Wednesday, May 31, 2006

When the VIX Itself Becomes Volatile

Yesterday, we had an enormous daily rise in the VIX of over 30%. To put this into perspective, we have only had one larger VIX spike in percentage terms since 1998, and that was the day we opened after 9/11. A chart in my most recent post on the Trading Psychology Weblog puts this into perspective.

What typically happens after a single day spike in VIX?

Since January, 1998 (N = 2108), we have had 27 occasions in which VIX has moved more than 15% in a day. All but one of these occasions were market declines. The next day, the market (SPY) was up by an average .37% (18 up, 9 down). That is much stronger than the average daily change for the entire sample (.02%; 1092 up, 1016 down).

Just as important, the five days following the VIX spike day showed much higher VIX (and price) volatility than average. The average daily VIX change over the next five days was 7.91%, and the average daily price change was 1.42%. Those compare with the averages for the sample of 4.13% and .89%, respectively.

After a five-day period of VIX volatility averaging 9% per day or more, the next day in SPY averaged a loss of -.18% (11 up, 15 down), but the next five days averaged a gain of .34% (14 up, 12 down). Once again, VIX volatility led price and further VIX volatility. Over the next five days, the average daily price change was 1.39% (vs. .89% for the sample), and the average daily VIX change was 7.14% (vs. 4.13% for the sample).

Overall, when the five-day average VIX change is 6% or greater (N = 287), the next five days in SPY average a gain of .67% (178 up, 109 down). This bullish tendency exists even when VIX levels following the VIX volatility are below 20. Interestingly, there is no edge one day out, however, in any of the data.

In short, volatility in the VIX begets volatility, with bullish implications five days out. That having been said, VIX percentage changes can be misleading, as recently pointed out by Adam Warner in his excellent blog. Nonetheless, look at the time periods with the highest five-day periods of average VIX change: April, 2005; September, 1998; March, 2004; and September, 2001. These were good times to buy stocks for an intermediate-term hold.