Sunday, September 16, 2007

Futures System Trading - How To Choose A System

If you are new to system trading, after reading my previous article ("Futures System Trading - Reality Check") you might feel a bit uneasy about the whole business. That's good because you should. There is a vast wild jungle out there with swamps scattered all over rather generously. One false step and you just said good-bye to a nice chunk of dough.

How then should you choose your system, you may ask. The short answer is: the same way hedgehogs multiply, that is, cautiously... The long answer is that you have three options and each of them can be good if used judiciously.

The first option and probably the best one is to find a vendor who offers his system through a broker (using Tradestation or Strategy Runner to generate orders) and charges you based on the actual profits his system makes in your account per month. That usually means a 10-20 % cut of real profits for the vendor. Vendors like that are few and far between and if you ever decide to choose one like that you want to make sure that you know how his system performed in the past in a real account and not on paper. The broker that handles vendor's business or the vendor himself should be able and even eager to provide this kind of information. If they can't, don't bother as this is usually an indication that you are dealing with some monkey business. If the system is new and there is only a limited amount of information about its actual past performance you may want to wait a quarter or two to see how the system is doing. Rush is never a good thing in these matters.

The second option is to buy a good system from a reputable vendor. You want to buy a system that is fully disclosed and it is very advisable to choose a system that has very little room for curve-fitting (no more than 1 to 2 parameters that are optimally adjusted in the backtesting process) over a system that has plenty of room for this. The latter are usually less robust than the former. If the system is not fully disclosed (i.e., it comes as a gray or black box) you will never know if it was optimized and to what extent. This is not good as it is rather easy to produce a system with a stellar past performance by curve-fitting it to the data. It is very naive to expect that the system designed this way will continue its stellar performance. The opposite is more likely, that is, the system, being not very robust, might unravel as soon as you start using it. Now, how to make sure that you are dealing with a reputable vendor? I would dismiss all hypsters as a rule. A good system can speak for itself, no hype is necessary. I would also avoid vendors who are not very forthcoming with information on the realistic system performance: for instance, they do not account in their advertising for the slippage and commissions in a realistic way. This can have grave consequences as the previous article was meant to show you. Particularly insidious can be 'non-fill' slippage occurring in systems that use limit orders. As opposed to regular slippage caused by the use of market or stop orders, the kind of slippage in question is not always easy to estimate and if not accounted for can lead to significantly inflated profits. It can even turn an essentially losing system into a great looking winning one.

I believe that the only honest way to account for this kind of slippage is by disregarding all the trades whose entry or exit prices were not penetrated by at least one tick. A robust system will survive this type of cleansing, a bogus one will not. I do this routinely with my systems, but alas, to the best of my knowledge, no one else does. If you are still wondering why, you may want to re-read my previous article. Another issue is regular slippage which should be estimated realistically depending on the particular market's liquidity. For instance, this kind of slippage is smaller for a market as liquid as the S&P500 emini futures (ES) than for the Russell 2000 emini futures (ER2) that also enjoys some popularity among traders. Finally, you definitely do not want to overpay for the system. I think that nowadays you should be able to buy a good fully disclosed system for less than $1000. However, most vendors still think that they can afford to charge more. I would avoid them. If a vendor really believes that he has a good system that is worth more, he can always generate a steady income either by employing the first option mentioned above or by leasing it (option three to be discussed next). Finally, it's good to check if a vendor offers a money back guarantee (at least conditional) for his system. Most will not, so those who do should, in my opinion, be given priority over the others. You can certainly agree that a vendor who offers some form of reasonable guarantee has more faith in his system than a vendor who shuns any idea of such a guarantee.

The third option is to lease a system on a monthly or quarterly basis. This is a good option, but very often not as good as the previous ones. Electing a system for trading in this way requires as much prudence if not more as in the other options, the reason being that when the year of using the system comes to a close you may end up paying much more for the system than you would by buying it outright and still have nothing to show for (see the previous article for an example of a situation like that). This is so, in part if not largely, because the subscription fees are absolutely not commensurate with the system actual performance, so be careful not to overpay. As a rule, I would avoid any vendors who charge more than $150 a month. The majority of them will hardly ever deliver profits to your account when all is said and done and so you want to be frugal as much as possible. Beware though of the common trap: people tend to think that if something is expensive it must be good. This is absolutely not true! A vendor who charges $300 a month for his system may not necessarily deliver greater profits than the one who charges only $150. The past hypothetical performance cannot be used as a justification for higher fees.

All systems are born equal every single quarter and the system is only as good as its next quarter and not its past 5 years. You also need to realize that unless a vendor backs up his claims of past performance with a Tradestation performance report, you should not put much faith in what he claims. However, even with the Tradestation report available you still don't know if the system has not been curve-fitted and so you might end up paying a lot for something that could be performing much worse than the past performance would indicate. It should not come as a surprise that this option is most frequently used by vendors. The reason is quite simple: they can keep milking you forever, no matter whether they deliver or not. Unlike in the first option where the vendor's fee is tied to your account's actual performance or in the second option where you can get a system for life for a one-time fee (and not only can you use it but even learn from it if the system is fully disclosed which is by far the best deal in this option), in the last option you are hardly ever in the winning position and so the only way to make sure that you do come ahead as a winner is to ensure that your subscription fee is as low as possible.

Waldemar Puszkarz, Ph.D., is a web veteran with 15 years of web surfing under his belt. By training, he is a theoretical physicist, but his interests are much broader than science and include trading financial markets, sports betting, poker, and researching online business opportunities. He is also an avid book reader and sports afficionado. Currently he is making his living mostly as a day trader. He has been in the trading trenches for almost a decade during which he has traded a variety of financial instruments. He is the owner and webmaster of ( which provides free common sense trading education and simple trading systems for e-mini and stock markets as well as reviews of honest online business opportunities in Meet HOBO ( section of his site.