Re: Estimating Final Odds (983 Views)
Posted by:
Mathcapper (IP Logged)
Date: December 14, 2015 01:05AM
I know it’s tempting to think about this in terms of individuals (like who and how could someone have possibly had such a combination?), but it’s better understood by looking at it from the perspective of the entire whole instead.
There are really two things at play here: 1) crowd wisdom and 2) market efficiency
Consensus estimates of large groups tend to be as good as, and often better than, the estimates of even the smartest individuals within the group (James Surowiecki wrote a pretty cool book on this called “The Wisdom of Crowds”).
What this means in terms of racetrack betting is that the market is pretty efficient. So for instance, if you look at all the horses that went off at 3-1 over history, you’ll find that they win around 25% of the time, just as the public estimated. Same goes for the exotics pools, both vertical and horizontal. A great deal has been written on this topic by academics, much of which has been compiled in a book called “The Efficiency of Racetrack Betting Markets” (highly recommended reading for both quant geeks and insomniacs alike).
Racetrack markets aren’t perfectly efficient of course. Individual results will vary widely, as has been discussed often on this board. And there are some biases. Favorites used to get underbet in the win pool, although that seems to no longer be the case. In the exotics, as Boscar noted, extreme longshots tend to get overbet by the public.
But the wisdom of crowds and market efficiency are the reasons why you see bets like the ones in the 9th at GP occurring with the frequency they do. It’s not so much that particular individuals have figured out how to structure their tickets in such an ingenious way as to hit these kinds of combinations, but rather that the probability (ie. odds) of that combination will tend to reflect the crowd’s opinion on the win probability of that combination.
In this case, based on the win odds of each of the top 4 finishers, using discounted Harville, the public’s win probability estimate for that super was 0.00063%, or about 160,000-1.
So the estimated payout for the $.10 super was around $16,000, or 5 to 6 expected winning tickets. Not surprisingly, given the public’s bias toward extreme longshots in the exotics, it actually came back at over twice as many tickets, or a little less than half the expected payout.
As a side note, it’s unlikely that the CRW’s - at least any of the ones whose systems are designed on betting overlays - had this ticket, unless they had the winner or one of those longshots as being big overlays themselves. For those guys, covering all potential combinations just to ensure all combos are covered is the antithesis of what they’re all about – finding overlays - and because of the public’s longshot bias, such extreme longshot combos like this one are almost always big underlays.
Rocky R