Here is how I would explain that VXX ‘works’ as a simple introduction.
Think of VXX in two parts.
Part 1 is the short term, where VXX tries to replicate short term moves in the VX future. Specifically the 30-day spot on the VX curve. I have never done any regression but I think the correlation is quite high, so I believe it does a good job here. So if someone wants short term replication of that part of the curve, VXX could be useful.
Part 2 is anything longer than I guess 3-4 days. There is a large long term cost to producing this short term replication. Because futures have expiry dates (there is no 30-day VX future), the VXX is comprised of 2 near term futures to get to the required average expiry. This needs to be constantly rebalanced, which is very costly due mainly to the contango of the volatility curve (there is also the bid-ask spread). Every day the VXX rolls 5% of its portfolio, selling the cheaper near month contracts and buying the more expensive next month contracts. Said another way, every day the VXX is selling low and buying high (and accepting the bid/paying the ask). In the long term the VXX has destroyed massive value.
Other notes:
- on backwardation: when the short term curve is high during VX spikes, this can be a benefit for VXX (the rolls will help)
- on reverse splits: because of part 2 and the VXX’s massive depreciation, it has seen numerous reverse splits to raise the price back to visually credible levels
While on this subject, I have a VXX question for ET: Does anyone know how much short term buying/selling flow can distort the VXX price? I suspect there may be short term effects but I have never quantified the size or length. But maybe they aren’t exploitable as I have never heard of anyone arbing this. Any thoughts/observations?