Attack of the Trading Robot

Posted by downtowntrader | 12/18/2008 08:51:00 PM | 0 comments »

The folks at Trade-Ideas have recently released an onslaught of videos showing their trading robot (jointly developed by in action, and it's sparked some interest in automated trading. The videos are great, and they do a good job of showing what is possible with the use of a trading robot. Dave Mabe of has also been writing recently on automated trading and has some great posts on his blog. As many of my regular readers know, I've been running the bot for a few months now. Recently, I've received a few emails asking several questions including;

  • Are the results real?
  • Does it ever blow up and have a really bad day?
  • Is your automated system profitable too?
  • How long did it take to develop your system?
  • Could you just use trade-ideas and trade the alerts manually?
  • Do you intervene with your system?
  • Can you use the canned alerts that come with Trade-Ideas?

First of all, I have no doubt the results posted by Trade-Ideas are real. If you take the time to properly develop, back test, and then forward test a system, there is no reason it wouldn't perform consistently moving forward.

So how do you develop, back test, and forward test a system? The first step is finding a suitable alert and filter combination from the Trade-Ideas suite. The way I handle this is to think of a market behavior I want to take advantage of, and then recreate it with the alerts available to me. I then back test it using oddsmaker, which will let me know if it is a profitable idea over the past 14 sessions. I use the back testing feature to act as a filter on whether a system is worth pursuing. However, I've never felt comfortable relying on strictly back testing, so I then I forward test it by running it with the paper trading feature of Interactive Brokers. I closely examine slippage, stop performance, influence by time of day etc., during the forward testing period. Another often overlooked, but important part of forward testing is watching how the system trades through different market environments.

Does it blowup. I've never had the robot blow up using real money. I've had pretty spectacular blowups while forward testing, but luckily by combining back testing with forward testing, many issues are avoided.

Is it profitable? Yes. As mentioned above, if a system can graduate from the forward testing phase, it means it is profitable.

How long did it take? Developing the system doesn't take too long once you have created a few. If you have a good idea, creating and back testing it can all be done pretty quickly. I like to forward test for at least one month and preferably two or three months.

Here is a chart created by that shows the results for one of the strategies I have developed for the robot. It shows the cumulative R's for the system from the forward testing stage to the current date. This is a long only, momentum based strategy. What I like best about this system, is that despite the long bias, it has been very consistent at avoiding draw downs. It has taken very few trades during the recent obliteration of the markets (needing almost three months to net 10R's), but it hasn't lost money and I have other short systems that take better advantage of weak markets. I think the chart shows that even in this environment, positive results can be achieved with proper development of an automated system.

Could you trade the alerts manually? Yes, in theory you could. In practice, not so much. Looking past the often discussed benefits of trading an automated system vs. a discretionary one, I find that the robot can trade systems that would be near impossible to trade manually. One of the advantages of using a robot is that it can act at lightning speeds and dissect huge amounts of data. It can enter several orders at a time and eliminates human errors on placing those orders. And I can attest to how fast the bot reacts to an alert and gets the order in well before you could manually.

Do you intervene with the system? While I have and won't rule out intervening, I don't do it that often. I've found that while you can step in and save some money at times, in the end it evens out with the times you exit early. However, since the systems have been extensively tested, I do have a feel for what market environments cause under performance, and I have no qualms with stepping in and flattening out.

Can you use the alerts that come bundled with Trade-Ideas? Yes you could, but I wouldn't recommend it. From my experience, most of the prepackaged alerts are a little noisy and would need a large amount of trading capital to use. However, they do have an extensive list for which you can base your ideas on.

Another benefit to trading a robot like this is for traders that have a non trading job or that have to be away from their trading screens, yet desire the safety of ending flat each day. You can create strategies for any part of the trading session and trade part time. I have systems that trade only the first 15 minutes and others that won't take a trade until 10:00am EST. The beauty is that you can create something that works around your schedule and can be as active as you want it to be.

If you're interested in learning more about Trade-Ideas, feel free to contact . Tell them I sent you and ask about a 7 day demo.

You can also find out more about the robot I use and stocktickr here.