What effects the odds mostly

[rmania]

Originally Posted by misscashalot

Generally what handicapping factor has the greatest influence on post time odds?

Don't know if this is considered a handicapping factor but that which has the greatest influence on post time odds is the morning line.

[wanderius_thrax]

maybe this will help a little. this is a sample is of 549 races of 6 furlongs or less where the horses on average have 1 or 2 wins. dirt races, various tracks around the u.s. it shows average correlation of various factors to 1) finishing position and 2) the odds

1 = perfect correlation. 1=1 2=2 3=3 4=4 etc
0 = no correlation. 3=6 1=4 9=2 etc.
-1 = perfectly inverse correlation 11=1 10=2 9=3 8=4 7=5 6=6

first number: correlation of ranking to actual finish position
second number: top figure horse actually won race
third number: bottom figure horse actually came in last
fourth figure: correlation of ranking to final odds


odds: .49 196 163 1.0
[read: correlated to the actual real life outcome better than any of the figures below. picked more winners than any of the figures below. picked a lot of last places horses accurately. and, naturally, it correlates perfectly to itself: 1.0]

bris prime: .46 167 165 .81
[read: correlates highly to actual outcomes of races, picked lots of winners. picked 2 more last place finishers than the real odds. and it correlates very strongly to odds: .81]

average speed last 3 races: .41 147 172 .66
morning line: .39 148 135 .79
last out speed fig: .38 162 152 .59
best speed figure out of last 3 races: .36 150 133 .63
second most recent speed fig: .28 126 148 .46
average finishing position (kind of complicated): .26 111 137 .47
top lifetime speed fig: .25 114 117 .41
last out beaten lengths: .25 101 128 .43
average lifetime earnings: .23 106 115 .48
trainer (this is a complicated calculation): .22 133 102 .41
third most recent speed fig: .22 114 106 .34
fps second call previous race: .19 142 84 .31
jockey (complicated again): .19 107 91 .37
average of bris pars (complicated): .17 109 91 .22
number of workouts and races in the past 150 days (most to least): .15 79 116 .21
pedigree: .14 109 83 .24
average purses of recent races (complicated): .13 90 89 .28
horse lifetime win %: .11 102 90 .28
average workout finish position: .10 97 85 .16
claims in past 10 races: .04 72 62 .06
last out purse value: .04 85 84 .12
post position (inside to outside): .02 72 69 .02
alphabetical 0 65 65 0
reverse alphabetical: 0 51 68 0

expected wins due to pure luck: 67.36

[ikeika]

The tote board and the handicappers in print. As far as I know, most people don't feel as though their opinions have any value and look to others. It's hard to go against that 4-5 flashing on the tote board or the newspaper handicapper who says he likes the 4 horse when you like the 3. How many times have you heard "I had that horse last night". Well, from the time they made their selection they then spent the next 24 hours talking themselves off of a winner and onto a loser.

[markgoldie]

Originally Posted by wanderius_thrax

maybe this will help a little. this is a sample is of 549 races of 6 furlongs or less where the horses on average have 1 or 2 wins. dirt races, various tracks around the u.s. it shows average correlation of various factors to 1) finishing position and 2) the odds

1 = perfect correlation. 1=1 2=2 3=3 4=4 etc
0 = no correlation. 3=6 1=4 9=2 etc.
-1 = perfectly inverse correlation 11=1 10=2 9=3 8=4 7=5 6=6

first number: correlation of ranking to actual finish position
second number: top figure horse actually won race
third number: bottom figure horse actually came in last
fourth figure: correlation of ranking to final odds


odds: .49 196 163 1.0
[read: correlated to the actual real life outcome better than any of the figures below. picked more winners than any of the figures below. picked a lot of last places horses accurately. and, naturally, it correlates perfectly to itself: 1.0]

bris prime: .46 167 165 .81
[read: correlates highly to actual outcomes of races, picked lots of winners. picked 2 more last place finishers than the real odds. and it correlates very strongly to odds: .81]

average speed last 3 races: .41 147 172 .66
morning line: .39 148 135 .79
last out speed fig: .38 162 152 .59
best speed figure out of last 3 races: .36 150 133 .63
second most recent speed fig: .28 126 148 .46
average finishing position (kind of complicated): .26 111 137 .47
top lifetime speed fig: .25 114 117 .41
last out beaten lengths: .25 101 128 .43
average lifetime earnings: .23 106 115 .48
trainer (this is a complicated calculation): .22 133 102 .41
third most recent speed fig: .22 114 106 .34
fps second call previous race: .19 142 84 .31
jockey (complicated again): .19 107 91 .37
average of bris pars (complicated): .17 109 91 .22
number of workouts and races in the past 150 days (most to least): .15 79 116 .21
pedigree: .14 109 83 .24
average purses of recent races (complicated): .13 90 89 .28
horse lifetime win %: .11 102 90 .28
average workout finish position: .10 97 85 .16
claims in past 10 races: .04 72 62 .06
last out purse value: .04 85 84 .12
post position (inside to outside): .02 72 69 .02
alphabetical 0 65 65 0
reverse alphabetical: 0 51 68 0

expected wins due to pure luck: 67.36

This is a beautiful piece of work and it certainly answers a good deal of what the original poster was asking.

But now for the "over-the-top-none-of-your-business" question (which I am compelled to ask as designated forum idiot):

You ran ROI numbers on these as well because we don't normally run such projects for the good of our health. And you ran ROI's for different combinations of the factors, probably starting with the highest raw ROI, which my guess might be the actual off-time odds. For example, what is the ROI effect when the favorite is also the Prime Power top choice, etc., etc.

So, anything you'd care to tell us?

[TrifectaMike]

Originally Posted by wanderius_thrax

maybe this will help a little. this is a sample is of 549 races of 6 furlongs or less where the horses on average have 1 or 2 wins. dirt races, various tracks around the u.s. it shows average correlation of various factors to 1) finishing position and 2) the odds

1 = perfect correlation. 1=1 2=2 3=3 4=4 etc
0 = no correlation. 3=6 1=4 9=2 etc.
-1 = perfectly inverse correlation 11=1 10=2 9=3 8=4 7=5 6=6

first number: correlation of ranking to actual finish position
second number: top figure horse actually won race
third number: bottom figure horse actually came in last
fourth figure: correlation of ranking to final odds


odds: .49 196 163 1.0
[read: correlated to the actual real life outcome better than any of the figures below. picked more winners than any of the figures below. picked a lot of last places horses accurately. and, naturally, it correlates perfectly to itself: 1.0]

bris prime: .46 167 165 .81
[read: correlates highly to actual outcomes of races, picked lots of winners. picked 2 more last place finishers than the real odds. and it correlates very strongly to odds: .81]

average speed last 3 races: .41 147 172 .66
morning line: .39 148 135 .79
last out speed fig: .38 162 152 .59
best speed figure out of last 3 races: .36 150 133 .63
second most recent speed fig: .28 126 148 .46
average finishing position (kind of complicated): .26 111 137 .47
top lifetime speed fig: .25 114 117 .41
last out beaten lengths: .25 101 128 .43
average lifetime earnings: .23 106 115 .48
trainer (this is a complicated calculation): .22 133 102 .41
third most recent speed fig: .22 114 106 .34
fps second call previous race: .19 142 84 .31
jockey (complicated again): .19 107 91 .37
average of bris pars (complicated): .17 109 91 .22
number of workouts and races in the past 150 days (most to least): .15 79 116 .21
pedigree: .14 109 83 .24
average purses of recent races (complicated): .13 90 89 .28
horse lifetime win %: .11 102 90 .28
average workout finish position: .10 97 85 .16
claims in past 10 races: .04 72 62 .06
last out purse value: .04 85 84 .12
post position (inside to outside): .02 72 69 .02
alphabetical 0 65 65 0
reverse alphabetical: 0 51 68 0

expected wins due to pure luck: 67.36

In a different thread I wrote the following:
"Avoid methods that are average based: NN's, GA's, etc.

Methods based on averages maximize calibration at the expense of sharpness.
Whereas the goal should be to maximize sharpness subject to calibration.

Each race is a unque event and should be viewd as such in determining probabilities and wagering strategies.

The tote odds are well calibrated pooling schemes. Maximizing sharpness will separate you from the pooling.

Anyone making their own oddsline (probabilities) should strife for a concave probability distribution."

Much of the stats you present from top to bottom attempt to calibrate with Bris Power Rating doing the best job.

The one factor, from those you have listed, that has a possibility of being the basis of maximizizing sharpness is

third most recent speed fig: .22 114 106 .34

Would you like to share the ROI for that stat?

If not I can understand why.

Mike

[Overlay]

Originally Posted by wanderius_thrax

maybe this will help a little. this is a sample is of 549 races of 6 furlongs or less where the horses on average have 1 or 2 wins. dirt races, various tracks around the u.s. it shows average correlation of various factors to 1) finishing position and 2) the odds

1 = perfect correlation. 1=1 2=2 3=3 4=4 etc
0 = no correlation. 3=6 1=4 9=2 etc.
-1 = perfectly inverse correlation 11=1 10=2 9=3 8=4 7=5 6=6

first number: correlation of ranking to actual finish position
second number: top figure horse actually won race
third number: bottom figure horse actually came in last
fourth figure: correlation of ranking to final odds


odds: .49 196 163 1.0
[read: correlated to the actual real life outcome better than any of the figures below. picked more winners than any of the figures below. picked a lot of last places horses accurately. and, naturally, it correlates perfectly to itself: 1.0]

bris prime: .46 167 165 .81
[read: correlates highly to actual outcomes of races, picked lots of winners. picked 2 more last place finishers than the real odds. and it correlates very strongly to odds: .81]

average speed last 3 races: .41 147 172 .66
morning line: .39 148 135 .79
last out speed fig: .38 162 152 .59
best speed figure out of last 3 races: .36 150 133 .63
second most recent speed fig: .28 126 148 .46

So it would appear that, in this relatively small sample of sprints, on a race-in, race-out basis (excluding angles or spot plays that might or might not appear in any given race), figures have a greater correlation with respect to both final odds and actual results than the other handicapping factors examined (not counting the morning line and the public odds themselves).

[Show Me the Wire]

Last race finish position and speed figures earned in last three races, especially on So. Cal racing.

[wanderius_thrax]

bris prime and the odds had the same #1 354 out of 549 times. a $2 bet on all of those would return $554.40, a loss of $152.60. that's not a great angle to pursue! but if you search for say, best avg speed, best fps to the second call, and first or second best recent earnings all in one horse, going off at even money or better, i do recall something like that turning a profit of 15% or so... those kinds of horses don't come along too often though. I'm guessing some of the "big players" wait around for opportunities like that?

i didn't return ROIs because, I didn't want to mislead anybody. But if you want to know which factors had the best rois over this particular sample, they were workouts (about +30% and pedigree about +10%), courtesy of some big winners... So take those with a grain of salt, because "alphabetical" only lost 8%, while reverse alphabetical lost 30%, and neither of them should have anything to do with anything. [i call them "reality check" factors]. With a larger sample those should be guaranteed to converge to an almost identical loss, considering the correlation is 0.

betting the top horse in the odds produces an 18% loss.

the third most recent speed fig loses around 25% if you were to blindly bet the top fig to win. It may be a sharpener to a power fig, but it's certainly not a stand alone blind-bet fig. I don't think any dumbed down fig can be much of a stand alone -- at the very least, a person should always pay attention to trainer changes, days since last race (over 90 days, win percentages start to decline), and in reality try to consider everything else relevant.

[Robert Fischer]

Finish position(and running lines in general with a general preference toward speed) of recent races
+
Class of those races