Tuesday, January 30, 2007

When Markets Are Flat, Look At What's Moving

Yesterday, the S&P 500 Index (SPY) closed very near where it opened the day's session. Such flat days represent a kind of short-term trading range. During the day's activity, we moved both above and below the open, but could not sustain either extreme. The question for the short-term trader thus becomes: What are the odds of breaking out of this short-term range with an upside vs. downside directional move? In the case of a day such as today, when we're beginning a Fed meeting and investors will be waiting for the Wednesday announcement, such a test of either the previous day's high or low may be all the directional activity we'll get.

There are two principles I follow when analyzing flat market periods: 1) Identify which sectors of the market *are* moving during the flat performance and then break down the sample of historical flat periods by this sector performance to see if there is a next-day directional edge; and 2) Identify the longer time-frame context in which the flat period is occurring and see if flat periods with a similar context (i.e., whether the flat period is occurring in a rising or falling market on a longer-term basis) display a directional edge. In this post, let's look at analyzing flat periods on the basis of what's moving among sectors. In my next blog entry, we'll examine the role of context.

Monday's action was noteworthy because we had a solid gain in the Russell 2000 Index (IWM) during the flat day's performance in SPY. So let's go back to 2004, identify all flat days in SPY, and then see if the relative performance of IWM makes a difference for next day performance.

What we find is that, since 2004 (N = 772 trading days), we've had 112 daily trading sessions in which, open to close, SPY has been up by less than .20% but not down by more than -.20%. Interestingly, the overall correlation between SPY and IWM daily price change in percentage terms from open to close, is a very high .87. But when we look at the 112 flat days in SPY, the correlation with IWM is only .11. This relative absence of correlation enables us to meaningfully break down the flat days based on IWM performance.

Overall, when SPY has been flat (N = 112), the next day in SPY has averaged a gain of .05% (58 up, 54 down). No real directional edge there. Now, however, let's break down the sample based on the action in IWM.

When SPY has been flat on the day and IWM has been strong (i.e., rising more than .40%, as we had on Monday; N = 17), the next day in SPY averages a gain of .27% (13 up, 4 down). Fully 14 of the 17 days trade above the highs from the flat day; only 7 of the 17 days break the prior day's low. When SPY has been flat on the day and IWM has been weak (i.e., falling more than -.40%; N = 30), the next day in SPY has averaged a gain of only .03% (15 up, 14 down, 1 unchanged). Only 19 of those 30 occasions have ended up trading above the high from the flat day; 13 of the 30 have broken the flat day's low price.

What this suggests is that when small cap stocks have been strong during a flat market day, there is underlying strength in the market. This buying interest tends to carry forward the next day, making it reasonable to look toward the prior day's high as a potential price target for intraday trading. Note that this hypothesis can be tested out by looking at other measures of market strength during flat periods, such as performance from other sectors (e.g., QQQQ) and indicator levels (e.g., advancing vs. declining stocks). Such validation is important when sample sizes are small.

As always, I don't take these as mechanical trading signals. Rather they are sources of directional hypotheses regarding near-term market action. If I see early selling drying up in ES when price is holding above the low from the flat day, I'm likely to be a buyer in ES. Conversely, if I see large traders hitting bids in ES on the heels of a spike in oil prices or interest rates, I will not blindly follow the hypothesis. Indeed, I might conclude that the market is not living up to historical norms and factor *that* into my decision making.

The beauty of this kind of historical market research is that it supplements just about any trading style. You may, for example, utilize CCI patterns (like Woodie's traders) or chart formations (such as gaps) for your market setups. Because most stocks are highly correlated with the movement of the S&P 500 Index, knowing the odds of a rising or falling market can help you decide whether or not to take a particular setup, and it can help you decide how much size you want to devote to that setup. My personal experience is that, when several different historical patterns are pointing in the same direction *and* I see market sentiment (NYSE TICK, volume at bid vs offer) pointing in that same direction, that is a very high probability trade and I want to use my maximum size to take advantage of that. Such trades--even just once a week--can account for a large portion of your overall profitability.