Sunday, October 22, 2006

When New Lows Swell, Is It Time To Sell?

My last post examined relative new 20-day highs and found an interesting momentum-based pattern in the data from 2004 to the present. In this post, we will look at relative 20-day new lows and see if there might be similar tradable patterns.

As before, relative new lows reflect the number of stocks making fresh 20-day lows minus the average number of new 20-day lows over the prior 20 sessions. When the relative new lows are highly positive, it means that many more issues are making new lows than over the previous 20 sessions. A highly negative reading indicates that new lows have contracted greatly from their recent average.

Since 2004 (N = 697 trading days), we have had 52 occasions in which new 20-day lows have exceeded their 20-day average by 750 or more. Five days later, the S&P 500 Index (SPY) was up by an average of .52% (35 up, 17 down). That is much stronger than the average five-day gain of .14% (391 up, 306 down) for the entire sample.

In other words, when new lows have expanded sharply, returns have been bullish over the subsequent five trading sessions. When new lows swell, it’s not been time to sell.

How about when the number of new lows contracts greatly?

Since 2004, we’ve had 60 occasions in which new lows have fallen short of their average by 600 issues or more. Five days later, SPY is up on average by .19% (36 up, 24 down), modestly stronger than the average five-day performance. Interestingly, however, when we look 9 days out, SPY was up on average by .54% (43 up, 17 down). That is much stronger than the average 9-day change of .26% (408 up, 289 down) for the sample overall. In short, it has paid to buy and hold after new lows contract sharply.

These results point to two bullish edges: 1) a reversal effect after new lows expand greatly; and 2) a continuation effect after new lows sharply contract. When new lows have neither expanded nor contracted significantly (i.e., have been within 200 issues of their 20-day moving averages; N = 291), returns in SPY have been subnormal from 5-9 days out.

Once again, this points to strong momentum and lack of momentum as important variables in short-term market returns. The 20-day new high/new low data each day are available from the Trading Psychology Weblog.