Brute Force Optimization of VRP Strategy

Of all of the simple VIX trading strategies we’ve tested on this blog, one of the most effective historically has been DDN’s Volatility Risk Premium (VRP). Recall that the strategy is shorting vol (ex. long XIV) when the VIX is trading at a premium to historical S&P 500 volatility, and long vol (ex. VXX) when it’s trading at a discount (1).

Below are the results of DDN’s VRP strategy as originally tested, trading XIV and VXX, from 07/2004 to present. Read about test assumptions, or get help following this strategy.


The original strategy used the following formula to calculate the premium/discount of the VIX vs historical volatility: the 5-day average of the [VIX index – (10-day annualized standard deviation of SPY * 100)].

In this post I focus on those two parameters, the 10-day standard deviation and 5-day smoothing average, and determine what the optimal combination of parameters has been historically.

The two tables below assume we ran our strategy using all possible combinations of lookbacks for the standard deviation (columns, from 2 to 15) and smoothing average (rows, from 1 to 12).

The first table shows each run’s Sharpe Ratio (return vs volatility), and the second table each run’s Ulcer Performance Index (return vs drawdown). High numbers are better than low numbers.



The best cluster of results is centered on lookbacks of 4-days for the standard deviation and 8-days for the smoothing average (highlighted in bold in the tables).

These values are considerably different than the original strategy’s 10 and 5-days. Below I’ve rerun our backtest trading XIV and VXX, showing the original strategy (grey) versus our new optimized strategy (blue).


The two variations of the strategy have performed nearly identically with the exception of 2013 when the two variations diverged sharply. The numbers…


Does a handful of trades in one single year justify calling one set of parameters better than another? Probably not.

In all likelihood, either variation of the strategy will be as good as the other in the future, and the strategy as originally presented is already very close to optimal. As with most trading strategies, the concept being exploited is much more important than the specific parameters chosen.

Having said that, I think that there’s value in presenting divergent views, so when this strategy (like most we cover on this blog) signals a new trade, we’ll be including an alert on the daily report sent to subscribers. This is completely unrelated to our own strategy’s signal; it just serves to add a little color to the daily report and allows 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

Wonk note: For readers looking for a more thorough understanding of the volatility risk premium and how it relates to future VIX ETP returns, I suggest our post Four Graphs to Rule Them All. Using the parlance established in that post, the VRP strategy shown here is attempting to use relationship #1 (VIX vs historical volatility) and its ability to predict relationship #2 (VIX vs future realized volatility), to hopefully (and more importantly) predict relationship #4 (VIX futures vs future realized VIX).

Posted in Strategy Backtests.