I've been wanting to determine whether the CAPS rating is really an effective determinant of performance. I have been gathering together market price data and collecting a database of CAPS ratings for statistical studies. I did a crude study of stock price change YTD vs CAPS rating and found a nice correlation, but unfortunately an INVERSE CORRELATION. It's so perfectly inverse that I almost can't believe it, but I can't see anything wrong with my methods. Here is the average stock price percentage increase year to date vs CAPS rating:0 (no rating or not in database) - +17.06% - 3739 stocks1 star - +16.88% - 752 stocks2 stars - +7.98% - 926 stocks3 stars - +3.63% - 959 stocks4 stars - -0.04% - 984 stocks5 stars - -3.8% - 843 stocksThe study is based on the current CAPS rating, not the rating in effect at the start of the period, so this would affect results a bit.If anyone has any other similar or conflicting results, I'd love to hear about them.Chris
Neat stuff, Chris. One caveat, though: What you might be seeing here is people rating highly stocks that have underperformed to such an extent that they now look cheap... ...and rating lowly stocks that have reaped large, perhaps unwarranted, run-ups in price.TMFDitty
Wow neat findings. It would be interesting to delve further into the data in order to see what's driving it. The first thing that comes to mind would be to inspect the distributions of percent change of stocks, perhaps looking at separate distributions for each rating category. My hunch is that if you look only at the stocks with one star (or no stars), they'll have very skewed distributions, with many of them showing losses or no gain, and then a couple really extreme outliers that are driving the mean.If that's the case, then my guess is you're stumbling across the artifact of combining small or micro cap stocks (which may be more likely to have less or no stars) in comparison to medium and large-cap. That is to say, compared to other stocks, small caps can't go down much, but they artificially have a large upside. A couple solutions might be to: Run non-parametric tests; group the stocks by their cap-size, and rerun analyses for each group separately; control for the number of CAPS players who rated the stock; instead of CAPS rating, look at CAPS rating change. Along the lines of the latter, it might also be interesting analyze a stock's CAPS rating in the same way as you'd look at its stock price: Number of ratings being like volume; and quantifying volatility in rating as well.Anyway, cool stuff thanks.
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |