Saturday, July 07, 2007

Anticipating Volatility in the Stock Market

Traders commonly focus on market direction, trying to anticipate bull or bear moves. Equally important, however, is anticipating market volatility. Without an understanding of how volatile the market is likely to be, it is difficult to set appropriate stop loss points and profit targets. But how can we know how much movement to expect in the market?

I decided to look at several possible predictors, going back to the start of 2004 (N = 883 trading days) in the S&P 500 Index (SPY).

Strongly associated with a day's volatility is the volume of trade for that day. Indeed, the correlation between SPY daily volume and the high-low trading range for the day session in SPY is an impressive .49. That means that about 25% of all variance in market movement can be explained by the participation of many traders--and large traders.

Another powerful predictor of daily volatility is the level of option volatility: the VIX. The correlation between the closing VIX level for the day and that day's volume is .43. Clearly, volatility among options premiums is associated with volatility in the daily trading range of stocks.

When we join those two factors, daily volume and VIX correlate in a multiple regression about .60 with the high-low range for SPY.

If we look at the prior day's closing VIX and the current day's high-low range in SPY, we find a correlation of .36. Interestingly, the current day's high-low range also correlates .23 with the size of the opening gap in SPY (i.e., the absolute magnitude of the move from the previous close to the current day's open). The correlation between the current day's high-low range and the prior day's high-low range in SPY is .20.

A multiple regression of the prior day's VIX, trading range, and the opening gap yields a multiple correlation of .38. That suggests that, of the preceding predictors, the prior day's closing VIX accounts for the lion's share of variance in the present day's volatility.

Accordingly, it makes sense to use the VIX as an estimate of the likely volatility for the current trading day, but to continuously update this estimate on the basis of the current day's trading volume. By identifying whether the current day's volume is above average, average, or below average and then seeing how these volume levels have interacted with VIX levels in the past, we can make reasonable estimates as to the likely movement for the coming day.

More on this strategy to come.


First Half Hour as a Volatility Indicator

Intraday Volume and Volatility

What You Can Learn From the Opening Minutes of Trading