I started with no experience on the Tradestation platform and came across a few hiccups during my month of learning, which cost my simulated account a substantial negative impact.
I started with $10,000 and ended the month down $242.33 to finish at $9757.67
Now, that's only a 2.4% loss for the month, which is't bad at all really, but when compared to the results that it should have been, the discrepancy is a little unsettling.
If you look at the far right totals for both the real and reported month end, you'll see that the strategy that was actually running produced a profit of $42.69 while the backtest result (performed at the end of the month) reported a $36.92 loss.
How did the actual performance have me losing $242.33 when the live running strategy actually reported a $42.69 gain? Let me introduce you to TradeStation's quirks.
For whatever reason (perhaps the performance capabilities of the computer I'm running things on, mixed with internet instability) at one point the strategy was giving trade signals that the platform was not executing... So that generates trades on the chart that never happen in real life. There were also problems that kept the strategy from generating trades that it should have as well. Those are the two reasons for the discrepancies.
One quirk I have noticed on my dedicated desktop is that when I load up TradeStation portfolio, there are almost always problems with the strategies initializing. I will have indicators not load on the chart, so I will have to disable and re-enable them before the platform recognizes the Robot. This problem has not cropped up on my more powerful desktop in the office, but I use that for many other things (Such as testing) and don't want to run the portfolio on that machine.
I do have a handy method in place for helping me keep track of the platforms performance. I have text messages setup to tell me when the strategy fills an order, and I know to expect another text immediately that says TradeStation filled the order as well. If I don't get that text, something is wrong and I need to go check the platform.
Anyways, my present conclusion is that the robots do perform as described and advertised in execution terms, give or take a few dollars here and there. I will more than likely be running these robots on a small live account soon with 10 times smaller positions just to see if live money execution is drastically different than simulated positions.
In the mean time, I have also decided to optimize these robots in accordance to the previous 5 years of market conditions. These robots were developed from roughly 2003 to 2007 to perform in this time period. I'd like to optimize them for the last 5 years and see how they do in earlier years.
I can understand ones apprehension to fiddle with something that works, but I think a lot of that has to do with a lack of confidence or ability in ones own mind. These robots do very simple things: They use Moving averages to determine trend, and then use Relative Strength of the current market to identify a good time to enter and sometimes exit. There are many useful variables in the programming to take advantage of, but my point is they arrived at their current profitability through the same work I am going to be doing to modernize them for the current time period:
Hours and hours of testing and paying close attention to the response of inputs tested. So far I have about 4 of the robots producing more profit with the same or less drawdown. It's something I will continue to work on for a very long time and perhaps as long as I am a robotic trader. I love data, and this beats manual testing any day.
As far as my non-trading life goes, I am now rocking a pink cast which will be removed in 6 weeks, No surgery required thankfully. My new tail section is arriving on the 10th and I'm dying to get back on the bike. I'll have to rig my cast up somehow to make sure it will not slip off the right peg. My new left side peg is here and ready to be installed. I'm going to the cinema-pub for valentines day to watch Zoolander 2! I will be back in the gym rocking it very very soon! Turns out the bus isn't so bad at all :)
No comments:
Post a Comment