Combining Supervised Forecasting with Feedback Control in Trading

Summary

The article replaces a moving-average component in an earlier feedback-control trading design with a supervised statistical model of EUR/USD. It outlines a workflow that gathers daily OHLC data, labels future prices at a defined horizon, compares candidate regression models, and trims older history based on forecasting error. The selected model’s output is then used within a trading application whose feedback controller adjusts behavior based on observed outcomes. The central idea is that a predictive model and a controller can play complementary roles: one estimates market variables while the other responds to system performance.

For its reported test, the article compares the combined system with its earlier baseline and gives improvements in net profit, gross loss, accuracy, trade count, and profit factor. These are results from the described experiment, not evidence of durable performance across markets. The account is limited by a single currency pair and test setup, and the excerpt provides little detail about robustness, transaction costs, or independent replication. The reported results should therefore be treated as an example rather than a validated trading edge.

Key ideas

• The design combines supervised forecasts with a feedback controller that responds to trading-system behavior.
• The workflow compares regression models using historical EUR/USD data and future-price labels.
• The article selects a shorter, more recent training history based on forecast error.
• Its reported comparison favors the combined model and controller over the earlier baseline in several test metrics.
• Results from one described setup do not establish robustness across markets or live trading conditions.

Tags

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