Wednesday, April 04, 2007

A Follow-Up On New Highs And New Lows In The Stock Market

My recent post noted that, following a surge in the number of stocks making new five-day closing highs, we tend to see follow through strength over a three-session period prior to subsequent reversal at a five-day period. An alert reader wrote and asked if the same pattern applies to surges in the number of stocks making fresh five-day closing lows.

The answer is yes, but the time frames are different.

Recall that we're looking at a portfolio of 40 stocks evenly divided among and highly weighted within eight S&P 500 sectors. Going back to 2004 (N = 813 trading days), we've had 38 occasions in which the net number of stocks making new lows over new highs reached 25 or greater. We'll call that the "weak day".

The following day, the S&P 500 Index (SPY) broke below the low of the weak day on 26 of the 38 occasions, for an average drop of -.28%. Across the entire sample, the low of the current day actually exceeds the low of the prior day by an average of .03% (443 up, 370 down). That tells us that, in the short run, broad weakness tends to spill over into the next trading session.

Interestingly, however, if we look on a close-to-close basis, we find that the day after the weak day, SPY averages a *gain* of .12% (22 up, 16 down). That is stronger than the average one-day change for the entire sample of .03% (444 up, 369 down). This tells us that follow-through weakness the day after a surge of new lows tends to reverse by the close of trade.

If we look five days out, we can see strong evidence of a reversal effect, as noted in a prior post. Following the weak day, the market is higher on 30 out of 38 occasions, for a solid average gain of .74%. By contrast, the average five-day gain for the entire sample is .16% (462 up, 351 down).

Once again we see a pattern of follow-through weakness followed by reversal. Notice with the days of surges in new lows, however, that the subsequent weakness tends to evaporate more quickly and the reversal effect is more pronounced. This is an important distinction between strong rising and falling markets that has important implications for trade planning.


Ryan said...


Be interesting to know if these reversal patterns are inverted when framed in the context of a primary bear rather than primary bull market.

Paul said...

Enjoy reading your blog - very good commentary & insights. A comment on your reversal pattern analysis.

Your period of anaylsis from 2004 co-insides with what is clearly a bull market that started c. early-mid 2003.

Would your conclusion (final paragraph) be different
if your sample trade data were based a period covering a bear market?

So I would be interested in your analysis of the 3 & 5 day reversal patterns during a bear market (say the period Q1 2001 through Q3 2003).

This would highlight any difference in 'behaviour' of the reversal patterns given the different market environments and perhaps a different conclusion?

Brett Steenbarger, Ph.D. said...

Hi Ryan and Paul,

Thanks for the great idea regarding a follow-up study. One difference between my work and that of many system developers is that I'm not trying to identify patterns that work in all markets. Rather, I look at the most recent market and find the regimes that are operative, with the assumption that these will carry forward into the immediate future.