Structural Limits of OHLCV-Based Intraday Signals in MNQ Futures: A Systematic Falsification Study

Abstract

This paper asks a straightforward question: do common intraday momentum signals built from price and volume data produce a tradable edge in Micro E-Mini Nasdaq 100 (MNQ) futures once realistic execution costs are included? I tested fourteen signal families using 947 trading days of five-minute data from 2021-2025. Every signal was evaluated using the same criteria: out-of-sample walk-forward validation, a minimum T-statistic of 2.0, at least 30 trades, positive net returns after a fixed two-point round-trip friction cost, and consistent performance across years. None of the tested strategies satisfied all of these requirements. Across all signal families, the maximum gross return before transaction costs ranged from roughly 0.07 to 1.50 points per trade, well below the assumed two-point friction cost. One signal family-gap continuation short-produces a T-statistic of 3.23 and a mean net return of 14.52 points but on only 22 trades across three years, falling below the minimum sample threshold and therefore failing deployment criteria. Two separately validated signals-the RTH Confluence Signal (T = 5.83, mean net +15.77 pts, N = 538) and London Session Signal B (T = 5.15, mean net +5.77 pts, N = 289)-are presented as positive controls confirming the methodology is capable of detecting genuine edge when it exists. The primary contribution is methodological rather than predictive. By applying a consistent evaluation framework across a broad set of commonly used intraday strategies, this study documents where these approaches fail under realistic trading conditions and highlights the importance of reporting negative results alongside successful ones.