Predictive Validity of ICT's AMD Model on Nasdaq-100 Futures

01 ABSTRACT

The author operationalizes the AMD model into a mechanical strategy and analyzes 484 manipulation signal days using chi-square testing, Benjamini-Hochberg correction, and regime-conditional analysis on 15-minute OHLCV data from January 2022 to July 2026. Key findings: the manipulation signal fires on 44% of trading days; 57% of valid signals see full target delivery before the 9:40 AM New York entry window; and the signal performs well only when aligned with the daily trend (62-71% correct) versus 25-30% when opposed. The author frames these findings as a testable hypothesis rather than a final verdict on the model's validity.

02 KEY FINDINGS

1. The AMD manipulation signal fires on 44% of trading days, but 57% of valid signals complete target delivery before the New York entry window, indicating a structural mismatch between the model's timing assumptions and actual price paths.
2. Directional information is highly regime-dependent: when aligned with the daily trend, accuracy is 62-71% (Cohen's h = 0.23-0.43); when opposed, only 25-30% (Cohen's h = 0.42-0.53).
3. All four trending-regime cells survive Benjamini-Hochberg correction, and key findings are robust to variations in the manipulation breach threshold (5-20 NQ points), Asia range filter percentile, and stop buffer.