AB VectorMomentum
- 지표
- 버전: 1.0
- 활성화: 5
How it works
No RSI, no stochastic. VectorMomentum measures momentum three independent ways and demands agreement:
– Regression-slope t-statistic — a linear regression on log price over N bars; the t-stat of the slope (slope divided by its standard error) gives statistically-significant momentum that self-normalizes across symbols and volatility levels.
– Volume-pressure-weighted rate of change — ROC weighted by relative tick volume, so momentum on dead volume scores lower than the same move on climactic participation.
– Multi-horizon agreement — the t-stat is computed on three horizons (e.g. 10/30/90 bars). Full agreement equals a high-conviction vector; disagreement signals decaying or rotating momentum.
Output is a single Momentum Vector from −100 to +100 with an explicit conviction band. The divergence module only flags divergences when the vector is statistically significant (|t| greater than 2), removing roughly 80% of the false divergences classic oscillators print.
Marketplace description
| Classic oscillators measure price displacement. VectorMomentum measures statistical significance. Its core is the t-statistic of a rolling regression slope — the same construct quantitative funds use for time-series momentum — combined with volume-pressure weighting and three-horizon agreement scoring. The result is one clean −100 to +100 vector that is comparable across every pair, index and metal, with a conviction band that tells you when all horizons align. The divergence engine only flags divergences that are statistically significant, eliminating the noise that makes RSI divergence untradeable. Non-repainting, EA-ready buffers, full alert suite. This is momentum, rebuilt from first principles. |
