Light, In the second sample the difference in win rate and roi between the set of those wearing blinkers and those not wearing blinkers is certainly smaller.
FWIW, my interpretation here runs along the following lines:
1. The second sample is restricted to starters whose odds are under 10-1.
2. The first sample contains lots of starters whose odds are greater than 10-1.
3. Therefore, the avg odds of the starters in the second sample are lower than the avg odds of the starters in the first sample.
4. It's been my experience that the lower the odds the harder it is to "move the needle." Think of "the "needle" as a gauge being used to measure win rate and roi when a sample of starters with known characteristics is broken out by some independent variable.
I'm not at all surprised the difference between blinkers false and blinkers true turned out to be smaller in the second sample than in the first sample. Why? Because the avg odds of the starters in the second sample were lower than the avg odds of the starters in the first. (It was harder to move "the needle" in the second sample than it was in the first.)
-jp
.
__________________
Team JCapper: 2011 PAIHL Regular Season ROI Leader after 15 weeks
www.JCapper.com
|