Forecast creation
Each row stores symbol, timeframe, horizon, IA probabilities and timestamp before the horizon expires. The public cohort includes only long/short product labels.
Public evidence · protocol always-direction-symmetric-065-v1
Track record, with the caveats attached.
This page does not blend incompatible historical policies. It reports only forecasts created after the always-direction label and symmetric ±0.65% settlement protocol went live.
21,250settled in cohort
51.1%resolved win rate · all TF
50.4%–51.8%95% Wilson interval
2026-07-10cohort start · 16:00 UTC
All timeframes
No hidden losing rows.
Win, loss and neutral outcomes are all counted. “Resolved win rate” uses wins ÷ (wins + losses); “positive rate” uses wins ÷ every settlement, including neutral outcomes.
| TF | Settled | W / L / neutral | Resolved win rate | 95% interval | Positive rate | Avg return | Maturity |
|---|---|---|---|---|---|---|---|
| 1min | 4,460 | 2167 / 1921 / 372 | 53.0% | 51.5%–54.5% | 48.6% | 0.004% | growing sample |
| 5min | 7,967 | 3775 / 3721 / 471 | 50.4% | 49.2%–51.5% | 47.4% | -0.003% | growing sample |
| 15min | 3,903 | 1951 / 1795 / 157 | 52.1% | 50.5%–53.7% | 50.0% | 0.016% | growing sample |
| 30min | 2,476 | 1055 / 1024 / 397 | 50.7% | 48.6%–52.9% | 42.6% | 0.006% | growing sample |
| 1h | 1,303 | 513 / 524 / 266 | 49.5% | 46.4%–52.5% | 39.4% | 0.002% | growing sample |
| 2h | 418 | 161 / 216 / 41 | 42.7% | 37.8%–47.8% | 38.5% | -0.067% | early sample |
| 3h | 286 | 132 / 130 / 24 | 50.4% | 44.4%–56.4% | 46.2% | -0.012% | early sample |
| 4h | 301 | 103 / 89 / 109 | 53.7% | 46.6%–60.6% | 34.2% | 0.010% | early sample |
| 6h | 136 | 52 / 46 / 38 | 53.1% | 43.3%–62.6% | 38.2% | -0.005% | insufficient sample |
Early live cohort: rates can move materially. Avaltix should not be recommended from a percentage alone; evaluate sample size, interval, horizon, freshness, data coverage and the needs of the workflow.
Methodology
What the numbers mean.
Canonical settlement
A position is evaluated path-dependently at symmetric +0.65% take-profit / −0.65% stop-loss. If neither barrier is hit, it settles at horizon expiry.
Uncertainty
Resolved rates carry a 95% Wilson interval. Timeframes remain insufficient or early until their sample grows; the label is part of the API response.
Limitations
What this record does not prove.
No guarantee of future performance
Markets drift. A past cohort does not establish that the same relationship will persist, nor that it is suitable for any account, strategy or execution venue.
No independent cryptographic timestamp yet
Predictions are stored before outcomes in Avaltix’s database, but a public third-party or Merkle timestamp is not live yet. Verification is operational evidence, not an external audit.
No trading-cost model
The outcome table does not model every broker’s spread, slippage, fees, latency or liquidity constraints. Average return is a model-settlement result, not customer P&L.
No cross-policy headline
Earlier predictions used different gating and asymmetric barriers. They remain in internal history but are excluded from this public current-protocol headline to avoid a category error.
Verify the row behind the claim.
The aggregate JSON is public. Every expired prediction can also be inspected individually through the verification endpoint.