Follow up to VIX-based Pairs Trading Strategy


This is a follow up to yesterday’s VIX-based Pairs Trading (Market Neutral) Strategy (which by the way, has been added to the new Addendum).

A refresher on strategy performance (red) vs the S&P 500 (grey) from 1989:

[logarithmically-scaled, growth of $10,000]

Additional Reading

The concept behind the strategy is at least 3 1/2 years old. Links to others that came before me: Leistikow and Yu (academic paper, 2006), CXO Advisory (paper review, 2006), and CSS Analytics (mash up with our own Generals Lead the Troops analysis).

Strategy Losing Effectiveness?

Just eyeballing the chart, it’s pretty clear that the strategy as I described it has been less effective in recent history. To illustrate, the chart below shows the strategy’s rolling 5-year annualized return (blue, left scale) and volatility-adjusted return (red, right scale) from 1989.

volatility-adjusted return = annualized return / annualized std. dev.

Since about 2004, the strategy has failed to produce big results. At least part of that was likely a result of the low-volatility bull market of the mid-2000’s; it’s especially hard to generate market-neutral returns in a low-vol. environment.

But there was lot of volatility in the S&P 500/Russell 2000 pair during our most recent 2008-09 crises and yet still the strategy didn’t perform. That might be an indication that it has permanently lost some of its mojo.

Strategy Development Stuff (Warning: Extreme Geekery Ahead)

What this strategy needs to be really effective is an additional “moderator”, or put another way, another nifty observation that helps better refine the relationship.

Since the original post, I’ve been putting some time in to trying to understand just how to do that. Here’s where my equity curve stands today…

[logarithmically-scaled, growth of $10,000]

I was able to flatten out the big “shifts” in the original equity curve, particularly around 1998 and 2004, and produce a strategy with much more consistent results.

Unfortunately I can’t share my specific tweaks because I think this idea has the potential to eventually become one of our proprietary strategies, but I can say this…

This improvement actually had nothing to do with an additional moderator (that’s still to come). There’s a flaw in the logic of the strategy, as I originally described it, that’s related to the asymmetrical nature of the VIX. That flaw will cause the original strategy to lose market-neutrality, depending on the market regime in effect at that moment in time, creating bends in the equity curve that are NOT the result of the strategy gaining/losing effectiveness, but rather, the result of changes in the regime itself.

And that’s all I can say about that (sorry to be so vague). The strategy still needs additional work and it’s still waning in effectiveness, but I think the tweak above is definitely a step in the right direction.

Happy Trading,

. . . . .

To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.

11 Responses to “Follow up to VIX-based Pairs Trading Strategy”

  1. 1 Blue cat

    What a tease!

  2. 2 Rod

    Interesting, as always. Apart from the very interesting ideas, I must say I really like the clean layout of your pages and charts. Some of the other sites (eg Trading the Odds) are hard to follow – they just dump numbers and dont summarise.

  3. 3 kalki

    What is your formula for calculating volatility adjusted returns? I mean volatility used is pnl volatility or VIX? If pnl volatility then what time frame over which you use calculate the pnl volatility?

    • 4 MarketSci

      RE to kalki: in this particular post, I took the standard deviation of monthly (log) returns and annualized them (x sq. root of 12). The graph above is a 5-yr rolling stat, so the vol. is measured based on the same rolling 5-yr period. michael

  4. 5 mike

    Glad to see a flurry of posts! Have you run this using the Nasdaq & SP500? How were the results? Would using futures give you the leverage necessary to make the strategy work?

    • 6 MarketSci

      RE to mike: good question. I’ve run this with other low/high beta combos (DJI/S&P500 vs RUT/NAZ) and the S&P 500/RUT combo seems the most consistent.

      Futures would give the leverage, but two potential issues: (a) trading frictions, and (b) tracking error versus the index (see note at end of first post). It would need to be modeled. Unfortunately, this is beyond the scope of what I do, but would be very interesting to see.


  5. 7 quantivity

    Nice post, Michael. Any vol wing trader knows there are numerous flaws with the original strategy, in both symmetry and linearity. Factoring RVX would likely help too. Drop an email if you want to chat further refinements.

  6. 8 Kevin


    Have you looked at a scenario that shorts during the Misaligned segment?



    • 9 MarketSci

      RE to Kevin: I’m assuming this comment was re: the test of CSS’s “perfect alignment” and not this post? Assuming so, I generally don’t like the idea of using long-term indicators to short the market. The downturns during misalignment have come but they’ve been far between. Better I think (if I were only trading this one strategy) to just move to cash in that phase. michael

  7. 10 Kevin

    Thanks. And yes, I did accidently post this to the wrong strategy review.

  8. 11 tom

    Just came across this post – any update to your market neutral strategy using the VIX level as a gauge?

Leave a Reply

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

You are commenting using your 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