I didn't put this in the harness forum because I think it deals with handicapping and betting more than the type of racing I have the data from.
With Woodbine Mohawk harness down most of this year I went through the laborious process of going through every line in my computer that I have made and completed a database of about 4800 horses mostly at Mohawk which is now Woodbine Mohawk, some from Woodbine when it was racing. This is over about 3 years. I did make some interesting findings. Nothing that I think is going to bring me over the hump of really beating this game. Just some general gist info that I think some may find interesting.
For reference, when I say as expected I mean they won based off their final odds not my odds. Thus the predicted win probability of a 2-1 shout would be 1/3/1.16. The whole point of this exercise was to determine if the markets are totally efficient or if the public/whales have some errors in their game or of course whether I have a clue myself. Obviously you don't have access to my line and none of us have access to the final odds until the race has gone off but the generalizations were interesting to me nonetheless. Now keep in mind that when I use public line I just have the final odds based on the toteboard not the actual odds. So if a horse went off at 5.9-1 I have him as a 5-1. This needs to be considered when looking at the numbers I will present below.
Horses under 2-1 pretty much won as expected by the public. I can make a horse 6-1 he can go off at 3/5 and he is going to run like a 3/5 shot not a 6-1 shot in general. Look at the roi on favorites from a track and that is pretty much what you are going to get. Did not seem to matter much if I hated the horse and very few were above my line.
Horses I made 25-1 or higher almost never won. Something like 9 out of 1100 of these horse won and I think at least half of the ones that did win were under 10-1 final odds. This group of horses performed well below public expectation even.
Horses I made 15-1 to 20-1 were in general poor performers. They won, just less often than expected.
So all that is super logical. Horses that the public thinks should win, do generally win and horses that look bad generally do not win.
So when I look to see how the public betting factors into everything I eliminate the horses that are over 13-1 my line and under 2-1 final odds. I pretty much know the story on those horses from the info presented above.
So with this remaining database here is what I came up with.
Horses bet to 1/2 or less of my line won about 80% of the time they were expected to. Yes they win often but as a whole they lose a lot more than track take. So while a 6-1 at 8/5 performs fairly well (still going to lose a little less than track take) a 8-1 at 3-1 is going to perform very poorly. They win often but as a group they are poor performers.
Horses bet down over 1/2 of fair value to fair value did a little better but were still big losers. They won about 86% as often as the public expected them to.
So at this point things are looking pretty good. With the exception of favorites winning as the public thinks they should and not as I think they should, over bet horse are not doing well, horses with a small chance of winning are not winning. This should be easy, make a line, limit myself to horses up to 13-1 my line bet value and cash tickets. Easy Game.
Not! There is more to the story and that is the horses that perform the best are the ones that are up to 13-1 my line are above fair value is a start. One big problem is the higher above my line they go the less often they win. Not sure my database is big enough (after all the eliminations to draw a completely accurate picture) but at this point I am finding that once the horse is over double my line the results go the other direction and they start to under perform big time. So a 10-1 at 12-1 might actually be a decent bet, that a 10-1 at 30-1 may not be and a 10-1 at 70-1 is most definitely not going to be a good bet long term. Hard to file in the logic department but that certainly is what the data shows.
So what is the conclusion. This is a tough game. NSS. The best performing group seems to be horses up to 13-1 my line up to double my line. They seem to hold their own (even went through 150 races to get wps payoffs on these horses-I cant get the results from before that at track info-they actually did a little better on the place end, 2nd best was show end and win end was a small loser. Win Place show was probably about break even. Not only that but since I do not have access to final odds, how often will I be betting horses I make 8-1 at 10-1 as the gate rolls and see the horse dip to 6-1 late and how often will I pass on a horse I make 6-1 at 5-1 only to see him win at 8-1.
Nothing here should be that surprising. I have long suggested that the game was very efficient thanks to rebates and whales. This study actually showed me it is a little less efficient than I thought it was,
but seizing a profit sans a strong rebate is a very tough task. That being said there is certainly enough fuel in the conclusions I made to improve my betting. FWIW. the horses that I found under perform on the win end do hit the board, so just because a 10-1 at 30-1 or 15-1 at 25-1 may not be a profitable win play long term, doesn't mean he should not be thrown into the exacta or tri or super. For whatever reason they just seldom win as I have found out the hard way
