Performing Optimization

From MultiCharts
Jump to navigation Jump to search

Once a strategy has been applied to a chart, strategy optimization can be performed.


Running Exhaustive Search

Optimization is performed from the Format Objects window:

To perform Exhaustive Search optimization:

1. Open the Format Objects window.

To open the Format Objects window, right-click on an empty area of the chart and then click Format Signals.

The Format Objects window can also be opened by one of the following methods:
  - Select Format in the main menu and click Signal.
  - Double-click any one of the signal's labels on the chart.
  - Position the mouse pointer over any one of the signal's labels on the chart; once the Pointer Pointer.png changes into a Hand Hand.png, right-click the label and then click Format Signals.

2. Select the Signals tab.
3. Click the Optimize button.
4. In the Choose the Optimization Type dialog box that appears, select Regular Optimization and click Next.
5. In the Set Optimizable Inputs section check/uncheck the check box to the left of the signal name and input name to enable/disable optimization for this input or check the check box to the left of the Signal Name column heading to enable optimization for all inputs (at least one input should be selected to perform optimization).
6. The Current Value column shows input values that are currently selected for the signals applied on the chart.
7. In the Start Value column, enter the desired starting values for each of the inputs.
8. In the End Value column, enter the desired ending values for each of the inputs.
9. In the Step column, enter the desired step size, for each of the inputs.
10. The Step Count column shows current number of steps for an input.
11. Select Exhaustive radio button in the Set Optimization Method section.
12. Select the target function to be optimized in Optimize by section:

In the Maximal/Minimal list, set the respective values to be selected at the end of optimization. From the next list one can select the function itself.
If Custom Criteria is selected, Edit button is available. Press it to write the script for the custom criteria. Read more about Standard and Custom Criteria.
In order to apply the desired function from code select Custom Fitness Value. The custom value shall be specified by SetCustomFitnessValue keyword in the script.

13. Select the Use Limitation checkbox to limit the output to a defined number of the best results; unselect the checkbox to list all the results in the optimization report.
14. When Use Limitation checkbox is checked Optimization dialog window shows current best results for selected сriteria during optimization.
15. Click Optimize N combinations to run the optimization and generate the Optimization Report.


Since MultiCharts 10 that is possible to start optimization from Chart Analysis toolbar (see MultiCharts_Work_Area#Hiding_and_Redisplaying_Toolbars) by clicking on the Optimize Strategy button: Toolbar2 ChartAn 125.png


Running Genetic Algorithm Search

Note: For detailed information on Genetic Algorithm properties, see Understanding Genetic Algorithm Optimization

Optimization is performed from the Format Objects window:

To perform Genetic Algorithm optimization:

1. Open the Format Objects window.

To open the Format Objects window, right-click on an empty area of the chart and then click Format Signals.

The Format Objects window can also be opened by one of the following methods:
  - Select Format in the main menu and click Signal.
  - Double-click any one of the signal's labels on the chart.
  - Position the mouse pointer over any one of the signal's labels on the chart; once the Pointer Pointer.png changes into a Hand Hand.png, right-click the label and then click Format Signals.

2. Select the Signals tab.
3. Click the Optimize button.
4. In the Choose the Optimization Type dialog box that appears, select Regular Optimization and click Next.
5. In the Set Optimizable Inputs section check/uncheck the check box to the left of the signal name and input name to enable/disable optimization for this input or check the check box to the left of the Signal Name column heading to enable optimization for all inputs (at least one input should be selected to perform optimization).
6. The Current Value column shows input values that are currently selected for the signals applied on the chart.
7. In the Start Value column, enter the desired starting values for each of the inputs.
8. In the End Value column, enter the desired ending values for each of the inputs.
9. In the Step column, enter the desired step size, for each of the inputs.
10. The Step Count column shows current number of steps for an input.
11. Select Genetic radio button in the Set Optimization Method section.
12. Selecting the target function to be optimized in the Optimize by: section:

In the Maximal/Minimal list, set the respective individuals, which will take part in further selection and crossover, for the selected parameter. From the next list one can select the function itself.
If Custom Criteria is selected, Edit button is available. Press it to write the script for the custom criteria. Read more about Standard and Custom Criteria.
In order to apply the desired function from code select Custom Fitness Value. The custom value shall be specified by SetCustomFitnessValue keyword in the script.

13. Set Basic Genetic Algorithm Parameters.
14. An optimum population size value is automatically placed next to the Set Population Size box; if a different value is desired, check Set Population Size box and select the value.
15. In the Crossover Probability field, select the desired crossover probability; value range is 0-1, with a default of 0.95.
16. In the Mutation Probability field, select the desired mutation probability; value range is 0-1, with a default of 0.05.
17. In the Convergence Type section, select Number of Generations or Proximal Convergence.
18. If Number of Generations is selected:

In the Number of Individuals in Population for Crossover field, enter the desired number of individuals in population for crossover.
In the Maximum Number of Generations field, enter the desired maximum number of generations.

Note: the above fields are grayed out if “Set Population Size” box is not checked in Basic Genetic Algorithm Parameters.
19. If Proximal Convergence is selected:

In the Number of Individuals in Population for Crossover field, enter the desired number of individuals in population for crossover.
In the Maximum Number of Generations field, enter the desired maximum number of generations.
In the Minimal Number of Generations field, enter the desired minimal number of generations.

Note: the above fields are grayed out if “Set Population Size” box is not checked in Basic Genetic Algorithm Parameters.
20. If Proximal Convergence was selected, enter the desired convergence rate into the respective field. A value, approaching 1 is usually selected for the convergence rate; the default value is 0.99.
21. In the Genetic Algorithm Subtype section, select Basic or Incremental algorithm subtype.
22. If Basic algorithm subtype was selected, select Yes or No for Use Elitism option.
23. If Incremental algorithm subtype was selected, select the Replacement Scheme between Worst, Parent and Random.
24. Click Optimize N combinations to run the optimization and generate the Optimization Report.
25. Optimization dialogue window shows Average fitness value for current population during optimization.


Since MultiCharts 10 that is possible to start optimization from Chart Analysis toolbar (see MultiCharts_Work_Area#Hiding_and_Redisplaying_Toolbars) by clicking on the Optimize Strategy button: Toolbar2 ChartAn 125.png


Custom Criteria

There are two types of criteria: standard and custom.
Standard Criteria lets the user select a criterion from a pre-defined list. For example, if the user selects Net Profit, the optimization will find the parameters that generate the highest Net Profit.
Custom Criteria lets the user create his own criteria. Instead of choosing from a pre-defined list like in Standard Criteria, the user can script his own criteria. E.g., suppose the user wants to find the parameters that maximize the Net Profit / Max Drawdown ratio. This ratio isn't pre-defined under Standard Criteria, so it won't be in the drop-down list. However, this ratio can be created using Custom Criteria.

It is also possible to add additional criteria such as number of trades and winning percentage.

E.g., it is possible to specify: the result have more than 300 trades and a winning percentage that's greater than 50%.


The scripting language is JavaScript.


Code Example

The code below creates the Net Profit / Max Drawdown ratio:

      
if (StrategyPerformance.MaxStrategyDrawDown != 0)       
{       
return StrategyPerformance.NetProfit / (-      
StrategyPerformance.MaxStrategyDrawDown);       
}


Besides NetProfit and MaxStrategyDrawdown, there are many other properties that can be referenced. The table below shows the available properties.
For example, to reference the number of trades, simply use StrategyPerformance.TotalTrades. To reference winning percentage, simply use StrategyPerformance.PercentProfitable.

Name of property in the object StrategyPerformance Description Available in MultiCharts Available in Portfolio Optimization
NetProfit Net Profit Yes Yes
GrossProfit Gross Profit Yes Yes
GrossLoss Gross Loss Yes Yes
TotalTrades Total Trades Yes Yes
PercentProfitable % Profitable Yes Yes
WinningTrades Winning Trades Yes Yes
LosingTrades Losing Trades Yes Yes
AvgTrade Avg Trade Yes No
AvgWinningTrade Avg Winning Trade Yes No
AvgLosingTrade Avg Losing Trade Yes No
WinLossRatio Win/Loss Ratio Yes No
MaxConsecWinners Max Consecutive Winners Yes No
MaxConsecLosers Max Consecutive Losers Yes No
AvgBarsInWinningTrades Avg Bars in Winner Yes No
AvgBarsInLosingTrades Avg Bars in Loser Yes No
MaxStrategyDrawDown Max Intraday Drawdown Yes Yes
ProfitFactor Profit Factor Yes No
ReturnOnAccount Return on Account Yes No
CustomFithessValue Last SetCustomFitnessValue() calculation result.

Will return 0 if SetCustomFitnessValue() was not calculated.

Yes Yes


Pausing optimization

Since MultiCharts 10 that is possible to pause the optimization process in order to temporary decrease the load on the CPU. To pause optimization click Pause in the Optimization Progress window. To resume optimization click Resume.