Automatic Pattern Search V 4.9
1133 Broadway, Suite 706
New York, NY 10010
(212) 656-1467; fax: (212) 656-1467
E-mail: sales@tradingpatterns.com
Tradingpatterns.com
Overall Rating: Good
level: Advanced.
Price: $1,495.00; data not included.
Recommended System: Windows XP; 512 Megs of RAM; 2GHZ Pentium 4 Processor, or compatible. One of the following third-party data sources: QP3, MLDownloader, Telechart, Pinnacle Data, eSignal, QCharts, Metastock, CSI or any other data source that can produce an ASCII file where the columns are separated by spaces.
Automatic Pattern Search (APS) from Tradingpatterns.com is best categorized as a scanning product — although it is unlike any other scanning product. You cannot use it to do classic symbol scanning like searching for overbought/oversold conditions, random line crosses, etc. Instead, you use it to scan for patterns that the software itself determines are relevant. It is a strange cross between black-box strategy software and market scanning software.
Installation and Data: Good
Installing APS is straight forward. You download it from the publisher’s Web site and run a simple installation program that automatically installs in demo mode. To unlock the software you have to exchange some codes with the company. This process is all done via e-mail — no phone number is available on the Web site. You will have to wait for a reply, that could take up to 24 hours. In this case, it took a couple of hours to get a reply.
As with many trading analysis products, APS does not ship with its own data. The product uses plain text files where the columns are separated by spaces. For most software trade data applications the default text export format is data separated by commas. So you will have to make sure you change that default “delimiter” to a space before exporting to APS. For applications that aren’t flexible enough to send data to a text file in the format that APS requires, there is a section on their Web site that provides instructions on how to convert the data.
For this evaluation we used daily CSI data. Their software package was able to automatically write data files in the APS format so we did not have to perform a two-step data conversion. APS can also handle intraday data.
Getting Started: Adequate to good
Once installation is complete you can use the sample data provided to open an existing workspace. You will be presented with a fairly complex screen with no idea what each section is used for. It took about 20 minutes to produce results after clicking the run button and the output screen was hard to understand, sending us to the manual. It took about three hours of research to get a handle on the product.
So what does APS actually do? It scans data files and produces a list of potentially tradable patterns. While what it scans for is a trade secret, you get to control certain key parameters.
Think of the product as a black box that knows thousands of patterns. The product will take your data and simulate trading those patterns, restricting the trades to the trade parameters you specify. It will then output those patterns that match your criteria.
Using ASP: Good
We used S&P futures data for our first search. A T/S file had to be created. T/S stands for “Target/Stop,” which is essentially a file that holds combinations of targets and stops.
Next a new search workspace had to be created. We specified our data file, the T/S file created and then set the other parameters mentioned above. We saved the workspace and then ran the search. Running the search can take a long time. In our case it took more than half of an hour for the specified parameters. This was on only one data file with one T/S combination. You can specify multiple data files with multiple T/S and other parameters in the search workspace and run them all in one shot. A search could take several days to run on just the 22 or so common U.S. futures markets.
The result of our search was a list of hundreds of patterns. It would take a long time to evaluate each pattern in depth to decide if it was truly tradable. The product does provide some rudimentary tools to get you started in the evaluation process. The results screen shows the file name, an internal pattern “index” number, whether the trade would be initiated on the open or close, the percent long and short trades, the total trades, the type of trade (long or short), the target and stop parameter and the last date a trade was taken on this pattern.
For every pattern identified by the program you can perform a rudimentary backtest using either the original data file or the data file in another directory. The output results of the backtest isn’t very sophisticated; you could never use it to determine whether or not to trade the pattern. Also, APS has no charting capability.
Using the Signals: Good to excellent
APS does more than simply generate patterns. It also helps track and trade the most promising ones. However, after you’ve determined which ones to trade on an ongoing basis you can add them to the APS system tracking library.
Patterns added to the system tracking library will automatically update with signals for the next day’s open or close, assuming that your data files are up-to-date as of the close of today. Some signals are conditional and the software will make clear what those conditions are for taking the trade on the open or close. For example, a market may need to close below a certain level the next day before a long trade is taken — APS will note that for you.
Short-term scanning is another way to use APS. It still uses its super-secret database/pattern matching engine but it only looks for patterns present as of today’s close instead of identifying patterns for a range of historical bars. This way, the product is also much faster because it’s only evaluating the end of the data file and not necessarily looking at a range of historical data bars.
Summary
A user of this product will not be able to pick it up and go. They will have to spend time wrestling with the help files and the Web site to figure out how to use it. It can easily generate hundreds of patterns from a single symbol, but the developers need to find a way to manage this volume of information. It needs more options to fine tune the patterns produced and a way to aggregate the same pattern across data files.
If you are looking for new ideas and patterns to trade, this product can give you a good starting point. However, it is easy to be overwhelmed by the sheer volume of candidate patterns that could be generated from a single data file, especially users new to trading. But anyone who has spent any time looking at systems should be able to whittle down the patterns to those that they are best able to trade, as with everything related to trading research, it will just take a lot of grunt work.
Nigel Bahadur runs the research and development group for LBRGroup. He can be reached at
