A deadly accurate method to
Spreading the Odds
in StkUltra allows us to create and test the accuracy of
the StkUltra odds line.
The accuracy of the StkUltra odds are controlled by the Odds Spread Variable which is variable 18. It is one of the variables in the Profile of each track. It is track specific, and distance specific within each track's profile.
Use the Power Users Console to change the odds spread variable. Click the purple Power-Users-Console command bar on the Command Central form of StkUltra to start the Power users console. On the Power Users Console, shown below, menu option (5) gives a quick analysis of the of the performance of the loaded profile. Make sure profile for the track you want to work with is active by entering I and the three letter track code for the track.
The console shown here is the Linux console, but it works exactly the same way as the Windows version.
Once you're sure you have the correct profile loaded, push the #5 key on
your keyboard and then push Enter to see the analysis in the console.
Spread set too highAt the bottom of the analysis is the Calculated Probability. In the example shown below, the odds spread variable 18 is set at 20, which is too high.
We know it is too high because the accumulated deviation numbers are all
negative. When the odds are spread exactly right, the accumulated deviation
numbers will be close to zero and will change from positive to negative
going from left to right.
Odds Spread too lowBelow is another screen shot of the Calculated Probability analysis. In this case the odds spread variable is too low. We know it is too low because the accumulated deviation numbers are positive all the way across. The accumulated deviation numbers should be evenly divided between positive and negative and should not deviate very much from zero.
You can change the odds spread variable #18 on that screen by entering 18 to get a Change-it-to prompt. At the Change-it-to prompt, enter the value for variable #18 and push Enter. Repeat this process with the goal to spread the odds as much as possible while maintaining accumulated deviation at an acceptable level.
We are experimenting now with automatic spreading of the odds during the training process. This requires much research, however, and this issue of StkUltra is the tool we are using to do that research.
In the example below, the odds are correctly spread. The numbers go from slightly positive on the left to slightly negative on the right. Increasing the spread more will cause the negative numbers to move further to the left, less will cause the positive numbers to move to the right. We like for the zero crossing point to be about mid way between the 1 and 20 columns.