How to Find Winning Anomaly-Based EAs! From an EA Developer’s Perspective

I will explain what a winning anomaly-based EA is, its characteristics, and how to choose one.

Anomaly-based EAs are automated trading programs that utilize patterns or anomalies in the forex market to execute trades. These EAs capitalize on specific times or dates when certain regularities in the market occur.

Let me introduce two representative examples of anomaly-based EAs.

The first one is the Tokyo Fixing Trade EA.

The Tokyo Fixing Trade takes advantage of the significant movements in the market during the fixing time, which is the reference rate set by financial institutions. This occurs at 9:55 AM Japan time. Trading with a focus on this period can yield efficient wins. The logic behind each Tokyo Fixing Trade EA varies among developers. The trading frequency also differs, ranging from daily trades to only a few entries per month. Generally, EAs that limit their trading days to Goto-bi (a five-day interval in Japanese business practice), month-ends, or weekends tend to perform well.

The second one is the London Fixing EA.

The London Fixing refers to the fixing time in the London market, which is at 4 PM London time. There is a well-known anomaly where the pound tends to be bought heavily at month-end London Fixing. The primary strategy is to target the time when the pound-buying intensifies before the London Fixing. Historical data shows that the pound tends to rise at month-end, allowing traders to profit by buying the pound during this period. Besides these, any pattern that seems to recur without a clear reason can also be classified as an anomaly-based EA.

One major characteristic of anomaly-based EAs is that if their logic no longer aligns with the current market, they will stop winning. Winning anomaly-based EAs analyze vast amounts of historical data to extract highly reliable patterns. The more accurate this analysis, the higher the success rate of the trades.

I developed an EA called “London Time Trade.” It enters a new trade with a counter-trend approach after 4 PM Japan time and closes the trade within the day. Since its release on May 24, 2021, it has experienced wins and losses, with a slight positive forward performance as of May 2024.

There are several reasons why anomalies might stop working.

London Fixing used to significantly impact exchange rates, but its influence has waned in recent years. Additionally, several factors might explain why the Tokyo Fixing Trade has been underperforming since 2023. One reason is the reduced trade volume and the development of financial markets, leading to a decline in dollar-buying by real-demand players during the Tokyo Fixing. In the past, exporters would buy dollars around the Tokyo Fixing for month-end trade settlements, but such transactions have decreased in recent years. Another reason is that the Tokyo Fixing Trade has become more well-known among traders, potentially diluting the anomaly’s effect. Furthermore, the globalization and digitization of financial markets have led to more rapid and significant fluctuations in exchange rates, reducing the effectiveness of anomalies tied to specific times.

To address the biggest weakness of anomaly-based EAs-losing their effectiveness-a few measures are necessary to maintain consistent wins.

One measure is risk management. Even the best anomalies won’t always lead to successful trades. Winning EAs have risk management features to minimize losses, including appropriate stop-loss and take-profit settings. Another measure is forward testing. While backtesting with historical data is crucial, it cannot entirely predict future market conditions. Ensuring the EA performs well in forward testing is essential when choosing an EA. Additionally, analyzing the fundamental reasons behind an anomaly helps determine its current effectiveness. Anomalies without a solid foundation are unlikely to be reliable in the future. Another measure is to avoid EAs with opaque logic. Choose EAs that clearly explain how they generate profits. Transparent logic allows users to understand entry and exit timings, aiding in decision-making.

As an EA developer, I advise creating a portfolio of various anomaly-based EAs. If an EA starts losing because its anomaly is no longer effective, remove it from the portfolio. For EAs with disclosed logic, consider fundamental analysis to decide whether to continue using them.

I have developed dozens of anomaly-based EAs, but their performance is generally inferior to those using technical indicators. The reason is that anomaly-based EAs are often optimized for past market conditions without having an edge in future markets. EAs without clear logic tend to fail in forward tests. Avoid using anomaly-based EAs that do not perform well in forward testing.

Leave a Reply

Your email address will not be published. Required fields are marked *