This article describes how to use ensembles of trading signals within Dakota 3 to build a trading system. Contributing market timing signals could originate from any source e.g. R, BioComp Profit, BioComp Patterns, BioComp Dakota 3. The signal generators included in the ATS Ensemble Signal Generators for Dakota 3 are used in the article.
Step 1. The first step is to add the signals that you want to use to Dakota 3’s list of Securities on the Data tab. Click on the Add Text File toolbar button as pictured below.
Browse to the location of the CSV file that you want to add and select it. A dialog will open that prompts you for the details of the data in the file. Select the field that contains the signal values and then select the Close radio button. You need to nominate one column as the column containing the Close.
The selected signal will appear in Dakota 3’s list of Securities after clicking on the OK button.
Step 2. Add the data series to trade and the signals to the Data to Use list by double clicking on the traded security first and then each of the signal series.
The traded series is the SP futures and appears first in the list. A total of 27 short-term trading signals, 3 intermediate-term and 1 long-term signal have been added, not all of the signals are visible in the image.
Step 3. Modify the Equity Settings. The trading signals are all designed to trade MOC with a 1 trading day delay. No commissions or slippage have been applied for this simulation.
Step 5. Delete the volume series on the Indicators tab. All variables that are listed after the traded securities Close are used as input variables.
In the above image the volume variable has been deleted from the list of variables. There is no need to create a target when using the ensemble signal trader signal generators.
Step 6. Swarm settings. Enter the number of bots to use per swarm and select the Swarm Adapter, Swarm Initializer and Bot Performance add-in to use. Select one of the ATS Ensemble Signal Traders to use as the signal generator. In the image below the ESTAdaptiveVote signal generator has been selected.
The above swarm has been configured to use 15 bots (not pictured).
Step 7. Set the signal generator parameter values and run the swarm.
The equity curve and trade statistics appear below.
- If each of the input signals in the ensemble are identical then there is nothing to be gained. Ideally each input signal should be a bit different to the others. The goal is to take advantage of the ‘Wisdom of the Crowd’ principal.
- The higher the voting thresholds the less time the resulting system will be in the market. This improves some trade statistics, but can reduce the overall profitability of the system.
- This is not the only way trading system signal can be combined. The ATS EST Signal Generators add-in for Dakota 3 includes a number of signal generators based on different concepts.