Using the WF Analyzer

<< Click to Display Table of Contents >>

Navigation:  Walk-forward Analyzer >

Using the WF Analyzer

The following guide describes how to conduct a walk-forward analysis using the Advanced Tester.

 

Please refer to the screen-shot below for the various settings available to conduct your walk-forward analysis.

 

 

wfa_ui_mainguide

 

1.MT4 Terminal - Select the MT4 terminal you wish to test with. This terminal should be not currently be in use as it will automatically be started by the Advanced Tester.

2.Expert advisor - Specify the Expert Advisor you wish to perform a walk-forward analysis with. If you have selected a valid MT4 terminal, a drop-down list of Expert Advisors available on this terminal will be listed.

3.Presets - Specify the Metatrader 'Presets' file that outlines the parameters that will be optimized. This file should exist in the <MT4 DataFolder>/tester folder. The drop-down list will show all .set files in this folder. You can learn more about presets by referring to the Metatrader manual on optimization (see MT4 > Strategy Tester > Expert Properties > Inputs > Load/Save).

4.Symbol - Specifies the symbol/instrument you wish to test with. The drop-down list will show a list of valid symbols available for your selected terminal.

5.Period - Select the period/time-frame that you wish to test in, for example M1 is the "One-minute" period.

6.Testing model - Select the testing model to determine what sort of data is used during your test*.

7.Sort by - Select the metric by which results are sorted. The selected metric will determine how your strategy results are prioritized and selected as the next candidate in the walk-forward iterative process:

 Balance - The balance of your account. Select this if your priority is to ensure the highest balance in your strategy results.

 Profit factor - The result with the best profit-factor is preferred.

 Expected Pay-off - The best expected pay-off is preferred

 Draw-down - The result with the least draw-down is preferred.

8.Fixed Spread - This is the value of the fixed bid/ask spread that will be used in the back-test. Note that this value will only take effect if the Variable Spread feature is not already activated in the advanced testing parameters.

9.Manually select best results - By enabling this option, you will be asked to select the "best-of-breed" optimization result from a list between iterations. The list shows all the results that occurred in the optimization. The selected result will be used in the following test iteration. If this option is not selected, the best results as denoted by the Sort by field will be selected.

10. Normal/Quick mode - Allows you to select the mode of testing. Normal mode is the usual process of performing a walk-forward analysis. The Quick optimize mode allows you to "re-tune" your strategy on the latest market data.

11. In-sample optimization - The period of days to allocate to the optimization run.

12. Out-of-sample optimization - The period of days to allocate to the out-of-sample test run.

13. Ratio - The ratio to maintain between in and out-of-sample optimization days. A typical value is 25% which implies that for every x days allocated to optimizing your strategy, x/4 days will be used to perform the out-of-sample test (for example, for every 4 days of in-sample, there will be 1 day of out-of-sample testing). By clicking on the link icon, the system will ensure this ratio is maintained when setting in and out-of-sample periods. If you wish to use your own customized settings, the link can be unchecked/disabled.

14. Iterations - The number of walk-forward iterations to conduct.

15. End Date - The final date that the walk-forward analysis will complete. All the iterations' periods will work backward from this date.

16. Testing Parameters - The advanced test parameters to use during the walk-forward analysis.

 

The terminal will need to have all the necessary historical data for the test period loaded before a test can be commenced. During the configuration of your analysis, a 'Status' box will denote the range of test dates that it will require. In order to conduct tick-by-tick back-tests, please make sure you have exported data using Tickstory first.

 

Further reading

Please consult the Metatrader Manual on Optimization for further information on their optimization feature.

 

* If you are using the 'Every Tick' model and you wish to test with 99% modeling accuracy, please make sure to export your test data range using Tickstory first. This will allow you to test with tick data and real spread information.