Future Robot2 feature: Longshot "Scores"

[SteveS]

Probably too late this time around, but referring to the NOV 06 article on Longshots - where you establlished a "SCORE", related to various HTR factors - Is this something worthy of a Robot2 inclusion??

SteveS

[km]

SteveS=

We did have a "Longshot Score" in the old Robot for awhile. Frankly, it was useless and somewhat of a head scratcher, didn't provide much help for forward prediction. With the newer Robot expanded so much it was really a matter of users learning to interpret the data themselves rather than providing a simple output that tries to roll ROI and I.V. into a single number.

For example, suppose these are results for two spot plays I'm testing

1)
Win 28%
IV 2.25
ROI 1.05
Score = 085

2)
Win 7%
IV 0.75
ROI 1.70
Score = 113

The "Score" for (2) is much higher with that big 1.70 ROI, but it's chances of going forward on face value are very limited and we need to look at other factors on the Learn All report to determine if it is feasible to use it, such as: is it a result of a few lucky bombs? Play (1) would have a lower "Score" but there is already good evidence that it is far more likely to carry forward and looks like a workable play to avoid losing streaks and downside ROI.

So the "Score" idea can throw us off because it is too basic and does not get at the heart of the real issue with data: how likely is it to go forward with the same result ? The 1.70 ROI already got our attention, we need to do some scrutiny of the output to determine if it is a fluke.