Thursday, February 08, 2007

How You Can Track The Stock Market's Large Traders

I have a backlog of emails and comments from the recent poll concerning integrating system and discretionary trading. Many thanks to readers for their interest. I'll look for further comments today and then will summarize them--and my reactions--in a post tomorrow.

As a gesture of appreciation, allow me to share a piece of my research in this and the next post. In the past, I have only shared this with specific prop traders I have worked with in a coaching vein. The idea of the research is to identify two (related) things that every short-term trader should know:

1) When are institutional participants active in the market?
2) When are we likely to have large price moves during morning trade?

Let's start with the basics. As a gross distinction, imagine that we have two kinds of traders in the marketplace. The first are market makers (locals) who provide liquidity. They are in the markets throughout the day, and they are in the markets every day. Having worked with prop traders who function as liquidity providers in the electronic futures markets, I can tell you that their participation in the market is relatively consistent from day to day. For simplicity's sake, I treat their involvement (volume) in the market as a constant.

The second group of traders are directional traders who enter the market when they perceive that we are trading away from what they deem as value. They may fade highs and lows, identifying value as somewhere between, or they may trade breakouts, placing value away from recent trading ranges. They generally enter the market on longer-term bases than the locals (scalpers). While directional traders include small individual traders, their volume is dominated by CTAs, hedge funds, mutual funds, and other large, institutional traders.

If we make the assumption that the participation of locals is relatively constant from day to day, then we can attribute surges or plunges in volume relative to average to the increased or decreased participation of large, institutional traders. When markets are trading with below average volume, they are dominated by liquidity providers. Because those traders generally work orders above and below the current market price, they are selling offers and buying bids. That tends to make them short when markets rise and long when markets fall. They profit, on average, when moves do not follow through and trend. This is why local-dominated markets (i.e., low volume markets) tend to trade in narrow, choppy ranges. There just aren't enough large, directional traders participating to move the market far from current levels of value.

Conversely, when volume is significantly above average, we have active participation of the large institutional traders. They are perceiving that value is away from where the market is currently trading, and they exert a directional pressure on the market as they lift offers or hit bids. When markets get very active, you'll often notice that the average quantity of ES contracts at each level in the order book (your depth of market display) often drops. This is because locals are pulling in their horns. They don't want to get run over by the large institutions, and they--on average--lack the firepower to hold the market to ranges where they would benefit from selling offers and buying bids. With locals pulling back, directionally biased institutions create short-term trends.

This, then, is why volume is so closely correlated with volatility. The presence of large institutional traders is what makes for trends. Knowing how volume compares to average tells you a great deal about *who* is in the market and *how* the market is likely to move. When volume is very low, there is little market movement. At those times, there may not be enough opportunity to justify the slippage and commissions of trading. In short, volume = volatility = opportunity. Knowing volume is knowing how much markets are likely to move, because you're tracking who is in the market.

Now for the details. You may want to print this out and keep it by your side when trading. That's what I do.

Going back to the past 105 trading sessions, the median morning trading volume (8:30 AM CT - 11:00 AM CT) in the ES contract is 442,369 contracts. The median trading range (high - low range) for morning trade is .47%. That is a little less than 7 ES points.

The correlation between morning volume and size of the morning trading range has been .73. That's quite high. If we divide our sample into quartiles based upon volume, the highest volume group of days (with a median 569,000 contracts) average a trading range of .71% (about 10 ES points). The next highest volume group of days (median of 485,000 contracts) averages a trading range of .51% (about 7 ES points). The third, next-to-lowest volume group (median of 407,000 contracts) averages a trading range of .38% (about 5.5 ES points). The lowest volume group (median of 280,000 contracts) averages a trading range of .35% (about 5 ES points). In other words, you get twice as much movement (range) in morning trading when you compare the highest volume days to the lowest volume ones.

Think: Can you see how this information helps you set profit targets for morning trading? Can you see why, yesterday, I took profits once the morning market had moved about 6 points from low to high?

Tracking volume *is* tracking the stock market's largest traders. When you see volume expand significantly and when you see that the volume is asymetrically distributed at the bid or offer, you know that large market participants with a directional bias are taking over the show. That is why, yesterday, I alerted readers to a pending breakout move. I'm not clairvoyant; far from it. But I could see that large traders were leaning one way and locals would not be able to fade that in the short run.

When there is low volume, there is relative consensus about market value; when there is high volume, there is uncertainty. Uncertainty is what moves markets and facilitates future trade. As we will see in my next post, the current period's volume and volatility is positively correlated with those in the next period. Knowing participation *now* informs us about the near-term future.

But how can we track volume within the morning to determine if volume is waning or picking up? I'll provide some numbers in my next post that you might want to print out.