Discussing the article: "Self Optimizing Expert Advisor With MQL5 And Python (Part IV): Stacking Models"
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: Self Optimizing Expert Advisor With MQL5 And Python (Part IV): Stacking Models.
Today, we will demonstrate how you can build AI-powered trading applications capable of learning from their own mistakes. We will demonstrate a technique known as stacking, whereby we use 2 models to make 1 prediction. The first model is typically a weaker learner, and the second model is typically a more powerful model that learns the residuals of our weaker learner. Our goal is to create an ensemble of models, to hopefully attain higher accuracy.
We shall turn our attention to forecasting the NZDJPY currency pair. We desire to algorithmically learn a trading strategy from the data we will collect on the symbol from our MetaTrader 5 Terminal. As humans, we may be naturally biased towards trading strategies that are aligned with our own beliefs and interests. Machine learning models are also biased. The bias of a machine learning model, is the extent to which the assumptions made by the model are violated. Our trading strategy will rely on an ensemble of 2 AI models. The first model will be trained to predict the future close price of the NZDJPY pair, 20 minutes into the future. The second model will be trained to predict the amount of error in the prediction made by the first model. This technique is known as stacking. Our hope is that by stacking 2 models, we will be able to overcome our human bias, and hopefully this will be enough to lead us to higher levels of performance.
Author: Gamuchirai Zororo Ndawana