Walk Forward testing can be performed with any suite name that has a system attached.  In other words the suite name doesn't need to be "Walk Forward" in order to use the Out of Sample (OOS) walk forward optimization procedure.

 

Walk Forward testing is a concept that uses a segment of historical data for system parameter optimization and then uses those optimized set of parameters selected using the maximum goodness statistic assigned in Trading Blox Preferences -> Data  Folders and Options -> Multi-Parameter Graph section.  Using the statistic index number assigned to the Measure of Goodness Index as the best performance guide, Walk Forward feature will then test an out of sample segment of data to generate a test result.

 

In the chart shown next the bottom yellow arrow represents an optimization test from 1998 to 2002.  When the system is set to step parameters, the test could be hundreds of stepped tests over this time frame to find the optimal parameters for the system. When the optimization ends the optimal system parameters are used in an out of sample test for 2002 to 2003.

 

because there is more data available the process will begin again with an optimization run from 1999 to 2003, and an out of sample test for 2003 to 2004.

 

Click to enlarge, Click again to reduce

Walk-Forward Step Process Example

Using Walk Forward Optimizations:

Select the Walk Forward Suite, and then select any working system name from the system listing you want to test.  If there isn't a listing for the Walk Forward, then create it.  If the Walk Forward Suite doesn't exist, create it.  Also create a system in the System Editor using the same name after you create the new suite.  When you create the Walk Forward system you will be notified there is a suite by the same name and the system will be considered a Global suite.

Walk Forward Stats 2 Suite

If it does exists and another system has been assigned, remove that system if it isn't the system you wish to optimize.

 

Go into the System Editor and check if the Walk Forward Stats 2 blox is attached to the Walk Forward system.  This blox module will change how a Walk Forward simulation reports its results.

 

Note:

Do not allow the Walk Forward Stats 2 blox to be attached to a standard simulation system as the error checking internal to this blox will cause Trading Blox to halt with error informing the blox is attached in the wrong place.

 

 

With the system selected its parameter section will appear and you can set the important system parameters to Step as necessary.  In this example the Donchian SmartFills - EntryExit Blox has been selected for our optimization test.

Donchian SmartFills EntryExit Stepping Parameters

Enable stepping for the parameters you want to optimize.

 

Checking Preferences:

The optimal run from the optimization runs is determined by the maximum goodness value. Goodness value is a statistic set in Preferences , and can be any of the built in statistics, or custom statistics.  Access to Trading Blox Preference settings is File menu Group section.

 

 

When Preferences settings appear, click on Data Folders and Options so you see this information:

Click to enlarge, Click again to reduce

Walk-Forward Test Selected Performance Statistic

Enter a statistic value if you need to change it.  Then click OK.

 

 

Setting Global Parameters:

In the Global Parameters section these Walk Forward Parameters are available:

Global Parameters Walk Forward Settings v4.3.x

Walk Forward Optimization (days), determines how many calendar days over which to optimize the parameters.

 

Walk Forward OOS (days), determines the OOS (Out of Sample) calendar days over which to use to test the optimized parameters.

 

Check the other Global Parameters listed to ensure they are they represent the values you want applied to this optimization simulation.  Check the Thread Count parameter field to ensure it is a value greater than 1.  Threads will allow the optimization to complete faster.

 

Note:

Saving of open positions from one OOS test to the next has limitation because any variables that might be tracking trailing stops and such are not brought over.

 

 

Creating a Walk Forward Simulation:

Go to the Suite menu and select the option Run Walk  Forward Test:

Run Walk Forward Test

 

Walk Forward Performance Results:

This table shows Summary Performance results from the Walk Forward Simulation:

Click to enlarge, Click again to reduce

Walk-Forward Stats Performance Summary Example

When the above report appears, it also shows the period results of each of the periods shown in the stepped value performance table near the top of the screen.

 


Edit Time: 9/20/2017 07:56:26 AM


Topic ID#: 234

 

Created with Help & Manual 7 and styled with Premium Pack Version 2.80 © by EC Software