Intraday Momentum Trading Strategy (19.6% Annual Returns)

This strategy, which we found in a research paper by Carlo Zarattini et al., titled An Effective Intraday Momentum Strategy for S&P500 ETF (SPY), focuses on exploiting intraday trends in the S&P 500 ETF (SPY) driven by persistent supply-and-demand imbalances. Unlike traditional momentum models that only trade at the end of the day, this approach identifies and enters trends as they develop.

We call the strategy an intraday momentum trading strategy. We have not backtested the strategy ourselves.

Related reading: –Day Trading: Momentum Strategy

The Core Trading Rules

The foundation of the strategy is the authors’ Noise Area, a mathematically defined equilibrium zone where prices are expected to oscillate without a clear trend. The authors were rather “cryptic” about the trading rules, so we recommend reading the code (see link below) or the original paper.

• Dynamic Boundaries: The boundaries of this area are time-of-day dependent and are calculated based on the average price movement from the open over the previous 14 days.
• Overnight Gap Adjustments: To ensure accuracy, the boundaries are adjusted to account for overnight price gaps, which often signal immediate imbalances.
• Entry Signals: A long position is initiated when the price crosses above the Upper Boundary, indicating abnormal buying pressure. Conversely, a short position is triggered when the price falls below the Lower Boundary.
• Stop loss: Volume Weighted Average is used as a protective stop (more below).
• Dynamic position sizing: Dynamic adjustment of position size based on daily market volatility
• Semi-Hourly Execution: To filter out “noise” and transient price spikes, trades are only executed on semi-hourly intervals (at the top or bottom of the hour).
• Commission fee of $0.0035 per share, which represents the entry-level commission charged by Interactive Brokers.

Here’s an example of such a boundary:

A key characteristic of this model is that its noise thresholds are dynamic and change throughout the trading day. As a result, the price movement needed to signal a demand–supply imbalance is not constant: it differs, for instance, between the first 30 minutes of trading and the first six hours.

In general, the average intraday move from the Open increases as the day progresses, typically reaching its maximum around 16:00. As illustrated by the simplified example in Figure 1, a decline of just −0.30% from the Open by 10:30 was sufficient to indicate abnormal selling pressure, only half of the move required to trigger the same signal by 15:30.

The authors provided two trade examples in their research paper:

If you want the code to better understand the trading rules, please look at this source:

• Python code

Trailing Stops and Dynamic Sizing

The sources highlight that the most effective version of this strategy utilizes sophisticated exit and sizing rules to protect capital:

• VWAP Trailing Stops: The model uses the Volume Weighted Average Price (VWAP) combined with the current Noise Area boundary as a trailing stop. If the price crosses either level, the position is closed immediately to mitigate the risk of sudden reversals.
• Dynamic Position Sizing: Rather than using a fixed amount of capital, the strategy adjusts exposure based on daily market volatility. It targets a daily volatility of 2%; if the market is twice as volatile as usual, the position size is halved.
• No Overnight Exposure: All positions are strictly closed at the market close (16:00) to avoid the risks associated with overnight gaps.

Performance and Results

Through rigorous backtesting from May 2007 to early 2024, the refined intraday momentum trading strategy delivered exceptional results net of commissions and slippage:

• Total Return: The strategy achieved a staggering 1,985% total return.
• Annualized Return: It maintained a consistent 19.6% annualized return.
• Risk-Adjusted Gains: The strategy produced a Sharpe Ratio of 1.33, significantly higher than the 0.45 produced by a passive buy-and-hold strategy on the SPY during the same period.
• Market Independence: The strategy showed a beta slightly below 0, meaning its returns are largely independent of broader market trends, providing excellent diversification.
• Volatility Resilience: Performance historically improves during high-volatility regimes, with the Sharpe Ratio escalating to 3.50 when the VIX exceeds 40.

Here’s the equity curve of the strategy:

The authors used a lockback period of 14 days. However, they provided a table showing that a wide range of lookback periods were profitable.

The paper was originally published in the summer of 2024 and included results up to early 2024. Since it was published, it has performed well in out-of-sample:

Additional Comments

The sources conclude that this strategy is highly effective because it capitalizes on the market’s under-reaction to news, allowing information to be slowly incorporated into prices through persistent intraday trends.

Additionally, the model’s success is bolstered by the delta-hedging activities of institutional dealers, which can amplify price movements when they are forced to rebalance their portfolios