IMPORTANT: Fixed a problem with BetaFunc and delisted stocks

Dear All,

We corrected an issue with the Beta1Y, Beta3Y, Beta5Y and BetaFunc() which was causing inactive stocks to return very low beta values historically. You may see some changes in the results for simulations which use those Beta factors, but they should be minor differences except in cases where the simulation is holding a small number of stocks and Beta factors are heavily weighted in the ranking system.

This issue was always present since we switched data provider, but after a recent release to fix future prices it affected more stocks.

On a side note…

be aware of the limitations of individual stock’s Betas. They are very volatile and may not be what you expect. Take IBM for example Beta’s for example

The first plot compares 5Y betas. Beta5Y uses weekly samples and BetaFunc(21,61) which uses monthly samples. Did you think that monthly samples would be so volatile?

The second plot compares 3Y betas , one is done weekly, the other daily. Daily one is a lot smoother.

It gets worse …

The fundamental chart doesn’t show all data points with long periods. Here’s IBM’s betas when you zoom in to a three year period

The monthly beta experiences harmonic behavior because 60 out 61 samples are the same only every 4 weeks. This may be ok if your simulation frequency is also 4 weeks, but with weekly simulations you will get very different betas at each rebalance.

Long story short, know the data. Always visualize it in the Fundamental Chart


1 Like