The good thing about value investing is that if you can handle a long-term approach, then
the evidence is that you will do well over time. Buffett did the "statistically
significant" test in his talk "The Superinvestors of Graham and Doddsville" where he
simply looked back at all the people he worked with in his first job (for deep value
investor Ben Graham in the 50s) and they were all in the 80s worth $50mm from their
investing prowess.

A couple of points though. Value investing did horrible from 1997-1999 but then did
wonderfully in 2000-2001 when growth fell apart. You had to be able to withstand that pain
(plus the pain of your dot-com friends making fun of you).

Second, while Joel Greenblatt's approach seems to be a nice simply twist on the value
investing paradigm it should be noted that he doesn't seem to use that approach for his
own fund and I think, more accurately, the topics in his first book, "You be a Stock
Market Genius" probably better reflect his personal investing strategy.

That said, his approach is simple and appealing and seems, based on his tests, like a
solid way to follow value investing principles.

I'm more in line with the thinking that the magic formula approach will lose its magic now
that it is out there for everyone to see and use. I've seen this happen too many times
with other approaches. There are many different schools of thought on this but if you
have something that works, you can share it in general but giving out the secret sauce is
another thing. I like Greenblatt's approach and think it can be a good FILTER going
forward (at least for me).

The fact that Greenblatt uses only 2 variables is important. If I recall, the Schwab
Equity Ratings (at least when first introduced) used 24 variables. To me, that sounded
like data mining.

On his Columbia website, Greenblatt makes a presentation explaining why he thinks the
Magic Formula will continue to be useful. (The portion on the Magic Formula begins at
about the 1:04 mark.)

http://www3.gsb.columbia.edu/courses/selection/describe.cfm?WHATCOURSE=B9301-066&GSB=YES&T
erm=20063

Good pionts all around by jason t. One thing I might add with regards to "the cat being
out of the bag:" In value investing, if you can identifiy situations early, this could be
a good thing. The idea behind value investing is that something is undervalued by the
market now but eventually enough people will realize this that the stock will go up. If
you identify a stock by a formula, buy it, then tell more people about your formula the
process will be accelerated.

Also, my personal experience with any rule based investment system is that if it stops
working its more because of changing market dynamics then people figuring it out. For
example, a system like this would be undone by people favoring growth over value as they
did in the late 90s, as opposed to too many sophisticated readers figuring out whats up
thanks to Stockpickr.com

PART 2


This is tremendously important. As to his methodology, his
ranking method involves both the "earnings to price ratio " ie. earnings yield and the
firm's "return on capital". "EBIT" (earnings before interest and taxes) divided by
"enterprise value" (market cap debt preferred stock) is used to arrive at his
"earnings to price ratio" and for "return on capital" ("ROC") he uses EBIT/(Working
Capital Property, Plant, and Equipment). Because the Compustat database contains all
these items, the study can easily be duplicated. This condition is vital to all
scientific work, Greenblatt's study meets this criteria.

As to your concern about the cat being out of the beg, well, that's a much harder question
to address. My unscientific opinion tends to lean to the side that says Greenblatt's
transparency with his approach will result in lower returns for the strategy over time.
However this does not mean that the strategy will not continue to beat a passively managed
approach, such as being long the S&P 500.

Overall, Greenblatt is to be commended by a fine piece of scholarship in an arena plagued
by charlatans and unrealistic assumptions.

I think that statistically speaking, Greenblatt's approach is very sound. First he only
utilizes 2 degrees of freedom, this makes "curve fitting" much more difficult, second he
eliminates bias found in most fundamental-based research by pulling all data from the
Compustat Point-in-Time database.

As Greenblatt describes:"A newly released database from Standard and Poor's Compustat,
called 'Point in Time', was used. This database contains the exact information that was
available to Compustat customers on each date tested during the study period. The
database goes back 17 years, the time period selected for the magic formula study. By
using only this special database, it was possible to ensure that no look-ahead or
survivorship bias took place."

Test

Who here thinks that Joel Greenblatt's Magic Formula is just data mining, or that his
study has statistical problems that make his findings insignificant? Who here thinks that
Greenblatt's formula will have some predictive value going forward, and why? To me, the
approach makes sense, but I'm not sure it will work now that the cat's out of the bag.