Does the Piotroski Score work for Large Caps too?
I got a question from a reader this week. He wanted to know whether the Piotroski score only works for Small Caps or not.
What does the table show us?
The values in the table for the columns MEAN and MEDIAN are correlation values. A 1 means that it’s perfectly positively correlated, i.e. high Piotroski score equals high return, -1 means low Piotroski score equals low return, and 0 means there is no correlation, i.e. Piotroski score gives you no information on returns.
As you can see, high Piotroski scores are more predictive of outperformance in small firms, and much less so in medium and large firms, but the Piotroski score still is not zero, which means it’s still useful to have it, even in larger firms.
In larger firms, it seems that low Piotroski scores are good at predicting bad returns, but high scores are not as good at predicting good returns. So with larger firms you can still use it to weed out bad eggs. The ‘High MINUS(-) Low’ row shows this: it indicates how big the difference is between highly and lowly scoring stocks. And that is still quite high, even in the Mid and Large Caps, even though it’s just half as strong as in Small Caps.
Conclusion: Based on Piotroski’s 2002 paper, the F-Score is indeed about twice as effective in finding good companies in the Small Cap area as in the Mid or Large Cap area. But it is by no means useless in the others, and especially helps weed out the bad eggs.
Don’t forget that the Piotroski score is first and foremost supposed to find good quality stocks in the group of cheap companies (i.e. with low high book to market / low price to book). So it’s not intended to be used with “glamour stocks”, which are usually part of the Large Caps. Based on this, you could even say that the Piotroski filter allows you to find:
– good quality small cap stocks (with high F-Score) that are cheap for no reason
– low quality large cap stocks (with low F-Score) that are expensive for no reason