0
$\begingroup$

I want to monitor HF/CTA long/short position and calculate beta on different HF indices in Excel/VBA, see graph below. I can't seem to find any papers on "Multi-factor based rolling beta", so my question is:

Is the multi-factor rolling beta simply the "=LINEST" excel-function where the given HF index is regressed upon multiple log-return series from e.g. commodity, bond and currency indices. And practically, the rolling aspect then comes the regression (over 30 days) being dragged down on the whole series (spanning approx 2 years in this graph).

If not, could someone please explain how multi-factor rolling beta is calculated or point me in a direction of how to find out?

Best regards

enter image description here

$\endgroup$
1
$\begingroup$

That's certainly the theory. But you very quickly run into massive multicollinearity issues trying to unpick between stock, bond, currency and commodity risk on a 30-day (in fact 21) window. The betas easily blow out to plus or minus infinity, which very quickly becomes very embarrassing.

So most analysts put in some kind of factor-based workaround, alluded to above in the "multi-factor based" bullet point. There a few ways to do this (PCA, synthetic risk-parity, even just preset asset "buckets"); but most try to create uncorrelated buckets of assets. If they're uncorrelated, there is no multicollinearity. So you can infer from the factor/bucket exposures, the exposures for the assets within your factors/buckets.

I used to build these kinds of models professionally back in a past life; and honestly reckon I'd spend as much time trying to work out why my model said X and the client's said Y (that could be sometimes the other way) than actually gaining any insight from them. But why then do them? Well, they are presentational catnip to investors.

Sorry if that's not a very proper quant ending there ;-)

$\endgroup$
1
  • $\begingroup$ Thanks Demully! :) Opposed to rates space where I'm comfortable in interpreting the first 3 factors, I'm not sure how the PCA factors should be interpreted here. The idea expressed in this post was to follow the CTAs exposure to a specific type of asset fx stocks, but this is not straightforward when doing PCA? I can understand that the first componant should be interpreted as some sort of "market exposure"? Also, what do you mean when you say asset buckets? Disaggregation of e.g. the commodity index into indices of gold, copper etc.? And how should these be weighted? Thanks $\endgroup$ – fdp1996 Jan 20 '20 at 11:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.