|
Systems Trading
Pitfalls in Systems Trading
by David Zaitzeff, futures broker at PFG, 800-656-0443 dzaitzeff@pfmail.com
The average futures trader is bombarded with touts and ads
as to various systems he might buy or use. This essay will
discuss the basics of why a trader would use a system, and
pitfalls in choosing or developing systems.
Many beginning futures traders lack a consistent plan--or
path of reasoning--by which they make their decisions. They
may make trading decisions after reading an article in the
newspaper, hearing some news on CNBC, or seeing the weather
report. I've known of clients who bought options chiefly because
it was the hurricane season.
The trading of some beginning traders is almost random.
Some traders choose their trades after receiving an email
touting the "trade of the day" or "trade of
the week" from their brokerage office. Now, some brokerages
have excellent recommendations, but not all do. Some brokers
as individuals have excellent recommendations, but not all
do. Some futures traders remind us of sports bettors in the
movie Two for the Money. Like the clients of certain sports
handicapping services, they are very happy after a few consecutive
wins. They may well increase the size of their positions.
Then, they are caught disastrously unaware as their broker
or service makes several losing trading recommendations.
As a result, to survive and prosper, many speculators begin
to look for a trading system or systems. In fact, some successful
traders are simply those who have found a reliable system.
Some simply stumbled into a reliable system. Others look for
months or years before they find what they want and need in
a system.
Using a system should result in two advantages. The speculator
himself--or the developer of the system--will have studied
or reviewed the previous several dozen or several hundred
highly similar set-ups. (Some system developers review several
thousand.) The one who has studied these several hundred set-ups
knows what now seems to be the likely unfolding of the pattern
he sees.
Based on the past behavior of the market, the speculator can
make an educated guess about the profitable way to trade,
given the current market pattern. And, he can also make the
educated guess that, even if this particular trade today doesn't
work out, the next five or ten trades taken based on the "system"
are likely to result in more profit than loss.
Also, when trading without a system, speculators are likely
to make poor decisions based on fear and greed. The market
goes against them and they liquidate their position at what
is often the high or the low. Some traders take positions
for which there is no good reason other than the excitement
of being in the market. For some, they enjoy seeing if they
can outguess the market in the same way that others enjoy
the slots or blackjack in Las Vegas. Their trading usually
does not last long.
So there are strong advantages in using a system. The speculator
hopes that the system will keep him from making irrational
decisions. And, he hopes that the system will have been tested
in the same way medical researchers test new drugs and will
perform in the future as tested in the past.
It doesn't always work out that way.
As a futures broker I speak at times with potential clients.
I've known several who have paid between $2,000 and $10,000
for a system, software or training program and afterwards
believed that it was advertised in a misleading way. After
buying it, they became convinced it did not do what its sellers
had alleged it would do.
Colleagues in a previous brokerage office spent money on futures
trading systems only to find they were unworkable.
In fact, some say that these persons are members of the $3000
club. The "club" is those who have spent $3000 on
a trading system.
By one standard, they are lucky ones, because they are only
out their purchase price, which is usually an amount in the
area of a "few" thousand dollars. And at least sometimes,
these systems work. What is more painful to hear is of potential
clients who've lost thousands upon thousands upon thousands
of dollars after putting real money at risk in trading the
signals from a series of different systems or CTAs. The first
system they tried was one that was supposedly reliable and
profitable. In fact, it lost money for them. And then they
went from it to a second, and then a third, and then a fourth.
How do we evaluate potential systems? What might make a system
"good" or bad? Is a system likely to work in the
future? How would we know? Can we even make a guess about
that?
Here are questions we might ask of possible systems, software
and CTAs:
Is the system or software subject to any independent tracking
of its signals? There are several services or websites whose
function is or includes tracking the performance of systems,
signals and/or CTAs. These websites are futurestruth.com,
timertrac.com, autumngold.com and http://www.thetaresearch.com
and several others. Is the system, its signals or the signals
of the CTA being tracked by an independent source? If not,
why not?
If the system or signal series was created more than a few
months ago, and if it is any good, it is likely that there
is independent auditing of its signals to demonstrate its
performance since the time it was created. And this is true
even of systems for which no one has yet traded real money.
Once I talked with a potential client who had been through
two systems and had just moved his money to signals being
generated by a "CTA." I asked if this CTA had his
performance being tracked and audited by the people at autumngold.com.
The answer turned out to be no.
Within a few weeks, the results in this fellow account gave
us a good clue as to why the CTA's performance wasn't being
reported at autumngold.com. It was terrible.
In the absence of independent auditing of results, the closest
proxy would be statements or factual information from the
brokerage which holds accounts which are trading the system.
What is the track record of the system or service? This is
a good question. A lot of people ask this question without
asking the question about independent auditing of performance.
In the absence of independent of proof of performance, one
should assume that there are liars out there who lie about
their own performance or the performance of their system.
It has happened to me at least three times when I was doing
some web research on some system or software that I found
reports of CFTC enforcement actions against the developer.
These actions alleged various forms of fraud or other violations
of the securities or futures trading laws. And these were
software developers or systems developers who were still in
business selling their services!
A few years ago I was checking on a system advocated by a
certain brokerage. I asked what its return had been over the
last year and over the last several years. The broker who
was recommending the service told me he didn't know, but if
I would track the signals for one month I would be convinced.
"Seeing is believing," he said.
Maybe seeing one month of signals is enough for some clients,
but I prefer to see the results over several years.
I know one fellow who makes market predictions or calls, many
of which have been recorded in books or newspapers. He claims
to have a track record far better than he actually has. He
does this by reporting falsely what he had predicted so that
his predictions match what the market later did. You wouldn't
spot the fraud unless you check the old newspapers or the
webpages which have those old newspaper articles.
Is the "track record" of hypothetical performance
in the past, or was it produced real time? When was the system
created and what has been the performance since then?
Unfortunately some futures brokers have misled some clients
in some instances. Potential clients have told me that their
futures broker told them that a certain program was great
and had produced returns of 100% a year over the past several
years.
In fact, the 100% annual returns were hypothetical returns,
not with real money, produced in real time.
And why does this matter? The behavior of the market is fully
able to change over time. For example, the year 2006 has seen
less volatility in the S&P and the currency markets than
have previous years. Trading systems which were designed in
years of higher volatility, and which were designed to exploit
volatility, have "gone bad" recently. The reason
isn't fraud on the part of the system developer, but the market
itself has changed.
If the trading results were produced real time, will a change
in market behavior ruin the results of the system? While this
has happened recently with certain volatility-based systems,
the same potential exists with many common programs of selling
stock index options.
Selling stock index options can regularly provide a monthly
increase in one's account, but with the risk of ruin if not
handled very carefully. The best example of this is the case
of Long-Term Capital Management. They were a giant hedge fund
with billions of dollars under management. On their staff
was one Nobel Prize winner, who was partly responsible for
the creation of the Black-Scholes option pricing model. LTCM
had many months of gradual growth. They (and others, for that
matter) were short option volatility going into August 1998.
In August 1998, nearly every position held by LTCM was going
against them by abnormally large amounts. Their mathematical
models had indicated that this event should never occur in
the lifetime of the universe. Yet, it was happening before
their eyes. Within a few weeks, the Fed organized a consortium
of banks to purchase LTCM and gradually close out all their
positions.
Other important questions relate to drawdowns incurred as
a result of trading the signals of a system. There are many
seemingly profitable systems which have had drawdowns of equity
of 50% or more. And, many traders wouldn't continue to trade
a system after having experienced a drawdown of 30 or 40%,
no matter how good it seemed in the past.
For some traders, the only tradeable systems are those which
have drawdowns of less than 15 or 20% of equity from any given
point in time.
Note that a system developer may, inadvertently or intentionally,
hide his drawdowns from potential clients. Here is one way
it is done. The system developer says, "I've recently
done an extensive study of the market and derived several
new indicators and a system for trading the market. And, over
the last several years, this system would have taken several
hundred points from the market in trading S&P futures.
Moreover, since my system began to function real-time, it
has continued to take substantial profits."
Does this sound impressive? If it is true, it is. However,
what is missing is this: If a person had been trading using
this system in the last six or ten years, how large would
have been any and all large drawdowns?
Moreover, suppose one had decided that as one's account increased
in size, that one would take progressively larger positions.
Many traders do exactly this. If one has increased his position
size and simultaneously entered a period of consecutively
losing trades, how bad will things get for one's account before
starting to win again?
There are at least two functioning websites right now which
offer stock market signals and which have had at times a seemingly
fantastic return. They have a seemingly fantastic return until
you also consider the actual or hypothetical drawdowns some
of their clients have experienced. And those drawdowns make
them more like the sirens of old than a rock on which a trader
could rely.
Another useful question to ask is whether or not the input
variables of the system (or model) have a rational relationship
to the price behavior.
Lets consider the following system: If the close of today's
S&P is lower than yesterday's close and lower than the
close of 3 days ago and of 4 days ago, and if the close of
3 days ago is lower than the close of 4 days ago, buy one
S&P the next day market on open and exit on the close.
A Tradestation simulated run indicated in January of 2006,
that "system" would have resulted in large profits,
hypothetically speaking, over the previous six years. Now,
the system works over the past in the S&P, but not in
the Russell 2000. Are the results of the system a fluke?
From January 2006 to September 2006, the system would have
produced a large loss.
Why? Is it simply a statistical fluke over period of years
and covering dozens of trades? Did the market change its behavior?
Is the loss in 2006 simply a drawdown in a system that will
be profitable in the long run? I don't know. I have no idea,
in fact.
What I do suspect is that if a system is being built, it has
a greater likelihood of "working" after it is built
if the inputs, variables or paramaters of the system have
some rational relationship to future price action. Otherwise,
the market itself may change. Then, the system might no longer
function profitably.
Traders must realize that markets do change over time. The
use of systems themselves changes the markets over time. Systems
may increase or decrease market volatility and change it in
other ways. Also, as any given profitable system becomes known
and widely used, market participants may act in ways to take
advantage of what they know to be the behavior of those following
the system.
Some markets are believed to have a consequent effect on other
markets. For example, some believe that the price of crude
oil and/or treasury bonds has a lagged effect on the prices
of stocks. If this is true, then, systems might be built to
exploit this. And, such systems would perhaps continue to
work, despite changes in market volatility.
Because markets might change and one system begin to fail,
one strategy that some would use is to allocate various portions
of one's funds to the signals of diverse trading systems.
There is a risk of loss in all trading. Information in this
report is from sources deemed reliable but not guaranteed.
David Zaitzeff is a futures broker at PFG. He may be reached
at 800-656-0443. dzaitzeff@pfmail.com
|
