Saturday, November 03, 2007

Using Stock Market Volume to Predict Stock Market Volatility

I recently posted on the topic of VIX as a predictor of daily volatility in the S&P 500 market and the role of volatility in money management. This next post in my series dealing with intraday market movement will explore volume as a predictor of daily volatility in SPY.

Relative volume is a cornerstone concept in my trading. Relative volume simply refers to whether today's volume is ahead or behind the average volume for a given lookback period at a given time of day. Thus, if we are trading with X volume at 10:30 AM, I want to know how X volume compares to the average volume at 10:30 AM for the past 20 trading sessions.

There are two reasons for wanting to know relative volume through the day:

1) As we'll see below, volume correlates with volatility. Knowing if volume is ahead or behind the 20-day average helps me estimate the likely size of the daily range. It also helps me estimate how much I can expect to take out of a given trade, the size I should trade, and where I should place my stops (see the prior post).

2) The lion's share of volume in the S&P market is attributable to large (professional) traders. When we see volume ahead of the 20-day average, it suggests that there is good institutional participation in the move we are seeing. If volume is below the 20-day average, it suggests that locals (market makers) are dominating the price action. Since my trading is predicated on following the footsteps of the large traders, I stop trading when relative volume drops significantly. There just isn't enough volatility and institutional participation to sustain meaningful moves, and the choppy action of the locals does not lend itself to short-term position trading.

So let's look at relative volume (how the current day's volume compares to the median volume for the prior 20 trading sessions) and relative range (how the current day's range compares to the median high-low range for the previous 20 sessions). Going back to 1996 (N = 2934 trading days), we find that the correlation between daily relative volume and daily relative range is a very high .66. About 40% of all variance in the day's relative range can be explained by the day's relative volume.

In practice, that means that I'm regularly updating volume during the day and comparing it to what the median volume has been for that same time of day over the past 20 trading sessions. This updating of relative volume through the day enables me to see if we are building participation or losing it; if we're likely to have above average volatility or below average for the day. Very often I will identify relative volume for the first 15 minutes of trading and come up with a reasonable estimate of what that day's volatility (size of range) is likely to look like.

There are also important serial correlations in relative volume. Yesterday's relative volume correlates with today's relative volume by an impressive .42. That means that we can estimate today's participation of institutional traders simply on the basis of yesterday's participation. Moreover, yesterday's relative range correlates with today's range by .26--not huge, but significant. Busy, volatile days tend to be followed by busy, volatile days.

Interestingly, overnight volatility is correlated with the volatility of the next trading session in the S&P 500 market (SPY). The absolute size of the market's opening gap (difference between the open and the prior day's close) correlates with the size of the range for the next trading session by .35. Thus, when we see above average volume for the previous day and an above average size opening gap for the current day, we are justified in expecting a busier, more volatile trading session ahead.

Let's say that an intraday trader is seeking moves of X points. Volatility defines how many of those trading moves will be likely to occur in a given trading session. In that sense, volatility helps define raw, gross opportunity. Volatility also helps define risk, the market's likely speed of movement (time it takes to make a move of X), and the likely amplitude of random movement around one's position (which affects stop placements).

It's important to a trader to know which direction a market is likely to move in, but it's also important to know how much a market is going to move. As we will see in upcoming posts, knowing how much a market is likely to move is a valuable aid in handicapping the odds of that market hitting various target price levels.


Five Guiding Principles of Short-Term Trading

Finding Opportunity in the Market