Program that makes an accurate final odds line

[midnight]

I'm not looking to abandon HTR, just to find something that does the following: makes a fairly accurate final odds line, in other words a line that's close to what the off odds are. I know no program can factor in things like public whim, touts, inside info, etc., but if it was fairly close most of the time, it would be a big plus. The morning line isn't bad, but there are a lot of holes in that (e.g. the 12-1 shots often go off at half of that or three times as much, the 2-1 shots might go off as low as 3/5 or as high as 4-1, etc.).

Any suggestions are appreciated.

[fred4now]

Are you trying to figure what the actual off odds are, or what the actual off odds should be? IMHO getting as close as possible to what they should be and then exploiting the differences is the key to the mint.

[midnight]

As close to what they'll actually be, but anything that gives a reasonable approximation for most horses would be good.

[tomcat]

Why?

[MikeDee]

Midnite
- simple situation - Many times you will see a horse as the off favorite and you go back and look at the data in htr and based on everything you look at there is noway that horse should be the favorite. Or you may see a real strong rk-1 horse and the public has the horse as a 3rd ranked favorite.
The crowd sees someting in these 2 horses where it is strong for one horse on not the other.
Whatever the public sees, it is not reflected in the data. So how can anyone write a program to predict what the crowd is going to do, when whatever the pubiic is using to make its' collective opinion is not in your data that you are using to calculate a win probability and a resultant odds line.

I don't think it is possible to make the line you are looking for

[fred4now]

I'm with Tomcat, just gots to know why?

I think we could help better if we understood what you are trying to accomplish.

[Evan]

I'd have to disagree and say I think it is a doable thing, but I don't think anyones really done it nor would it have much value. Take the ml and an accurate cumulative rating such as the K and weight them together. You could even look at each ML maker at each track historically and instead of using ml, run a projection based on the ML's history and weight this with something like the K. If you were really ambitious, you could try to guage how the public at that track historically bets different connections. Don't think the final result would be too inacurate. But why not just use the ML? Any weighting of the ML based on data factors would just add more correlation with your own handicapping.

[tomcat]

Evan, I kinda agree, for instance when I subract odds from ML, I find the further from the ML the odds get, the less likely the horse will prevail.
I have a cut off in most situations of 15.

[beherethen]

In the old days-I used to get the DRF at 8pm the day before the races and visit the track about 5 days a week. I tended to focus on one track. At any rate, most of the regulars could make pretty good estimates of final odds. So what I'm saying is get a copy of the form-the local newspaper selectors-a weeks previous charts and using availble data, start guessing on the final odds. I mean make a list and the compare it with the actual final odds. Just practice-in a few days you should be able to make workable guesstimates, at least at one track.:)