Missing Signals Targets

Hi There,

I have been looking at the Neutralised Signals Targets provided by Numerai and it seems like there are quite a few “holes” in the data. While I’m looking at just filling the missing periods with 0.5 values, I’m worried that keeping those periods or the stocks with missing data will lower correlation/ranking accuracy.

For example a stock like AAON which has been trading normally throughout has 6 months missing in 2010 and nearly 5 months missing in 2012. (see here: Google Colab)

I’m curious to know how most of you have dealt with these type of missing targets? I’m also curious to know why/how these targets are missing?



1 Like