Thursday, September 28, 2006

Does It Pay to Buy Multiple Market Highs?


Amidst the attention being given to the new highs in the large cap indices, it's worth asking the question, "Does it pay to buy new price highs?"

In the chart above, I'm taking closing daily prices of SPY and simply calculating the number of days in the past 20 in which we've made new 20-day highs. I subtract from that figure the number of days in the past 20 in which we've made new 20-day lows.

Going back to 2004 (N = 670 trading days), we've had 205 occasions in which the net number of new high days has been five or greater. Ten days later, SPY has been down by an average of -.06% (106 up, 99 down). That is considerably weaker than the average ten-day gain of .38% (282 up, 183 down) for the remainder of the sample.

Note that, at present, we've had a net number of new high days of 8. When that has occurred since 2004 (N = 65), returns in SPY have been similarly subnormal 5-10 days out (31 up, 34 down).

Conversely, when the net number of new low days in SPY is five or greater (N = 55), the next ten days in SPY average a gain of 1.03% (41 up, 14 down)--quite a positive edge.

Quite simply, it's made us money since 2004 to buy the market when SPY has been making multiple new 20-day lows. It has actually cost short-term traders money to buy the market when SPY has been making multiple new 20-day highs.

What that's really saying is that we haven't been in a trending market over this time frame. A trending market is one in which buying highs or selling lows is significantly more profitable than the reverse. There were times during the 2003-2004 runup in which buying multiple new high occasions was profitable. That hasn't been the case, however, for going on three years and-- given divergences in the market among sectors (small/mid caps, NASDAQ)--I'm not inclined to pronounce that "this time will be different".

Note in today's Weblog, however, that a peak in multiple new high days tends to precede short-term market price peaks. Given that we just hit a peak of 8 on Wednesday, it is also reasonable to conjecture that we have not yet put in a top for this particular market move. It's when we've seen price highs at progressively lower net numbers of new high days that, since 2004, the market has tended to correct significantly.