"Another way to think about the trade is this: by combining a short position in VXX with a long position in the front-month futures, we are in effect creating a residual exposure in the value of the second month VIX futures contract relative to the first. So this is a strategy in which we are looking to capture volatility carry, not at the front of the curve, but between the first and second month futures maturities. We are, in effect, riding down the belly of volatility curve." I think I mentioned this in another thread at some point in time but any kind of overlaid short VXX/long VX strategy is really just playing games with VX calendar spreads eventually. Let's say you started today with the following data straight from iPath (http://www.ipathetn.com/US/16/en/details.app?instrumentId=259118): Index Components(as of 02/21/2018) Index Components Weightings % CBOE VIX Future MAR 18 83.75% CBOE VIX Future APR 18 16.25% and expirations of: VX H2018-CF 20180321T0000.00 (Wed) VX J2018-CF 20180418T0000.00 (Wed) Ignoring position sizing of the VX lots vs VXX percentages (which may very well be where any money in this arises from), if you start out with short VXX + long VXH18, you're basically long H18 at 16.25% (-83.75 + 100%), short J18 at -16.25% (VXX handles that). As time goes on and VXX becomes 50/50, you're then long H18 at 50% and short J18 at 50%. At 20/80, long H18 at 80%, short J18 at 80%. I can't see how scaling into a long M1:M2 spread is supposed to magically make money with a curve typically in contango. If you instead went long VXX and short VX H18 you'd scale into a short M1:M2 calendar spread instead. BTW: one of his charts shows VXX-VX, not VX-VXX although I can't tell if that's just coincidence or what he really backtested. Seeing as how the entire thing talks about short VXX the entire time it would seem not to be the case. I don't understand where the alpha is here.