By D. Thomakos - code with prompts to Claude
In one of my previous post, the Cybernetic Speculator, we had the opportunity to work with a model that used the flexible specification of Wiener integrals, resulting in a non-linear, kernel-based, forecasting engine. In this post, I continue in the same spirit (but not in code and approach) with another flexible inference method, the Gaussian Process Regression - which results in a GPT: Gaussian Process Trading! You should be getting the book from the previous link and you can find more advanced code here too.
What you get in this post: Now, we shall not spend the time here to go over in detail the methodology used for this post because you can read it all - with considerable side extensions, make sure you catch them all! - in the document that you will find at the end of this post. Plus, if you scan my github repository you will find more than one version of the code used for this post with pleasant results and surprises for your modeling. And what is it that, beyond code, that you get in the post? A powerful forecasting and trading methodology that is (a) expandable in terms of the variables that you can use (b) flexible to handle a number of different specifications (advanved kernels and methods of estimation) and (c) a complete backtesting engine (one more time) with an easy interface and many more suggestions for future use. Don't wait - get your hands in the technical summary below and the code and play (speculatively always) around with your own GPT!
