Discussing the article: "MQL5 Wizard Techniques you should know (Part 80): Using Patterns of Ichimoku and the ADX-Wilder with TD3 Reinforcement Learning"
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Check out the new article: MQL5 Wizard Techniques you should know (Part 80): Using Patterns of Ichimoku and the ADX-Wilder with TD3 Reinforcement Learning.
This article follows up ‘Part-74’, where we examined the pairing of Ichimoku and the ADX under a Supervised Learning framework, by moving our focus to Reinforcement Learning. Ichimoku and ADX form a complementary combination of support/resistance mapping and trend strength spotting. In this installment, we indulge in how the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm can be used with this indicator set. As with earlier parts of the series, the implementation is carried out in a custom signal class designed for integration with the MQL5 Wizard, which facilitates seamless Expert Advisor assembly.
In the family of RL algorithms, Twin Delayed Deep Deterministic Policy Gradient (TD3) is emerging, in some circles, as a solid candidate for financial applications. TD3 is designed for continuous action spaces, making it particularly well-suited to trading problems where position sizing and timing are not binary but require fine-grained control. If we compare it to its predecessor, DDPG, TD3 brings in crucial stability improvements such as using more than one critic, adding noise smoothing to the target actions and also delaying policy updates to prevent overfitting to transient fluctuations.
Our core aim with this article, as always, is to demonstrate how TD3 models, when trained in python, can be merged into MQL5’s Expert Advisor framework, for prototyping. Specifically, we aim to demonstrate how a TD3 actor network that gets exported to the ONNX format can be wrapped and consumed inside a custom signal class. As always, these signal classes then get assembled into a trading robot using the MQL5 Wizard.
To anchor this discussion with some more ‘practicality’, the article will conclude with a review of some forward tests performed in MetaTrader’s Strategy Tester. Three reports are analyzed each covering the 3 signal patterns we had chosen for further study in the ‘Part-74’ article. These were the signal patterns 0, 1, and 5.
Author: Stephen Njuki