Hello, I am trying to get something together to eventually be a competitor in the contest.
With that said, this is the most recent results of my Algo, and I was wondering if anyone is willing to provide feedback, as I am new to finance in general, but like writing in python, so forgive me if my model is archaic at the moment, or looks like patchwork, because it is.
Any input is greatly appreciated, as I don't even know if this model is worth putting in the competition as I have gotten most of the requirements satisfied thus far, but the returns seem shabby, and I am struggling to keep it within all 7 of the risk requirements and maintain positive returns. Is this common? I have some more improvements to make, but I am trying to see if this is even on par for course.
I am attempting to do a stocks on the move model, and project which stocks will move based on all the pipeline data, using machine learning. I pulled out the
class Prediction
as the one I had was making it practically give away the stocks, so any input or suggestions are welcomed! Any machine learning algorithm should be easily adapted into the structure. I am having trouble getting the algorithm to run at returns higher then 5% but it handles large amounts of capital decently. Is this anomalous, or something worth looking into further?
I am am trying to get help specifically with taking my momentum custom factor, and calculating the changes in the momentum, and multiply it by the volume, or another factor to represent the 'mass' of the stock, which might be more accurately represented by another dataset to which I am completely open to suggestion. I would then like to calculate the stocks "acceleration' per se. Then to factor this in to the alpha for the optimize portfolio. My other thought is to take and measure the acceleration of a few of the various universe's and get a market momentum factor and use that in the calculation of the machine learning process.
I am working now to create the class for the acceleration and other derivates, and to attempt to put them in a numpy series for factoring.
