Genetic Alpha Generation - Part 1
Data structure to represent alphas and a plan for the project
It’s time for a new project! In this one we will be building a genetic algorithm that automatically generates alphas.
Is the goal of this project an alpha generating machine that will make us a ton of money? While that would have been nice, unfortunately no. It’s more of an exploratory data analysis project and for confirmation. Do we see known alphas appear in our genetic algorithm? Can we infer better ways to handle those alphas? etc.
In this part of the project we will be getting all the data necessary, represent alphas as trees and define the fitness of an alpha.
In future parts we will be writing the genetic algorithm to actually make our alphas evolve and then we will be running a bunch of analysis on the generated data like how the fitness develops over time, if we get more momentum or mean reversion signals etc.
To not miss out on future articles on genetic alpha generation consider subscribing:
https://www.vertoxquant.com/eaf1ee4d
Table of Content
Data
Alphas as a Tree
Generating random Alphas
Fitness and avoiding overfitting
Final Remarks