[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[EquisMetaStock Group] optimize--MG's question



PureBytes Links

Trading Reference Links

Good afternoon, MG. Your questions are interesting ones as usual.

Because of your major contributions to the forum, your background and
your experience, I would like to respond to all of your questions with
a reasonable amount of detail, which would probably provide more
energy for further discussions that might benefit others as well as
ourselves. However, I am somewhat limited in my detail as a result of
having to respect Roy's newsletter and the people who pay a very
modest fee to subscribe. 

I agree that subjectivity in a trading system is a tricky subject.
I've always believed that a mechanical system is highly unlikely to be
traded as a strict mechanical system when that system is in the hands
of an individual. It's very difficult to remove judgement or emotion
from someone who is not only executing the system but tracking it's
performance with their personal funds. Anyone who has the iron will
and stomach to stick strictly to a mechanical system as the drawdowns
grow, is someone who has the discipline to trade without such a
system. I think various research projects by Future's Truth Magazine
has shown this to probably be true. 

On the other hand, mechanical systems make sense for money managers
with larger amounts of capital and staff resources. The corporate
structure is usually diversified, disciplined and monitored
sufficiently to get the best from a mechanical system. 

For me, I feel the human mind is the fastest, most flexible CPU there
is and I don't want to remove it from my decision making even though
sometimes the human CPU can be significantly disrupted by all kinds of
events. Unfortunately, I don't see Kalman filters as an indicator of
how much I should train my neural nets. Everyone in my family thinks
they're in charge of that task. 

In my trading, I look for four kinds of markets: uptrends, downtrends,
consolidation with leadership in tact and consolidation with rotation.
Consolidation includes sideways markets, normal pullbacks and various
other gyrations that aren't trends. 

Regardless of the market conditions, I use the same methods for
finding my prossible trades. Mine is a combination of TA, quants and
fundementals along with a pure TA exploration that is very effective.
I don't vary that technique because I haven't found a method that
produces a better pool of candidates.

The size of the pool of candidates changes as market conditions
change. In an uptrend, I'll find 50 to 100 trades that look
reasonable. I have a set of chart rules that I use to evaluate which
of those candidates I'm willing to take a risk on. My chart rules
don't change according to market conditions, either. 

In a consolidating market with leadership in tact, the pool of
candidates might shrink to less than half as many as when the uptrend
is clearly visable. Obviously downtrends stop my long trading and I go
short. Markets consolidating with rotation, or alternatively that are
directionless, calls for no trading, or very, very short holding periods.

How much subjectivity is there in the chart rules? I would say that
there is little flexibility and subjectivity in creating the pool of
candidats, but somewhere in the area of 40% in the evaluation of a
chart as to which of those possible trades to take. 

I've used systems testing to create the part of my system that selects
stocks for inclusion in the pool. I've tested that system in all four
market conditions and I know the trade statistics that are likely to
occur in each case. I've developed a method for figuring out which
market condition exists, and I have a pretty good idea of what to
expect from the pool based on how I see the market at that moment. 

However, when the 40% (plus or minus of course) subjectivity comes
into the final selection of which stocks to buy, then the systems
tests are of little value. I've figured out those trading statistics
by analyzing the final trading results. This gives me feedback on how
my chart reading rules are working.

I have a standard chart template I use to look at stocks. I only use a
couple of indicators to confirm what I think I'm seeing. While I don't
use the indicators as a reason to buy, I sometimes use them as a
reason not to buy. I'm mostly looking at price volume action, with the
indicators as a back stop in case I'm reading that wrong.

In addition, to the total number of stocks that look like good trades
changing according to market conditions, my rules for size, exits and
stops change also. 

If the market is in an uptrend like it was in Nov and Dec of 2005, I
use longer MA's, loser stops and less rigorous exits. In January I
tightened the stops and exits and shortened the MAs. In April I didn't
trade. My holding periods change accordingly as you would expect. 

I'm mostly a long only trader. I do short the indexes when I see a
downtrend, but I'm even more conservative in my shorting. With
commissions so low and execution times so high, there's little
economic reason not to be conservative when it so cheap to buy back in
if you're stopped out too early. It takes some degree of discipline to
go back to a trade after being stopped out, but my number one rule is
don't lose money.

I have to live strictly off of the returns on my capital account so
preservation of my capital account is a rule with zero subjectivity. 

A lot of people want to automate the chart reading criteria that I use
to narrow my lists of trades among the choices I have. I don't think
that works very well because the subjective evaluation is necessary to
respond to different market conditions and to consider all of the
variables and their constraints. 

If you read Roy's newsletter you will see how Pareto's principle and
my education in the mathematics of operations research has shapped
what I do. I have an explanation in there about why I don't believe
maximization or minimization despite the high usage of these
principles in econometrics and production management. (Even an
excellent, expert, programmer such as yourself would find some
interesting and new things in there. I do and have increased my code
library a lot because of it. You would even get a few new systems
develop ideas out of it.) 

I don't optimize much using the systems tester. As I said in my other
post I optimize mostly to evaluate robustness. I use the systems
tester as a relative comparison and to figure out how much the trading
statistics change from one market condition to another. 

When I do optimize I optimize the sytem to perform the best in
uptrends, and then I add rules to restrict it's ability to trade in
other market conditions. 

I do not feel that optimizing or testing across a wide range of market
conditions as a means of optimization is a good idea. I do test across
several years to look at overall systems performance, but I also test
year by year to correlate the performance with the market conditions.
I also test on various subsets of the universe of stocks. I use in
sample and out of sample data, especially when I want to evaluate the
performance of the system in a particular market type, or to determine
the effectiveness of my filters at preventing trading when the trading
statistics for that market condition are below my acceptable thresholds. 

I see trading as more about strategy than about indicators so I don't
waste a lot of time with indicators. Trading to me is the management
of probabilities. Unfortunately the mathematics of probability is one
of the hardest math techniques for people to understand. Too many
people assume they understand probability because they can get a feel
for a one in ten chance or 40% losers and 60% winners. To me, that's
not probability, it's performance statistics. 

As you know, probability is understanding independent events and
dependent events and how to determine the possibility of different
combinations occuring. 

Well, this is a long post so I won't go into any more detail. The
detail and code is in Roy's newsletter, but I hope that this much of a
discussion has at least been worth the time it takes to read it. 

This stuff works for me. Obviously there are thousands of ways to
trade. I've looked at as many of them as I could, and what I do is the
best fit for me, and it produces great results on my scale of results
measurements. My three primary requirements are: don't lose money,
consistency is more important than big numbers, and make enough to pay
the bills and increase my capital account to offset inflation and any
slips in rule number one.

Thanks again for your contributions. While a lot of them are way over
my head, I'm sure many people appreciate the time and effort it takes
to post them. 














------------------------ Yahoo! Groups Sponsor --------------------~--> 
What would our lives be like without music, dance, and theater?
Donate or volunteer in the arts today at Network for Good!
http://us.click.yahoo.com/Tcy2bD/SOnJAA/cosFAA/BefplB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/equismetastock/

<*> To unsubscribe from this group, send an email to:
    equismetastock-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/