Contributions : 1c Stock screening and image
Proto-momentum model by Jan (Leif Ericssen), 05 March 2000
Jan tells us here about his proto-momentum model, a front-end screen to generate a 1st level list
of stocks to a behavioral analysis strategy.
I've been working on a stock selection approach/model based on a concept I call
The basic idea is to anticipate/ identify quality growth stocks that will gain "mind share" in
The approach is based on searching for stocks with factors that are potential trend drivers,
rather than selecting existing glamour stocks through momentum strategies such as CANSLIM.
My M.O. starts with firms with positive earnings and low/no debt, figuring that a financially
sound and profitable business is a safe and logical start for finding a superior firm.
From there I'm data mining comparative earnings records and standard estimates, looking
for significant patterns that indicate a likely chance (in terms of expected value, not odds-on)
of positive surprises.
I'm also looking at sales/ earnings/ equity growth rates, ROS and ROE to see if there are
any clues that can help in the filtration process.
Then I adjust the results for PEs and rank them.
My rationale is that PEs are both
* a fundamental reference measure (what's a $ of this firm's earnings going for?)
* and a psych/tech indicator (mkt view on future prospects and risks).
I look at relative strength (RS) figures calculated by IBD as a guide to mind share like
Arbitron or TV ratings.
My logic here is a simple trend following rationale: "nothing attracts a crowd like a crowd".
It seems easier to forecast reactions in a context of increasing perceptions than of
My approach favours RS against the S&P index.
The relationship between the stock's RS and it's industry's RS is noted, but not part of the
system's ratings yet.
And, I'm studying the stock's "risk-adjusted" price performance compared with baseline
sets in its industry, capitalisation level, and PE range as a factor.
Last of all is the stock's image (glamour potential), which is qualitative and where human
judgment comes in, and my model would leave it there, generating a list to select from.
the proto-momentum approach looks for stocks:
(a) currently reasonably liked in the market
(b) of financially healthy firms
(c) reasonably priced on trailing earnings
(d) with a fair potential for an earnings break outplay and/or upgraded earnings estimates.
In developing the proto-momentum concept and models, I appreciate the insights
that have been shared by several friends in m.i. and I would like to thank Peter
Greenfinch, whose research in stock imaging concepts gave me the idea of extrinsic
value for stocks.