (4h), 98% (1d) - hindsight, since D is only labelled after price bounced. At the alert: 1h 44%, +0.10R (355 trades), -0.03R vs random; 4h +0.34R (32 trades); 1d +0.57R (131), +0.21 vs random, not significant. Limit at D: 1h -0.02R vs random, 4h -0.25R. At Harmonic Master timing: 1h 41.6%, +0.04R (vs HM's own +0.20R); 4h -0.04R (43); 1d +0.43R, ~+0.12R of it from the pattern.
- How much to trust it
- Low: code and FINDINGS.md deleted 2026-10-01; numbers survive only in the session transcript and a memory note
- Did it change the live rule?
- dropped
- Notes
- No repo source. If published, label as 'from session log; study code deleted'.
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).