Tuesday, June 26, 2007

The Language of the SPX:VIX Ratio

A fellow blogger at The dk Report inquired about the analytical value of the SPX:VIX ratio and what it may be telling us at present.

The simple answer is that I don’t know…but I’ll see if I can articulate my uncertainty a little more eloquently in the space below.

Taking a step back, the reason I first even bothered to mention the SPX:VIX ratio on this blog was to address some comments by Ron Sen at Technically Speaking, who blogs about the SPX:VIX ratio on a regular basis and asked rhetorically back on February 19th if this ratio was an effective risk metric. My rejoinder was that it does not make sense to compare a trending number like the SPX with an oscillating number like the VIX. Given that the long-term gains in the SPX approximate 10% per year, I somewhat flippantly suggested that a 10% trend line might be added to this ratio. Having tossed this idea out, I proceeded to look for some insight using the distance between the ratio and the trend line in the wake of February 27th and again when the markets seem to have settled down a month later.

Regarding the 10% trend line, however, I would be remiss in pointing out that my choice of the 10% value was relatively arbitrary (though supported by Ibbotson data) as was my selection of a starting point for the trend line, which turned out to be the default setting for a StockCharts.com monthly chart I created at that time. I used September/October 1991 as a starting point; if you use a different starting month, you can come up with a substantially different trend line.

While keeping one eye on the SPX:VIX ratio, in the last several months, I have spent more of my time and energy researching various correlations between the SPX and the VIX – and will present some findings and interpretations on this subject in the near future. I have already posted about one finding: that when the SPX and VIX are highly positively correlated for short periods of time, this does not augur well for the markets.

High on the priority list for my one man R&D department is to develop a model for long-term volatility forecasts. The VWSI and mean reversion analysis that I have relied heavily upon to this point is best suited for a 5-10 trading day time horizon, though they are sufficiently robust to be applied out to about 20 trading days. It is possible that an analysis similar to that of the SPX:VIX plus a 10% trend line might be appropriate for long-term volatility forecasting, but this remains to be seen. I will also look to tweak the VWSI in order to enhance its predictive power going out 1-3 months. I consider the VWSI and mean reversion to be excellent tools for trading during the current options cycle, but in terms of volatility, I think the Holy Grail lies in being able to look out a quarter or two. This will certainly be an area I look at closely going forward. As always, I encourage reader input.

1 comment:

  1. Gann, Jeff Cooper, and others argue about time being the most important factor in these matters.

    I've tried all kinds of cycle analysis, especially admirable is the work of Eliadis, but it hasn't helped me...other than occupying time.

    For what it's worth, SPX:VIX is out to almost 3 standard deviations (20,3) and the VIX is over 25 percent out from the ten period average.

    Is greed on the horizon or closer?

    ReplyDelete