|
PureBytes Links
Trading Reference Links
|
Thad:
>I am looking to do some back testing on James O'Shaughnessy's
>Reasonable Runaways long term stock strategy which is covered in his
>recent book "How to Retire Rich".
>The stock selection criteria are as follows:
>1. Market Cap > $150 million
>2. Price - Sales Ratio < 1
>3. Buy the 25 to 50 stocks from 1 and 2 above that have the best one
>year price appreciation.
>4. Rerun the selection process once a year.
As mentioned, at least some of O'Shaughnessy's funds have performed very
poorly, and with large DD's. But let's put that aside, on the assumption that
you have a strong enough stomach to embrace the long-term perspective that he
espouses.
IMNSHO, there is a flaw in the system you describe. Choosing low PSR
companies
will skew you toward low profit margin industries. After all, an industry
that
generally gets huge revenues and relatively small earnings, due to the very
nature of the business (supermarkets come to mind), will have low PSR ratios
simply because the R is so large. This doesn't necessarily mean that the P is
a bargain, which is the underlying intention of O'Shaughnessy's use of a low
PSR system.
O'Shaughnessy himself, in the recent TASC interview, has acknowledged that low
PM in itself is not a profitable strategy. In fact, the results are pretty
much random. Therefore, he is shooting himself in the foot by skewing the
above system to low PM companies, thereby adding a random filter and
inadvertently crowding out some of the better low PSR companies.
I have made this point on Compuserve, where one of the engineering types did a
study and confirmed that there was indeed a strong inadvertent bias toward low
PM using O'Shaughnessy's criteria.
The fix for this flaw is to replace "low company PSR" with "(low company
PSR)/(industry PSR)", thereby normalizing the criteria so that you invest in
only the exceptional PSR stars, regardless of industry. I do not have the
capability of backtesting the results, but this is (at least intuitively) the
correct formulation.
O'Shaughnessy has refused to respond to this point, even though he has the
capability to easily backtest this. He followed and participated in the
discussion on Compuserve until this point was made, and then grew silent. He
would not respond to direct emails. I again made the same point as a response
to his TASC interview, and TASC confirmed that they had passed the msg to him,
but again he refused to address it. Unfortunately, TASC did not gently insist
on a reply.
I like O'Shaughnessy and own a couple of his books. I think he's a very very
smart guy. But it concerns me that he seems less than eager to embrace
corrections and improvements, or at least respond if he doesn't agree.
Personally, I am always eager to have someone point out that I may have made a
mistake in a system design. It's happened to all of us, and I'd rather be
momentarily humiliated than permanently poorer.
For all these reasons, I am now very reluctant to have O'Saughnessy manage my
money. This is of course, just my personal opinion. And for the record, I
have nothing against Jim personally. I don't even know him, other than
exchanging a couple of emails with him long ago when I tried to do what you
want to do, and he was very nice about explaining his data sources to me:
S&P
Comstock, as I recall. S&P told me that it's $12,000 a year (I believe...but
it was a while back), and S&P won't even give you the full database they gave
him, they reserve it for their consulting practice.
But if you want to and are able to replicate his work, I suggest the preceding
correction to it.
Regards,
Paul
|