Impact of the VIX Term-Structure on VXX, et al.
Estimating the impact of the VIX futures term-structure on volatility ETFs like VXX or XIV should be uber important to volatility traders (including recent converts like me), because most of the time it acts as a strong tailwind in favor of either trading short volatility (ex. short VXX/long XIV) or long.
Geek note: I’m using “term-structure” loosely to mean the impact of roll yield, time decay, and all other factors not related to immediate changes in the VIX itself.
Below is my work-in-progress estimate of the daily impact of the VIX term-structure on VXX returns since 2006 (reverse results for XIV). Strong negative values (read: contango) act as a drag on VXX returns, and strong positive values give VXX a boost.
VXX was launched in early 2009 and, as the graph above shows, since then there has been a consistent drag on returns from contangoed VIX futures (read more).
But had VXX launched just six months earlier (during the financial crises in 2008), the opposite would have been true for a while, with a daily boost of over 2% (independent of the movement of the VIX).
Geek note: for ease of readability, the graph above actually shows the rolling 21-day (1-month) average estimate, not a spot estimate for that particular day.
A Work-in-Progress
I’m not ready to share the math on this one just yet because (a) I’m still fleshing out my thoughts on the subject, and (b) I’d like to hear from readers re: others’ work they’ve seen on the subject (I’d much prefer to stand on the shoulders of giants).
I think these estimates are ballpark correct, and I’m using them to bias the direction of my vol trades today, but I also think there’s room to improve the math.
When we’re at extremes like we’ve been recently (with strongly contangoed futures) this is a moot discussion – trying to trade short-term volatility ETFs like VXX long is like walking upwind with an open parachute – but that won’t always be the case.
Being able to accurately estimate the daily impact of the VIX term-structure helps to ensure we stay on the right side of the trade both in actual trading and when backtesting vol-trading strategies.
Happy Trading,
ms
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: VIX & Volatility | 9 Comments
aren’t you a little late to the party?
A little late to what party? Has there been a rash of research done into the tangible daily impact of term structure over the last 5 years on short-term volatility ETFs that I’m unaware of? That’s a party I’d like to be invited to.
Like the idea. Not overly familiar with VIX futures, but is there enough liquidity in the futures that aren’t the nearest expiry, to give you reliable data here? Not overly clear what *you* mean by term structure here, or how you get your graph. If data was an issue, are options a bit more liquid given their use in hedging?
Very interested to see where you go with this!
Hello Russfx – I’m using the term loosely (and I should think of a better one). Really I’m trying to capture everything other than what you would expect if VXX were perfectly VIX mimicking (after accounting for changes in delta of VXX to VIX). So direct term-structure impacts like roll yield, in addition to time decay, any other drift between futures and the VIX, fund expenses, and (what else?)
This is still a work-in-progress, but the way I approached the problem so far is based purely on VXX/VIX daily returns, not using the actual first/second month contracts.
michael
Michael,
Interesting post. Where can I find historical information on the VIX term structure?
Hello Bill – as mentioned in the comment above, I’m using just VXX/VIX prices for this estimate. But if you’re looking for historical data on the VIX term-structure in the true sense of the word you can use the CBOE tool at:
http://www.cboe.com/data/VolatilityIndexes/default.aspx
michael
Yes, viewing the term structure is the appropriate way to grab a gauge on the underlying. As it’s a futures curve and contango/inversion tells alot as to how these ETF’s will move. From casual conversations with some traders, the trade of late, which u mentioned)until the ME set on fire has been bear futures spread (short the front month/Long the deferred month) & they have been cleaning up until now. Who knows maybe this spike is another sell opp as vol is mean reverting. Anyone’s guess. Great post and thought provoking as usual, MarketSci.
Don’t stop the info/data carnival.
Hi Michael.
Do you think the VIX:VXV ratio could be used to get a fair estimate of the VIX term structure?
Hello OBT: I haven’t taken that approach, so I can’t give an intelligent response.
I would think that the VIX:VXV ratio might give you a general idea of the nature of the term-structure (i.e. whether it’s in contango or backwardation), but there are just too many other variables in order to make an estimate of the specific % impact of the term-structure.
I think comparing VIX to VXX for example is more telling. And of course, comparing the front/second month VIX futures even more telling that that.
michael