Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Betas in the risk model return as inf or nan

Hi,

Has anyone run into an issue with the risk betas returning inf or nan? I'm having a hard time fixing this issue.

ValueError: NaN or Inf values provided to FactorExposure for argument 'loadings'.
Rows/Columns with NaNs:
row=Equity(33655 [HYG]) col='momentum'
row=Equity(33655 [HYG]) col='short_term_reversal'
row=Equity(33655 [HYG]) col='size'
row=Equity(33655 [HYG]) col='value'
row=Equity(33655 [HYG]) col='volatility'
Rows/Columns with Infs:
None

It occurs around July 2007 as part of a weekly rebalance. I've tried this:
context.risk_factor_betas = pipeline_output('risk_pipe').replace([-np.inf,np.inf],np.nan).dropna()

1 response

Hi David.

Try this :
context.risk_factor_betas = pipeline_output('risk_pipe').dropna()