Quote:
Originally Posted by o_crunk
Yes check the data. Do something simple like sample the last 1200 horses who are racing, say, lasix free (or some such other low value move). You will certainly *not* find 300 favorites in the last 1200 instances of that such move or many other popular moves. I did not find this to be particularly abnormal - there is only one favorite in the race. There are many starters in those races. If you think breaking through the gate is random, there is probably a range of scenarios where only 11% of favorites make up the sample, particularly when you factor in the average field size for the sample is closer to 8 than it is 7 (remember the query goes back 10 years).
|
Why mention the irrelevant bullshit?
Racing "lasix free" is not a "move" in and of itself.
(and then include a data set listing twelve runners who were "declared non-starters") (it is a foregone conclusion that almost no runners who are "declared non-starters" have much hope of
winning those races)
Runners "breaking through the gate" are relatively random, determined mostly by chance.
The scratching of such runners leans significantly toward the outsiders, because the mutuel department can do away with them with relatively low cost to the handle.
Ergo, the instances of break-throughs who went on to run in those races would lean toward a higher percentage of favorite than the random norm. (some allowances must be made for some occasions when bettors do have time to identify the culprit AND cancel tickets - perhaps even rendering SOME of those original favorites no longer the 'favorite' at off time).
So the study was hundreds of races, where with your use of "8" for average field size, it should have landed randomly at or even near to 12.5% favorites... and yet the culprits in the study couldn't even reach 10% favorites, despite human factors (the mutuel department having a stake in what occurs) likely to make the favorite percentage even higher than pure randomness.
The data is most likely flawed for reasons which aren't disclosed by the study's author.