I have also been baffled by high new highs numbers in so-so days. My observation is that they are 1-day delayed respect to a high in the index, (or a good day) and my working explanation is that they are caused by a day late reaction in those thinly traded stocks that trade 2 or 3 times a week.
No hard data to support it, but it's the only explanation I could come with... <g>