Thursday, September 27, 2012

The F-Score

At the risk of being burned at the stake, I am going to speak some heresies about the F-Score.  

First, statistics -- F-score, NCAV, magic formula, P/B, golden cross, whatever -- may be useful for stocks taken two handfuls at a time, but they are not, by themselves, useful for evaluating any individual stock.

Second, the F-score is a momentum indicator. It has only the most occasional and coincidental relationship with quality or value. 

Look at any of the indicators/criteria, and they are, with the exception of CFO>Net income, flimsy at best, nonsensical at worst: 

  • no net pension liabilities, 
  • no changes in net pension liabilities as a result in changes in the prices of securities prices, 
  • no leases, 
  • negative working capital is bad, 
  • excess cash is bad, 
  • dilution is always bad, etc. 

Terrible stuff. It is what Baudrillard, bless him, would call a third order simulacrum.

To my mind, the F-score relies on two things: (a) selling into hype; and (b) timing of cylicals. 

Nothing else. 

Compare it to the most well-known of momentum strategies and I think you'll see what I mean:

 Data source:

Where there is no danger of overexcited buyers -- Japan, for example -- it doesn't work.

Still, I get it. I see its appeal in the same way that I see the appeal of buying stocks whose 50 MA crosses the 200 MA, or whose relative price strength is greater than 1, or whatever else.

But, in my view, it is a mistake to use the F-score at the level of an individual stock just as it is incorrect to think of it as a proxy for quality. 


  1. This is true. The fscore has value, however, when you use what it was intended to do. I swear all these guys throwing it around have never read the academic paper that introduced it to the world. I bet you haven't either.

  2. It wasn't intended to DO anything. It was intended to DESCRIBE a statistical phenomenon.

  3. Hi red. Very thought provoking, and mostly I agree. However I still like the F_Score and since my reasoning took 750 words in the end I've posted it on my own blog: !

    I don't take up the quality issue there, the post was just getting too long. I believe the F_Score is a measure of quality as well as momentum. Deciding, which is the more significant is tricky. Seven of the nine factors measure momentum in that they measure change in some way but the two most predictive factors don't. They just ensure the firm is profitable in cash and accounting terms.

  4. Thanks, Richard.

    I'll do a follow-up.

  5. Red, it was definitely intended to DO something. Piotroski came up with it while looking for a way to pick cheap companies that were less likely to go bankrupt than other cheap companies. Yeah, it's a nice set of measurements anyways but they were mostly picked at random- the point wasn't "these 9 things when put together are the perfect secret sauce for investing"

    You'd know this if you'd actually read the guy's research. It's a great paper. I'd recommend it.


Note: Only a member of this blog may post a comment.