Saturday, December 22, 2007

Returns Following Surges in New Lows

In the last post, we looked at returns in the S&P 500 Index following surges in the number of stocks making short-term highs. Now we'll look at surges in the number of stocks making new short-term lows and see if those affect returns over 5 and 20-day horizons.

Going back to 2004 (N = 981 trading days), we have 55 occasions in which the net number of new five-day lows minus new highs among the 40 stocks in my basket has been 25 or greater. Five days later, the S&P 500 Index (SPY) is up by an average of .86% (42 up, 13 down)--much stronger than the average five-day gain of .10% (513 up, 413 down).

When we look 20 days out, we see continued outperformance following surges in the number of stocks making new five-day lows. Twenty days later, the average gain in SPY is a very healthy 2.04% (45 up, 10 down), much stronger than the average 20-day gain of .48% (581 up, 345 down) for the remainder of the sample.

This makes conceptual sense. When traders panic and sell stocks indiscriminately across sectors, this marks a period of maximum bearishness. With no more bears available to sell, short-covering aids value-oriented buyers and we see favorable prospective returns.

Putting together the findings from this post and the last one, we can see that, on a 20-day horizon, returns have been best when a plurality of stocks are simultaneously making new highs or new lows. When they are making fresh highs, momentum tends to carry them further before there is a major correction. When they're making fresh lows, reversal effects occur more quickly--at five and 20-day horizons.

Understanding these patterns helps short-term traders make decisions regarding trading vs. fading market strength and weakness.

RELATED POSTS:

Falling Markets and New Lows

When New Lows are High
.

5 comments:

Ziad said...

I'm interested to know if the overall market context has an affect on these findings. i.e. The findings were that broad rises tend to continue in the short-term and broad declines tend to reverse. But is that necessarily a inherent difference in fear vs. greed, or could it be the fact that since 2004 we have been in an overall bull market where the bias is naturally up (thus causing a bullish move after rises and falls)? One way to answer this would be to see if the results are the same for a period like 2000-2002. If during that period we see opposite results, it means that its the overall trend that affects short-term continuation patterns following broad moves; if the results are the same it implies that fear and greed show distinctly different behavior in the stock market.

MidKnight said...

Hi Brett,

I like this sort of analysis you do here. A couple of questions about this post and your prior one if I may.

1) In the prior post you compared the 5 day new highs against the average 5 day change. But in this post you compared the 5 day new lows against the average 5 day gain. I was curious why you decided to compare them against a different benchmark?

2) In both this post and the prior post you are comparing the 20 day high/low against the average 20 day gains, yet the percent gain is different. Why is that?

Thanks in advance for your reply. I wish you and those dear to you a tremendous holiday season.

My very best regards,
MK

Brett Steenbarger, Ph.D. said...

Hi Ziad,

Yes, patterns of returns do differ across bull and bear markets, so the selection of lookback periods is important.

Brett

Brett Steenbarger, Ph.D. said...

Hi Midknight,

The problem is in my wording; in both cases I was comparing the new highs/lows to the next five-day returns. My comparisons were between the surge days for new highs and the remainder of the sample and the surge days for new lows and the remainder of the sample. Two different remainders, hence different returns for comparison purposes.

Brett

Brett Steenbarger, Ph.D. said...

Hi Ziad,

Yes, patterns of returns do differ across bull and bear markets, so the selection of lookback periods is important.

Brett