Has anyone done any forward testing analysis on Automatic Pattern Search results with delay filter applied? I find the delay concept very appealing but I wonder whether it results in optimized formations.
So, you finally got a copy of the program? If not, I think you should try an "inexpensive" one that also generates trading systems automatically: http://www.tradingsystemlab.com/howtoorder.aspx lol
I would love to try that one but they don't have a demo you can download and they require customers to sign an acceptance of restrictions regarding external use. Too expensive though but hedge funds can afford to spend this amount for sure. APS from tradingpatterns.com at the other end of the price spectrum does the job of automatically configuring trading systems but limited to price patterns. I wonder if the expensive one comes up with fitted systems. But no way of checking that. John
I think it is more an issue of cost than anything else. 60k for a program isn't exactly peanuts for anyone, even for hedge funds, and I noticed they require a fee of 20K per year after the first 2 years. APS can still get you tons of patterns at a fraction of the cost. I don't know if the data mining used by APS is much slower than the genetic programming used by TSL. Tradingpatterns seems to have come up with a very clever solution for speeding up the search using multiple copies of the program running on different cores of the same CPU: http://www.tradingpatterns.com/About_Us/articles/multicpu/multicpu.html I think the approcah is interesting and other software vendors should consider it as an ad-hoc/cheap solution to the multiple core technology without having to rewrite their program code to do that. If you have two unrelated tasks to run, like displaying two different charts with indicators on them, then two copies running in parallel on two different cores can do the job a lot faster by displaying one chart each. My 2 cents. Ron
I have APS and I think it is a fine piece of software. I just think they should add the profit factor as one of the criteria used in the search. This parameter is more important than success rate or profitability.
I found the article about the performance of the QQQQ patterns APS came up with in 2002 very interesting and I wonder whether someone who is an expert in applying statistical analysis can figure out a way of determing in advance the likelihood that a specific pattern stays profitable in the future: http://www.tradingpatterns.com/About_Us/articles/article3/article3.html I think figuring out a way to rank patterns according to likelihood of consistent future performance is one key to success. Any ideas? John
FYI - From the Trading Patterns Web Site: 08/10/2009: Important announcement. Tradingpatterns.com will cease retail license sales and all technical support operations on August 21, 2009 at 5:00 PM EST. I understand the web site will remain up for a while. There are some interesting articles available for download. Jack
He writes on the website: "APS code is up for sale. Introducing brokers click here". The APS program is out-of-date and will require serious improvements if another party attempts to market it.
Actually the latest version was released sometimes June of this year and it is fully compatible with VISTA 32-bit and 64-bit. Maybe the code is old, if you mean that, but a good programmer, like a young guy who worked for me in the past, can re-write it in C# with some moderate effort. For those who know something about code it looks like Delphi to me. I have followed this story developing and it seems that Michael Harris, the owner of the code, is making a reasonable move. Why selling to retail for $1,500 or about when you can work with hedge funds providing customized solutions and make multiples of that on a per sale basis? I see things are changing in the trading software industry. I sense that traders will either have to pay a lot of money to get software in the future or open accounts with a specific broker to get it for free. Tradestation started that, look at what is happening with QuoteTracker now. No more for free. Thousands are left wondering whether they should pay or open an account with the broker who owns it after they have invested time to learn how to use it. "We want to let you know that beginning on September 21, 2009 the free version of QuoteTracker you currently use will no longer be available. But you can continue to get FREE access to QuoteTracker by opening a TD AMERITRADE account. Plus, we'll upgrade you to Premium QuoteTracker." This message was received yesterday.