VXX/XIV Pairs Trade? (An Unfinished Thought)

20Jan11

For the uninitiated, the ETF VXX “tracks” the VIX, while the new ETF XIV (launched 11/30/2010) tracks the inverse daily change of the VIX.

I put “tracks” in quotes because there are fundamental issues with both ETFs that mean they don’t actually follow the VIX all that well (read more from VIX & More).

But despite tracking the VIX badly, VXX and XIV have tracked each other’s inverse pretty closely. In the graph below I’ve assumed that we were long 50% VXX and 50% XIV, rebalanced daily at the close (frictionless), from 12/01. Because VXX is long and XIV short the VIX, this is a pairs-trade of sorts.

 

Geek note: XIV acted strangely in its first couple of days of trading so I’ve started the graph above on 12/01. Weird starts are common in new indexed products – chalk it up to working out the kinks.

The pair has (so far) oscillated around a flat line, but occasionally deviated sharply, possibly offering a pairs-trading’esque opportunity.

In the graph below I’ve applied to a 10-day moving average to the “neutral” pair shown above and assumed we went “long” (long VXX/long XIV) at the close when the neutral pair was below the MA, and “short” (short VXX/short XIV) when above. Results are frictionless.


[linearly-scaled, growth of $1]

In this hypothetical world, the strategy has been uber effective, returning roughly 61% annualized with 76% of all days positive. Note: there’s nothing special about 10-days; any MA length has been similarly effective.

There are three reasons though that this is an unfinished thought and I’m not all that excited (yet) about the results:

1. Both VXX and XIV are difficult to short, so playing the “short” side of the trade would be difficult (having said that though, the “long” side of the trade has actually been the more effective of the two).

2. A bit of slippage wouldn’t have killed the strategy (there’s a lot of meat in these returns), but it’s definitely a concern, especially for XIV which is, for the moment, not very liquid.

3. And the most obvious, with just two months since XIV launched, there just isn’t enough data (yet) to assume that this relationship is going to hold or that these sharp deviations between the two ETFs will continue once XIV liquidity improves.

I mention the idea here as brain food and to keep it on the radar as something to revisit in the future. As always, more to follow!

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.



9 Responses to “VXX/XIV Pairs Trade? (An Unfinished Thought)”

  1. 1 eber terandst

    Interesting, but the differences (the average trade) seem to be tiny. Are you sure this will take any realistic friction ?
    eb

    • 2 MarketSci

      Hello Eber: in a liquid vehicle, yes. Average trade return over this short test = 0.44% (average hold time = 2.3 days). Of course, XIV isn’t very liquid (see note #2 above) at the moment, part of the reason I think this is just something to keep on the radar to revisit in the future. michael

  2. 3 john malie

    Maybe I am not understanding but isn’t “short VXX” the same as “long XIV” and vice versa. So the (short VXX/short XIV) trade is the same as (long XIV/long VXX) is it not?
    So both trades are the same?

    • 4 MarketSci

      Hello John – short VXX is the same as long XIV (and vice-versa) “conceptually”. But they’re two different ETFs provided by two different firms and there’s going to be drift between them. The point of this post is that that drift (at least so far) could have been traded profitably. michael

  3. 5 theta

    Nothing strange here. This is normal consequence of the fact that XIV has to adjust its short VXX holdings on a daily basis to maintain the same absolute dollar exposure at the end of every trading day. As a result it will have to buy more VXX when VXX rises and sell VXX when VXX drops, in other words buy high and sell low which means its logarithmic performance will be lower than the inverse of VXX. I’m saying logarithmic because of course if VXX drops 50%, XIV will probably rise more than 50%, but less than 100% which would have been its return in an ideal scenario of VXX dropping a little bit every day.
    It’s exactly the same situation with any pair of “normal” plus “inverse” ETF, there are several out there, for the SPX, some sectors, etc. The effect is amplified for 2x and 3x leveraged ETFS as the cost of rebalancing in up and down days is magnified accordingly.

    • 6 MarketSci

      Hello Theta – I have to disagree. While I think that’s one reason we’ll see drift between the two over time, here we’re talking about 1-day divergences that are far too large to be explained away this way.

      My own guess? The lack of trading volume in XIV is resulting in a less then accurate closing price (i.e. a closing price that couldn’t reasonably execute) and that as volume in XIV increases (which I expect it will) we’ll see this arbitrage opportunity disappear.

      michael

  4. 7 ron

    How have the updated results been from this pair trading strategy ?
    Column should not be an issue now.

  5. 8 ron

    *Volume


  1. 1 VXX/XIV Pairs Trade? (An Unfinished Thought) « MarketSci Blog | leroygardner.com

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s