|
PureBytes Links
Trading Reference Links
|
Jim,
You have asked for suggestions or comments on why we would believe that
what you are doing is wrong. So let me comment.
Your objective, as I understand it, is to reduce your stock universe
based on fundamental criteria in order to apply technical trading
techniques. Obviously, in doing so, you want to increase your technical
trading odds.
While I do not think that fundamentals are useful in short term (a few
days max) trading, I do agree that the combination of fundamental
criteria and technical indicators is a key to better medium to long term
trades. The problem is to gauge the fundamental criteria you want to
use. Market Capitalisation, Price-to-Earnings, Price-to-Book,
Price-to-CashFlow, Price-to-Sales, Dividends yields, Earnings-per-share,
Return-on-Equity have documented individual significance, but optimising
their combination, if at all possible, is not an easy matter.
It is a fact that all indicators - fundamental and technical - have
rather low predictive value. Their correlation with future stock prices
is in fact extremely low. In that case, it can easily be shown that
increasing the number of indicators will as a rule reduce and
effectively cancel the value of the prediction.
While you can reject a number of securities on the basis of fundamentals
and additional technical criteria such as high volume or price growth, I
cannot see any way to rank your selection depending on a multiplicity of
criteria, as suggested in your note. Perhaps it would be useful to
clearly separate the process of security rejection in a first step, and
then select the securities to include in your database on the basis of a
single preferred indicator in another.
Best regards,
André Scholberg
Jim Greening wrote:
>
> All,
> I'd like to start a discussion on how to narrow the universe of
> stocks to a few potential trading candidates that have certain
> characteristics that have resulted in successful trades in the past.
> I'll explain what I do and hope you chime in with comments on
> additional suggestions or comments on why you believe what I'm doing
> is wrong.
> First I believe that there are many different selection
> techniques that work and there are times when one technique will work
> better than others. For example, large cap momentum selection
> searches have worked exceptionally well for the last three years and
> small cap value techniques have not. However, if you look back in
> history, that's very unusual because small cap value techniques have
> worked better on average over the last fifty years. I also believe
> that the mirror image of a search for long candidates isn't
> necessarily a good search for short candidates, but I don't want to
> get into that today. I test the market indices to decide if my new
> positions should be long or short and my new positions have been long
> positions for some time now, but that is also the subject of a future
> post <G>. To keep this post fairly short, I want to concentrate on
> searches for long stock candidates only today.
> I use several different searches to select the few (less than
> 200) stocks that I keep in my MetaStock database. My searches are
> only to find stocks to add to my database and I only add a few each
> week. (I also prune my MetaStock database periodically.) I look at
> the results of the searches and use my 20/20 eyeballs for my final
> decision. I used to use Telescan for this by recently dropped them
> due to cost and bad data. I'm now using QP2 for my daily and
> historical data. It does have a built in search capability using a
> combination of fundamental and technical data. However, I'm just
> learning its programming language and it doesn't have a lot of
> criteria to search on. Still, I believe it has great potential and
> understand that they may add additional fundamental data after the
> first of the year. In the mean time I'm using Microsoft Investor and
> eSchwab search capabilities.
> What I'd like to do here, is not give the search formulas, but
> rather discuss the criteria I use for the various searches. I'll be
> glad to share the QP2 searches after I finish them. I've found that
> for fundamental criteria, the price/sales ratio by itself works best
> for me. I can enhance it's performance by also using debt/equity, and
> various earning and revenue growth criteria. Today I'd like to
> describe a momentum breakout search CANSLIM that I use for both large
> and small cap stocks; a bottom fishing search BotFish; a value growth
> search,ValuGro that can be used for both large and small cap stocks;
> and a small cap search, SmalCap. You will see my favorite criteria
> used in most of them.
> CANSLIM is a momentum search that I've been using for a long time
> that is based on parts of the CANSLIM strategy. What it looks for is
> stocks that are trading at a new 52 week high on high volume and meets
> several other conditions. In all my searches I begin by limiting the
> stock to a price above a minimum and below a maximum. For CANSLIM
> that's 5 and 144. Then I ask the search to list the 52 week high and
> the 52 week low for my info on making a final decision on adding the
> stock or not. The search continues by limiting the selection to those
> stocks which are in the top 20% considering earnings per share growth
> and the top 20% considering price growth over the last year. Then I
> ask for group rank, fundamental rank, 30 day average Volume, and 3/30
> day Volume to be as high as possible. This doesn't limit the
> selection, but helps rank the selection and I only take the top 20. I
> also ask for analyst rank to be as good as possible, institutional
> holding to be as low as possible, market cap be as low as possible,
> and insider trading to be as high as possible (buying). Finally I ask
> that the stock has set a new 52 week high within the last 5 days.
> BotFish is a search for stocks that have been beaten down and
> look like they are bottoming. It looks for stocks that are trading at
> 50% or more below their 52 week high but still have decent
> fundamentals and are showing signs of recovering. I again start by
> asking for a price between 5 and 21. I also ask it to list the 52
> week high and low. Then I ask that yesterday's close be lower than
> 1/2 the 52 week high. That makes sure that we are looking at only
> beaten down stocks. To insure good value I ask that the Price/sale
> ratio be less than 1 although I sometimes raise that to 2.1. I may
> also use the same criteria for price/book if that's available. I
> further ask that the price/sales be as low as possible. I may or may
> not use a debt/equity below 1 requirement. I then ask that the annual
> revenue growth be greater than 13% and that it also be as high as
> possible. I may or may not use the same criteria on earnings growth
> depending on how much that limits my results. Finally I ask that the
> 3 month relative strength be as high as possible to show early signs
> of turning up.
> ValuGro looks for stocks that are growing strongly yet are still
> good values as shown by low price/sales ratios. It starts off asking
> for a close greater than 3 but then asks for a price as low as
> possible. I also ask it to list the 52 week high and low. I then ask
> for a price to sales ratio that is well below the industry average.
> This way I get the benefit of a relative low price/sales but don't
> eliminate stocks in groups that have higher price/sales ratios. I
> then ask that the price/sales be as low as possible. I then ask that
> the debt/equity be below 1. I ask for a market cap above 250,000,000
> to eliminate micro caps. I eliminate stocks whose price dropped in
> the last quarter. I then ask that the quarterly price change, the
> annual revenue and earnings growth be as high as possible.
> SmalCap looks for small cap stocks that are showing exceptional
> earnings growth but are undervalued. It starts by displaying the
> previous days close and the 52 week high and low. It asks for a
> market cap less than 500,000,000. It asks that the 3 month relative
> strength be higher than the 6 month relative strength. It asks for
> annual EPS growth greater than 30% and projected growth greater than
> 30%. It asks for price/sales less than 1. Finally I display P/E,
> ROE, and Debt/Equity.
> I run each of these searches weekly. I then look at charts of
> the resulting lists on QP2 and decide if I'm interesting in adding any
> to my MetaStock database. If they look good I'll export the
> historical data to my MetaStock database and run my MetaStock system
> tests on them. If they don't return better than a 100% profit on one
> of my system tests, I'll delete them form my MetaStock database. The
> point to remember is that these searches are not a stock selection
> system by themselves, they are merely the first building block in my
> stock selection system.
> Any comments or suggestions?
>
> JimG
>
|