For example, once the Philadelphia Semiconductor Index (SOX) hit 430-ish in mid-February, I thought it had gone far enough. It then came down to tag the uptrend line and went on to make a higher high, and it's finally broken the uptrend line -- as you can see -- and has a potential head-and-shoulders top in place. If you review many of the individual stock charts in the semiconductor group, I am sure you'll see how many failed to make higher highs within the SOX in mid-March.
The number of stocks making new highs does matter. Over time, a decline in this statistic means lower highs, and that points to the formation of tops.