Adaptive Expert Advisors and Forward Testing in Currency Trading

Summary

In this interview, a participant in the Automated Trading Championship describes how he approaches building currency trading robots. He favors programming strategies because discretionary trading makes it difficult for him to stick to a plan, and says that finding robust rules and parameter values is harder than implementing them. He warns that optimizing for maximum historical profit can lead to poor future performance and recommends forward testing as a check. His contest system trades three currency pairs on the daily timeframe, placing pending orders based on current prices and prior days’ OHLC values. Position sizes adjust to daily movement and existing positions, while other pending orders can limit losses.

The account offers a practitioner’s example rather than a controlled evaluation. The author says he tested the system on two years of history followed by forward testing, but the interview gives no detailed performance results. The contest lasted less than three months, and its results should not be treated as evidence of long-term profitability. He also advises beginners to study platform documentation and start with a simple moving-average crossover robot.

Key ideas

• Historical parameter optimization can lead to overfitting, so forward testing is a useful additional check.
• The described robot uses prior daily OHLC prices to place pending orders across three currency pairs.
• Its position sizing responds to daily price movement and the state of other open positions.
• The author chose automation partly because he found it difficult to follow a discretionary trading plan.
• The interview provides personal experience, not a controlled demonstration of lasting profitability.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.