A Twist on TradingMarkets.com’s RSI(2) Strategy

This is a twist on a VIX ETP strategy originally proposed by Trading Markets, and further tested by MarketSci. It is a Martingale-like strategy where the trader takes a position in XIV (inverse VIX) when volatility becomes overbought, and then progressively increases position size if volatility continues to rise.

Strategy results, based on TradingMarkets.com’s original strategy, trading XIV from mid-2004 to present:


Original strategy rules (1):

Divide the portfolio into 6 equal units. Buy 1 unit at the close if the RSI(2) of XIV closes below 10. Add 2 additional units if XIV closes below your entry price at any point. Add a final 3 units if XIV closes below your second entry price at any point. Close all positions when RSI(2) closes above 50.

Wonk notes: Transaction costs and slippage are assumed to total 0.1% of each trade (0.2% round-trip). Return on cash has been ignored. Data prior to the launch of each ETP has been simulated back to 2004. We’re able to do this accurately using a combination of the indices and the futures data on which these ETPs are based. Read more about simulating data for VIX ETPs.

* * *

The strategy spends little time in the market, but has produced mostly consistent results. The problem lies in those nasty drawdowns during the 2007/08 crash. These types of catastrophic events are of course the bane of a Martingale-like strategy.

But VIX futures during the 2007/08 crash were often strongly backwardated (favoring VXX, not XIV), meaning any long XIV position at that moment would have been extremely risky.

So in this twist to TM’s strategy, we test only trading XIV when the VIX futures term-structure agrees with the trade (i.e. is in contango). Results of this revised test are in orange, versus the original strategy in blue.




The revised strategy preserved most of the original strategy’s gains, while sidestepping the worst of the drawdowns. Again, most of the original strategy’s biggest losses came during periods of strong backwardation where the trader should logically have been wary about putting on a position in XIV, so despite a small sample size, I think the twist shown here has wings.

I’m using our own internal metrics to judge the state of the VIX futures term-structure, but results that are similar in spirit to these would be had using even simple approaches like the one covered in our previous post.

I present this test for discussion’s sake, but personally speaking, the idea of trading a Martingale-like system on something as volatile and potentially dangerous as VIX ETPs, makes me terribly nervous. These products can move against us very quickly, and I think, over the long-term, we’re better served looking to reduce position size as trades move against us rather than increasing it.

* * *

When the strategies that we cover on our blog (including this one) signal new trades, we include an alert on the daily report sent to subscribers. This is completely unrelated to and separate from our own strategy’s signal; it just serves to add a little color to the daily report and allow subscribers to see what other quantitative strategies are saying about the market.

Click to see Volatility Made Simple’s own elegant solution to the VIX ETP puzzle.

Good Trading,
Volatility Made Simple

(1) In TradingMarket.com’s original post, they shorted VXX when the RSI(2) of VXX closed above 90. In the real world, shorting VXX can be unreliable, so I’ve flipped the strategy on its head and assumed we went long XIV when the RSI(2) of XIV closed below 10. These are essentially the same strategy, with our version being easier to actually execute.

Posted in Strategy Backtests.