Back testing from 2004
Been looking at short the front month, long the third out (or fourth if you count front as 1st). 1x Leverage, CAGR 33%, Max DD 45%. Compares favourably to simple short of front month at 1x leverage which has a very similar CAGR but a 92% max drawdown.
To be clear this rebalances on a daily basis: 50/50 between the returns from the long and the short sides. IE each day the return is composed of 50% of the return from shorting the front month and 50% of the return from shorting the third month out.
The long side is hard to match in ETN's ETCs. I have not looked into the reason why.
I'm engaged on a machine learning exercise on VIX futures contracts and this is a spin off I happened to notice.
I'm not planning to trade this or look at it in any further depth at the moment but bearing in mind people can upload futures from Quandl to Q this might be of interest.
Note that I used individual futures contracts - I am not a fan of back adjusted price series for futures these days. If you use individual contracts watch out for contract rolls and deal with the gaps appropriately or the results will make no sense.
No doubt some clever soul with more patience than I will try it in Q!
