ATS Bot Performance Add-Ins for Dakota 3

The ATS Bot Performance Add-Ins for BioComp Dakota 3 extends the set of standard bot performance metrics that are installed with Dakota 3. Bot performance metrics measure the performance of individual trade bots within a swarm. The performance measure is used by the swarm adapters when updating the adapted parameter values of the trade bots within a swarm.

The image below is of the ATS Bot Performance Add-ins as they appear within Dakota 3.

ATS Bot Performance Add-Ins for Dakota 3

ATS Bot Performance Add-Ins for Dakota 3

Bot Performance add-ins may have one or more parameters that can be set by the user. All of the ATS Bot Performance add-ins enable the user to specify the performance lookback period. The parameters for the PropOfPerfect (proportion of perfect trading) Bot Performance add-in appear below.

PropOfPerfect Bot Performance Add-In Parameters

PropOfPerfect Bot Performance Add-In Parameters

Each of the Bot Performance add-ins are listed below along with a brief description of how they are calculated.

AvgReturnDivStdDev: The Average trade return divided by the standard deviation of trade returns over the performance lookback period.

AvgTradeReturn. The average trade return over the performance lookback period. Computed using log returns.

EquityStraightness. The signed correlation squared of the equity curve and the equity curve resulting from perfect trading over the performance lookback period. Output range is -1 to 1.

MaxStreakRatio. The number of maximum consecutive wins minus the number of maximum consecutive losses divided by the sum of the maximum consecutive wins and losses over the performance lookback period.

NRSlope. The normalized slope of a straight line fitted to the equity curve over the performance lookback period. Computed using log returns.

NRSlopeMultStraightness. The normalized slope of a straight line fitted to the equity curve multiplied by the unsigned equity curve straightness over the performance lookback period.

ProfitFactor. The gross profit divided by the gross loss. Computed using log returns.

PropOfPerfect. The proportion of perfect trading. Output range is -1 to 1.

PropOfPerfMultStraightness: The proportion of perfect trading multiplied by the unsigned equity straightness.

PropOfPerfMultStreakRatio: The proportion of perfect trading multiplied by the streak ratio. If the proportion of perfect is positive then the streak ratio is the number of maximum consecutive wins divided by the sum of the maximum consecutive wins and losses. If the proportion of perfect is negative then the streak ratio is the number of maximum consecutive losses divided by the sum of the maximum consecutive wins and losses. Output range is -1 to 1.

PropProfitableTrades: The proportion of profitable trades over the performance lookback period.

SmoothedPropOfPerfect: The proportion of perfect trading measured using a smoothed and centered price series over the performance lookback period.

RSlope: The slope of a straight line fitted to the equity curve over the performance lookback period. Computed using log returns multiplied by 100.

RSlopeMultStraightness. The slope of a straight line fitted to the equity curve multiplied by the unsigned equity curve straightness over the performance lookback period.

TotalReturn: The sum of the log returns over the performance lookback period.

ProfitLossRatio: The average profit divided by the average loss. Computed using log returns.

A description of the bot performance metric parameters follows. Not all of the following parameters apply to each of the bot performance add-ins.

Lookback Period: The number of bar to bar measurements used to compute the trade bot performance.

While in Position: Indicates if the periods when the trade bot did not have a position in the market should be included in the calculation of the metric.

Metric: The performance metric can be set to either Direction or PriceChange. If set to Direction then the performance metric is the number of periods that the direction of price change was predicted successfully minus the number of periods that the price change predictions were incorrect all divided by the lookback period. If the Metric is set to PriceChange then the performance metric is the sum of the absolute value of the bar to bar price changes where the bot successfully predicted the direction of price change, minus the sum of the absolute value of the bar to bar prices changes where the bot was not successful, all divided by the sum of the absolute value of the bar to bar price changes over the lookback period.

Weighted: If weighted performance is applied then more recent bars are assigned higher weights in the same fashion that a weighted moving average is calculated.

Kind Regards,

James

Leave a Reply

Your email address will not be published. Required fields are marked *