Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Treasury cleanup #793
Rewrites most of
The bulk of the work here is in rewriting the treasury loader and benchmark loader to use
The benchmark returns loader has some more significant deviations from the previous behavior, but I don't see anything wrong with the code. @ehebert do you know of a good place to find reference values for S&P500 returns to check my current results against the previously-generated results?
One idea for double checking against old calculations would be to look at the output/cache in ~/.zipline/data using the different versions.
I've already done that in both cases. For the treasuries there's exactly one case where there's a potentially significant (on the order of 0.002) difference for a single date on the 3month curve, which is due to the fact that data.treasury.gov and federalreserve.gov produce different results there. I'm not sure how to verify that one or the other is correct.
For the returns benchmarks there are more nontrivial differences, which is why I was looking for a canonical place to compare our current return calculations vs the old ones.