Mr. Index Trader is a trading system developer whose systems have frequently appeared in the Top Gainers tables on Futures Examiner. His systems include Frequent Trader, Flash, and the newly released Turbo Trader. Some of his systems are traded exclusively with Fox Investments. You can refer to his website, MrIndexTrader.com for more details. In this interview by Ken Morin of Futures Examiner (2005-Dec) he shares some of his insights about trading systems, and some of his future plans.
Futures Examiner (FE): Can you give us some background information about yourself?
Mr. Index Trader: Before I turned to trading system development, I was a computer
systems engineer. That explains my analytical and mathematical background, which is so
helpful in developing trading systems.
FE: Can you tell us why you use a pen name instead of your real name?
Mr. Index Trader: Pen names or pseudonyms are a common thing for authors, actors
and musicians to use. They allow a person who wishes to maintain privacy, to share their
creative output without losing anonymity. In the case of Mr. Index Trader the name is
an alter ego, like Superman (!!), and when I am not trading or programming, I can return
to my family as Clark Kent.
FE: How do you evaluate the "quality" of a trading system?
Mr. Index Trader: That is a great question. So many people out there aren't
willing to trade, or auto trade someone else's system. They think that designing a
system is simple, and that they have to make one themselves, but they have no idea how
hard that is. I work full time as a system trader / designer and come up with about one,
maybe two systems per year.
One of the first things I look for is an equity curve that looks like a ruler; i.e. very linear, and consistently profitable. I don't like systems whose equity curve has tapered off in the last year or so of low volatility.
I also look for a decent profit per trade after slippage and commissions. As an example, the Frequent Trader hypotheticals vary from about $80 to $100 per trade before subtracting slippage and commissions. The profit per trade has skewed a bit downwards in the last two years, as volatility has been low. If you see a system that shows hypotheticals of $40 or less per trade, even if it claims to use limits - I don't feel its a tradable system. [FE: The Average $Profit-or-Loss /trade values reported on Futures Examiner are net of slippage and commissions. For Frequent Trader, the actuals currently (2005-Dec) range from about $20 to $60 /trade.]
Statistical validity of a system is in direct proportion to the number of trades over which it has been tested, so I like highly active systems because the sample size is bigger. One of my personal rules is, I never consider trading a system that has been tested with less than 300 trades. The beauty of Russell 2000 is that it trends so well, and systems that trade every day give a sample size of 900 or so trades in a backtest. This makes for a statistically robust model. I would consider a bond system that trades 2 commodities with 150 trades each, as long as the same parameter values were used for both commodities. But watch out for a system with 600 trades that consists of 6 systems all in one, because that's an average of only 100 trades per system, which violates my 300 trade rule.
I also look for a high profit to drawdown ratio. For example, Frequent Trader Standard miniMD with zero slippage and commissions gave a profit of $77K (over about 4 1/2 years) and maximum drawdown of $1660. That's a 46 to 1 ratio. [FE: Futures Examiner reports the average annual profit (currently 194% for Frequent Trader Flash miniMD hypothetical), and maximum drawdown (currently 31%), which yields a ratio of about 6 (all data is after slippage and commissions). The corresponding Sterling ratio of around 20 (shown on the Statistics page) provides a similar calculation (the ratio of average annual profit/loss to average annual maximum drawdown). This Sterling ratio is currently one of the highest on Futures Examiner.]
I also look at out-of-sample performance. This can be obtained either via backtest, or from post-release performance (which by definition is out-of-sample).
Walk-forward testing is a type of out-of-sample testing that can be used to evaluate the forward-trading ability of a system. For example, a system might be optimized over a 4-year window, and then forward tested over the subsequent year's data. And this process can be repeated a number of times. See the following link for details. Good systems are extremely hard to achieve, and looking at the carnage of systems over the years provides evidence of this!
Finally, I also look for published actual-trade results. This is critical as it gives the system credibility, and it hopefully shows the system made money in out of sample testing. There are also plenty of systems that make money in theory, but not practice. For example a system that makes $40 per trade on limit orders. If only 80% of the limit orders are filled, it means you miss out on 20% of the trades, which are more likely to be the winners.
FE: What warning signs do you look for that a system may soon "break"?
Mr. Index Trader: Concern about a system "breaking" is valid, and I note that the last
few years have been horrendous for many trading systems. Many systems out there are not
very good, and many of the previously good systems have died (i.e. lost a lot of money)
in the last few years. Even if a system has been good in the past, market conditions can
change. Many people look at past drawdowns, and use money management and Monte Carlo
analysis, but then assume that future performance will be the same as past performance,
which can be painfully wrong. I believe that you should assume that future drawdowns
will exceed past drawdowns. The bad news is that there is zero guarantee that the system
with the best track record in the world will continue to work tomorrow. The good news is
that by asking the right questions, a knowledgeable system trader can often avoid systems
that may perform poorly going forward. See the previous question on evaluating the
"quality" of a trading system.
FE: What are your philosophy and goals regarding trading system design?
Mr. Index Trader: My philosophy is that specific market trends can be identified and
profitably exploited by those who invest in statistically validated research. Money flows
from those without knowledge to those with knowledge.
For a client choosing a trading system, this would mean to ensure that the system has been tested in a statistically valid manner, as discussed above regarding the "quality" of a trading system. To recap, this means that the equity curve should look more or less like a straight line; the profit per trade should be high; the system should be tested over at least 300 trades; it should have a high profit to drawdown ratio; its out-of-sample performance should be good; it should perform well in walk-forward testing; and there should be published actual-trade results.
FE: Some developers like to release new versions of their systems periodically.
What are your thoughts on this?
Mr. Index Trader: This is a common thing to do, either to provide several variations of
a system, or to provide an updated version of the system. For example I have several variations
of Frequent Trader, and plan to release two variations of Turbo Trader. The use of
variations allows traders to diversify somewhat by trading 3 slightly different systems
rather than trading 3 contracts on one system, and this can provide greater smoothness
in the overall equity curve. It also means that for a given system, the combined users can trade
3 times as many contracts in total before slippage becomes an issue. Regarding new versions,
I feel that good trading systems should not need frequent optimization. However, when a
developer has made improvements to their code, they may release an updated version so people
can trade their best work. This is fine so long as the developer hasn't curve fitted the
new code to just a recent amount of historical data, or the new version may not perform
well for very long.
FE: Do you trade your systems with your own money?
Mr. Index Trader: Yes, I do. Everything I produce is tested for a few months in my own
accounts before release. And after release, I continue to trade everything too.
FE: Frequent Trader been a very profitable system. To what do you attribute its success?
Mr. Index Trader: About two years ago after painful S&P 500 trading losses, I concluded
that the Russell 2000 was the best vehicle for trend following systems, and focused
purely on it. For pure system traders, I believe that the Russell 2000 trends much
better than currencies. Today, there are many more Russell 2000 systems than currency
systems. (Though there might be more currency traders than Russell 2000 system traders.)
Frequent Trader is a family of fairly simple trend following systems, but has some
significant advantages as the systems look at other things going on in the market,
apart from the price action, that tell where the market is going. It's profit to
drawdown ratio is superb, and it's risk per trade is small.
FE: Are there any other systems that you would like to tell us about?
And how do they relate to Frequent Trader?
Mr. Index Trader: My newest system, Turbo Trader, has been running for just over two
months. It enters the market much earlier than Frequent Trader, and is far more active,
trading up to 3 times a day. Turbo Trader can do about 30 trades a month, while the
Frequent Trader systems typically do about 18. I'm looking forward to results being
published on Futures Examiner. Turbo Trader is also available for any broker or client
to trade - it is not exclusive to Fox. I spent many months trying to get a completely
new Russell 2000 system with no results whatsoever. I was feeling very discouraged,
but I learned a great deal about how the Russell 2000 trends. All of a sudden a new
system was born, combining my new insights with things that I knew from the Frequent
Trader systems. Turbo Trader made $3,000 in about a month, then lost $2,000, and then
zoomed up to make almost $5,000 in my test account - all in two months. It then hit
a decent drawdown as the Russell 2000 was not trending well in November and early
December. For my systems to make money, the market must trend well, but unfortunately
this doesn't happen all the time. After such glorious profits in October, I was bracing
for a difficult November, and had not increased my position size with October's profits.
[FE: Traders take note: It can be sooo tempting to increase your position size when
a system is "hot", but this may be just the time that a drawdown will occur.
It has happened to me!]
FE: What is your favorite feature of the Futures Examiner website?
Mr. Index Trader: Very simply, the combination of actual results and hypothetical
results. Futures Examiner is doing people a real service by posting both. I believe
that using actual results in combination with back-tested hypothetical results, is a
good approach to evaluate probable future performance. I look for good hypotheticals,
some good out-of-sample results, and I also check what typical slippage is on one or two
months actuals vs. hypotheticals. [FE: We like to plot them both
on a multi-chart, and see how well the actuals correspond with the hypotheticals. This
is easily done from the Summary Results display by checking the "Multichart" box for
each actual and each hypothetical that you want to chart, and then clicking the
"Multichart" link near the top of the page.]
FE: Are there any other comments that you would like to make?
Mr. Index Trader: If you want to make money trading, then I believe that it's a safer
approach to figure out who is making money, and how they did it, as opposed to looking
at hypothetical results or a course that looks good. Brokers rightly tell me, they get
tired of looking at hypothetical graphs based on in-sample backtests: they all go up
and sort of look the same. Actual results are seldom as good.
[FE: That is exactly why I developed the Futures Examiner website. First, so I could
look at various hypotheticals and compare apples with apples. By this I mean that
results for every system are net of all costs, including the cost of the system, and all are presented
in exactly the same way, including statistics that are calculated the same way for all.
The hypotheticals may still all go up, but when you compare them on a multi-chart and
look at their statistics, they no longer all look the same! And secondly, by also publishing
actuals for the same system, its easy to see if the actuals track the hypotheticals,
and if they do then I consider the hypotheticals to be more meaningful.]
FE: What are some of your favorite books about trading?
Mr. Index Trader: Reading is important for intellectual growth and understanding.
However, the trading books that I have read rarely gave information that had been
statistically proven using a sample size of > 300. I place little value on magic
indicators that are demonstrated using a sample size of one. The shown example of one
nearly always works. If may be correct, but if its not proven on a large sample size,
I place no weight on the research. I like
TheGrailSystem.com genetic algorithms
that can give you thousands of systems based on a particular sample of data, but
nearly all the systems lose money when they are tested out of sample. Of course a very
few of the generated algorithms are good, and make money out-of-sample also, but my
point is to illustrate why I like to use large amounts of out-of-sample data. Small
sample sizes tend almost always to fail in forward testing.
Wealth-lab
is also a very powerful tool for analysis, but one has to be careful to ensure that
some of the Wealth-lab published code doesn't accidentally cheat by peeking into the future.
FE: What future plans do you have that you can share with us?
Mr. Index Trader:
FE: Thank you, Mr. Index Trader, for your insights into trading systems and their design. I hope that your thoughts on evaluating the quality of trading systems will be helpful to system traders. I also look forward to seeing results from your new Turbo Trader system posted on Futures Examiner, and I wish you all the best in the future.
Last updated 2005-Dec-21