Thursday, February 12, 2009

Historical Patterns in Markets and Trading Hypotheses

Yesterday morning, I posted the following via Twitter:

7:24 AM CT- When Supply > 200 since late 2002, next day in SP averages gain of .69% (14 up, 7 down), but edge is bearish 4 days out.

Regular readers of the blog are aware that Demand and Supply represent indexes of the number of stocks that close above and below the volatility bands surrounding their short-term moving averages. A reading greater than 200 for Supply is unusually weak; only 21 previous days since late 2002 saw a similar proportion of stocks closing below their moving average envelopes.

Interestingly, the pattern showed that the next day following such weakness tends to produce a bounce, though the edge turns bearish thereafter. This pattern made sense to me: after a big selloff day, short-term participants are leaning bearish and have to cover on any market firmness. This contributes to the firmness and leads to a market bounce. The greater trend, however, is bearish, as the bounce is usually modest relative to the magnitude of the selloff that produced high Supply. Thus, the trend tends to continue after the bounce day.

What does knowing such a historical pattern accomplish? At the very least, it can serve as a heads up to prevent traders from blindly assuming that a very weak day will necessarily spill over into weakness the following day. Beyond that, the historical pattern can suggest worthwhile trade ideas. Seeing that the ES was holding overnight above its previous day's low, buying on weakness for a break above the overnight high was a reasonable short-term trade idea.

In my new book, which will be out next month, I devote an entire chapter to historical patterns in markets and how to identify them with simple tools: a data feed and Excel. The key point I make in the chapter is that such pattern identification represents qualitative, hypothesis-generating research, not necessarily hypothesis testing research. In other words, the historical patterns should be treated as market hypotheses that we test out with our trades; not as established, fixed conclusions that we trade mechanically, heedless of present price and volume action.

If you knew, for instance, that 2/3 of all warming days after a cold spell tended to be rainy days, you'd think seriously about bringing an umbrella to work. That information would be a heads up to prepare you for what might lie ahead. Similarly, historical patterns in markets provide possible road maps that reflect where markets have headed under conditions similar to the ones we're observing today. The key to utilizing this information is to treat it as a meaningful hypothesis, not as an opinion to marry. Indeed, if markets start behaving contrary to this road map, that too is important information: situational factors are at play that are leading the market to deviate from its historical expectations.

I will continue to offer some historical patterns via Twitter when they make particular sense to me. (Free subscription via RSS). If you're interested in generating more market hypotheses, other sites that conduct worthwhile historical research include Market Tells, SentimenTrader, Quantifiable Edges, Market Rewind, VIX and More, BZB Trader, Market Sci, and Ripe Trade.