Saturday, October 21, 2006

When New Highs Get Higher, Should You Be a Buyer?

And when new highs go into the cellar, should you be a seller?

In a recent post, I took a look at what I called the relative new high ratio and found an interesting pattern of price changes following extremes in the proportion of stocks making new highs relative to their moving average. The Trading Psychology Weblog tracks fresh 20-day new highs and new lows across the major exchanges. Might we learn something about momentum and short-term trading patterns from relative 20 day new highs?

For this investigation, I simply took the number of new 20-day highs and lows and subtracted from them the 20-day moving average of 20-day new highs and lows. This gives us a reading of when those new highs and lows are expanding and contracting relative to the past 20 trading sessions. The data go back to the beginning of 2004 (N = 697 trading days).

Interestingly, when there is a strong expansion of new highs, with the number of relative new highs exceeding +750 (N = 43), the next five days in the S&P 500 Index (SPY) average a gain of .13% (25 up, 18 down). This is virtually the same as the average five-day gain for the entire sample (.14%; 387 up, 310 down). In other words, when new highs expand sharply, there is no bearish edge whatsoever. Stocks continue rising at their normal rate.

When the number of new highs expand, but less vigorously (N = 270), the next five days in SPY average no gain at all (144 up, 126 down)--a subnormal return from average. This finding fits with my other research suggesting that rises on strong momentum tend to persist, whereas rises on modest momentum yield no bullish edge.

Now let's take a look at sharp contractions in the number of new 20-day highs. When these have exceeded -600 (i.e., when the number of current new highs falls short of the moving average of new highs by 600 or more; N = 44), the next five days in SPY have averaged a gain of .59% (32 up, 12 down). That is considerably stronger than the average five-day performance since 2004. In other words, when stocks retreat en masse from their new highs, that's been a good time to buy. This also fits with recent research that finds that great times to buy are when few stocks are attracting favor.

Quite a bit of evidence, across different time frames, converges to suggest that it is the trajectory of a market rise--its momentum--that accounts for whether that rise is likely to persist in the near term or not. On Monday, we saw a solid expansion of relative new 20-day highs. As long as that expansion persists, we can't legitimately conclude that this is an overbought market with a high probability of decline.

So to summarize: When new highs get much higher, it's OK to be a buyer. But when new highs go into the cellar, you don't want to be a seller.

My next post will take a similar look at the new low data and see if any patterns emerge from the expansion and contraction of new lows.