logo

@misango

logo

Frankline Oyolo, Misango

kagglelinkedinorcidgithub
Published on

Intraday Momentum Breakout Strategy: A Volatility-Targeted Approach to E-mini Futures Trading

Intraday-TradingMomentum-StrategyFutures-TradingRisk-ManagementVolatility-Targeting

Abstract

This paper presents a comprehensive analysis of an intraday momentum breakout strategy for E-mini S&P 500 (ES) and E-mini NASDAQ-100 (NQ) futures. The strategy employs volatility-based "noise area" boundaries — percentile-based rather than standard deviation — to identify momentum opportunities while maintaining strict intraday-only execution. We implement a volatility-targeted position sizing framework with a maximum leverage constraint of 8x and target daily portfolio volatility of 3%. Through rigorous backtesting from 2011–2026 with 1-tick slippage per side, the portfolio achieves 247% total return, 8.5% CAGR, and a Sharpe ratio of 1.18.

Introduction

Intraday momentum strategies exploit short-term price movements that persist over minutes to hours. Unlike daily momentum, intraday approaches eliminate overnight gap risk while leveraging the highly liquid E-mini futures markets ($100B+ daily volume). The strategy targets the regular session window, where momentum effects are empirically strongest, and mandates a flat position by end of day under all circumstances.

Motivation

Three empirical observations motivate the design:

  1. Volatility clustering: Intraday price movements exhibit GARCH effects, creating predictable momentum patterns
  2. No overnight risk: Intraday-only execution eliminates exposure to overnight gaps — which account for ~35% of buy-and-hold drawdowns in the same period
  3. Noise area innovation: Unlike Bollinger Bands (standard deviation), percentile-based boundaries adapt more robustly to regime changes in intraday range distributions

Key Equations

Noise area boundaries (90-day lookback, percentile-based):

where is the daily high-low range.

Signal generation (with 2-bar confirmation and volume filter):

Volatility-targeted position size:

Leverage cap (hard limit at 8×):

Daily loss limit:

Flash crash circuit breaker:

Algorithm Blueprint

Algorithm — Intraday Momentum Breakout — ES & NQ
Input: 5-minute OHLCV bars, initial capital, intraday range history
Output: Signals, position sizes, end-of-day flat book
AT 9:30 AM ET:
compute 90-day intraday ranges
set upper boundary from 75th percentile range
set lower boundary from 25th percentile range
estimate instrument volatility with EWMA
DURING the session:
FOR each 5-minute bar:
IF close > upper boundary AND volume > median volume AND no open position:
confirm over 2 bars and generate LONG signal
IF close < lower boundary AND volume > median volume AND no open position:
confirm over 2 bars and generate SHORT signal
size position from target volatility, price, and leverage cap
RISK MANAGEMENT:
IF daily PnL breaches loss limit: flatten positions and halt entries
IF 5-minute move exceeds 3%: activate circuit breaker and pause trading
IF drawdown exceeds 20%: require manual intervention
EXIT PRIORITY:
exit when price re-enters noise area
otherwise exit after maximum hold
always flatten by 15:45 ET
ALLOCATE capital across ES and NQ components and rebalance if drift > 5%

Results

Annual Returns by Year — Portfolio (2011–2025)

Strategy Component Comparison

| Metric | Portfolio | NQ Mom. | ES Mom. | NQ Long | |---|---|---|---|---| | Total Return | 247.3% | 312.5% | 98.7% | 189.4% | | CAGR | 8.5% | 10.1% | 4.7% | 7.2% | | Sharpe Ratio | 1.18 | 1.05 | 0.67 | 0.94 | | Sortino Ratio | 1.67 | 1.52 | 0.91 | 1.38 | | Max Drawdown | -16.4% | -21.1% | -15.2% | -16.3% | | Win Rate | 54.3% | 52.8% | 50.1% | 61.2% | | Profit Factor | 1.64 | 1.58 | 1.32 | 1.71 | | Total Trades | 3,847 | 2,134 | 1,713 | 412 |

Slippage sensitivity: each additional 0.5 tick reduces CAGR by ~2.7% and Sharpe by ~0.3. Walk-forward out-of-sample Sharpe: 0.94 (20% degradation vs in-sample — within acceptable threshold). COVID 2020 drawdown: -16.4% portfolio peak-to-trough, with volatility targeting automatically reducing average leverage to 1.8× during the crash vs 3.5× normal.

Contributions

  • Percentile-based noise area framework replacing standard deviation bands, providing more robust regime adaptation
  • Volatility-targeted position sizing with 8× hard leverage cap, demonstrated to automatically de-risk during crisis periods (COVID 2020)
  • Walk-forward validation over 15 years (2011–2026) spanning multiple regimes with 20% acceptable out-of-sample degradation
  • Quantification of the overnight risk elimination benefit: strategy avoids 100% of gap events that account for ~35% of buy-and-hold drawdowns

Paper

Author

Frankline Misango Oyolo Quantitative Research Division, Arithmax Research Frankline@arithmax.com — Published: March 2, 2026