First serious run of 11/4 neural network gave EER (Expected Excess Return) unsimilar to Matts original EER.
calculated average absolute difference of matts values versus mine. Though not too different on average, a visual pair-wise comparison said otherwise. Why are they SO different?
EER not same
A second serious run on 11/20 also showed similar differences in pairwise comparison
I am starting to think that this might be reflective of Matt’s own troubles of differing results, and that he began utilizing his leave-one-out method to solve the problem
I will need to focus on using leave-one-out in order to compare neural net runs of the same data, from run to run. They shouldn’t differ drastically. Otherwise, forecasts will likely differ too much run to run on the same data and setups.
Script written for portfolio optimization
Documentation of the linkage of how Matt tied the NN forecast into the Sharpe optimization algorithm GMQP and how he got results as to which stocks to buy.