I\'m trying to use the sample_cov function from PyPortfolio\'s risk_models, but for some reason it returns almost all nan on a certain df. I have a df of the S&P500 clos