No. of Recommendations: 42
Almost seven years ago SIPro introduced an F-score field in their database. It is based on the work of Joseph Piotroski of the University of Chicago. He found that while value stocks outperformed growth stocks a few outperformers skewed the portfolio returns. A majority of value stocks underperformed the market.

Piotroski, an unusally practical sort for an academic, put together a financial scoring system to weed out the dogs that had a larger chance of underperforming, being delisted, going bankrupt, etc. High scoring companies in general are in better financial health.

Up until now I have had little success in using the F-score in an MI screen. However, the key appears to be applying it to high-debt companies. Indeed, with this group one can produce both effective long and short screens. Not to mention a hedge with the highest CAGR that I've seen for a hedge.

The high screen:
Price > 10
Volume > 20K/d
Total liabilities Q1 > 0.75 x Total Assets Q1
F-score (12mon) > 5
Sort high 26wk

The low screen:
Price > 10
Volume > 20K/d
Total liabilities Q1 > 0.75 x Total Assets Q1
F-score (12mon) < 4
Sort low 26wk

Here are the results since 3/02 for a five-stock portfolio with a three-month hold:

High Low Low as short
CAGR 24 -39 33
GSD 39 64 45

These are quite respectable CAGRs for a quarterly hold, especially since over the same period SPY would have given you -1/13.

However, the GSD numbers are too high for comfort. The volatility can be cut in half if we hedge the High screen with the Low (as a short).

High hedged with Low
CAGR 34%
GSD 20

If one doesn't want to go to the trouble of shorting individual stocks, the hedge could be done with an ETF that is an inverse of SPY. Here the results are lower but with a much smoother ride.

High hedged with SPY (short)
CAGR 15%
GSD 13
RRR3 0.92

So, what has it done for you lately? Here are the 2008 results for the High screen and the Low as a short:

Start High Low Hedge
08-01 - 5% 26% 10.5%
08-02 2 39 20.5
08-03 13 25 19
08-04 29 5 17
08-05 23 30 26.5
08-06 1 36 18.5
08-07 -32 48 8
08-08 -39 45 3

DB2
Print the post Back To Top
No. of Recommendations: 0
Seems like a narrow spread for the F-score.

Also, how does it do without the momentum component?

D
Print the post Back To Top
No. of Recommendations: 0
Impressive. I've been looking for a good long/short strategy that was derived from the same underlying concept. I've never found one as good as this. Could you share a job #?
Print the post Back To Top
No. of Recommendations: 0
Seems like a narrow spread for the F-score.

Piotroski's original work just divided down the middle (0-4 and 5-9). I just added an excluded middle.

Also, how does it do without the momentum component?

I haven't tried other final sorts. Any suggestions?

DB2
Print the post Back To Top
No. of Recommendations: 0
Impressive. I've been looking for a good long/short strategy that was derived from the same underlying concept. I've never found one as good as this.

It certainly has the highest CAGR for any hedge I've worked with. Going along with that, the volatility is a bit on the high side.

On the other hand, the worst 4-quarter rolling return was 0% (starting in 10/02 -- everything levitated in 2003). I could live with that.

Could you share a job #?

I don't have a job number as I used my own data disks. If you run the screens at the Keelix backtester, let us know if the results are similar.

DB2
Print the post Back To Top
No. of Recommendations: 0
Also, how does it do without the momentum component?

I haven't tried other final sorts. Any suggestions?
>>
My favorite is always P/FCF.

Nice screen, just I like to strip momentum from value to see how it does without.

D
Print the post Back To Top
Advertisement