Final Odds vs Morning Line

[traynor]

Has anyone done a serious comparison and contrast between the morning line and race results? I work a lot with incremental and final odds compared to finish position, but I have never paid much attention to morning lines.

However, I have noticed recently that (at certain Canadian tracks at least) entries with morning line figures over a specific amount rarely win. At those tracks, regardless of what the numbers and ratings might be on the entry, long odds tend to indicate a long way from the finish line when the race is over.

Any experience (or opinions) in this area?

[Ray2000]

I've never done a serious study, just a few "take a quick look"

Very track dependent and I can never be sure who's making the line, might be more than one guy or robot at any particular track.

These two posts are 2 and 4 years old, wouldn't put much faith in them

A small sample for wagering the Morning Line Pick by track

http://www.paceadvantage.com/forum/s...ad.php?t=96976


correlation to final odds by tracks

http://www.paceadvantage.com/forum/s...ad.php?t=67920

[Dave Schwartz]

I have done quite a bit but never in harness racing.

My interest has never been as you mentioned in your original topic but, rather, the relevance of using ML to predict final odds. Clearly, there is a high degree of correlation between ML and final odds.

Since Final Odds correlates so strongly with win pct, then it is logical that there would be an ultimate correlation between ML and win pct.

My experience with tbreds, would indicate:

1. Look at races sorted by field size.
2. Graph the hit rates.
3. You should see a serious drop off at half the field plus 1 entry.

Thus, in a race with 8 betting interests, you should see that there is a precipitous drop off after rank of 5. (Ties can create a problem. It is helpful to have a 2nd factor that correlates with the tote to break ties.)

A second way to look at it:
Normalize the hit rates (IVs) or betting value (BVs) to 100%. Look at what happens below precisely 83%.

IOW, the bottom 17% seem to perform horrendously.

[traynor]

Originally Posted by Dave Schwartz

I have done quite a bit but never in harness racing.

My interest has never been as you mentioned in your original topic but, rather, the relevance of using ML to predict final odds. Clearly, there is a high degree of correlation between ML and final odds.

Since Final Odds correlates so strongly with win pct, then it is logical that there would be an ultimate correlation between ML and win pct.

My experience with tbreds, would indicate:

1. Look at races sorted by field size.
2. Graph the hit rates.
3. You should see a serious drop off at half the field plus 1 entry.

Thus, in a race with 8 betting interests, you should see that there is a precipitous drop off after rank of 5. (Ties can create a problem. It is helpful to have a 2nd factor that correlates with the tote to break ties.)

A second way to look at it:
Normalize the hit rates (IVs) or betting value (BVs) to 100%. Look at what happens below precisely 83%.

IOW, the bottom 17% seem to perform horrendously.

Final odds are a mixed bag:
In 8739 Pace Races Average Odds for First were 3.59
In 8739 Pace Races Combined Odds for First were 10.13
In 8739 Pace Races 66.64 % Lower 10.54 % Equal 22.82 % Higher 54.14 % Over 2-1 This 22.79 % Under 1-1 This
In 8739 Pace Races Purse 30.00 % Lower 53.70 % Equal 16.29 % Higher

In 9904 Pace Races Average Odds for Second were 5.73
In 9904 Pace Races Combined Odds for Second were 13.12
In 9904 Pace Races 52.71 % Lower 17.84 % Equal 29.45 % Higher 77.96 % Over 2-1 This 7.16 % Under 1-1 This
In 9904 Pace Races Purse 25.51 % Lower 57.59 % Equal 16.89 % Higher

And on and on, fairly symmetrical on down the line. The sticky part is in the bolded part above. At a few tracks, the win rate for entries going off below even money is very high. Much higher than 22.79. At some tracks an entry going off at 3/1 or better can be (relatively) safely tossed because the bettors at that track rarely let a potential winner go off with all that money on the table. Meaning the final odds for winners is usually (consistently) much lower than the overall average.

I don't really have the time to agonize over individual race odds before wagering. I would like to find some kind of correlation between the morning line and final odds that is dependable enough (and accurate enough) to use for wagering. It seems it may be a worthwhile area to explore.

[traynor]

Originally Posted by Ray2000

I've never done a serious study, just a few "take a quick look"

Very track dependent and I can never be sure who's making the line, might be more than one guy or robot at any particular track.

These two posts are 2 and 4 years old, wouldn't put much faith in them

A small sample for wagering the Morning Line Pick by track

http://www.paceadvantage.com/forum/s...ad.php?t=96976


correlation to final odds by tracks

http://www.paceadvantage.com/forum/s...ad.php?t=67920

I agree wholeheartedly. However, that may be the very thing that makes research in that area profitable. One of my favorite tracks tends to have some VERY tight bettors. Anything over $8 is a rarity. It would really simplify my life if I could find a correlation between the morning line and final odds, at least for the tracks I bet on most. I don't mind the work, as long as the results are profitable.

[traynor]

Interesting stuff, Ray. Thanks!

Once again, layering seems to be a key issue, Lumping too many things together seems to blur (possibly useful) distinctions.

[mrroyboy]

Well I have noticed over many years at east coast tracks like Yonkers and Med horses with ML 4-1 to 8-1 when bet way down to near 2-1 win a large amount of races. I never tracked the ROI's but if I like one of these horses I will often raise my bet. I guess at Harness Racing money does talk.

[traynor]

Originally Posted by mrroyboy

Well I have noticed over many years at east coast tracks like Yonkers and Med horses with ML 4-1 to 8-1 when bet way down to near 2-1 win a large amount of races. I never tracked the ROI's but if I like one of these horses I will often raise my bet. I guess at Harness Racing money does talk.

It seems to be the same situation at Saratoga.

Mar092014-Stga-R01-Trot-PP6|1|2.40| 6-5|0.55
Mar092014-Stga-R02-Trot-PP5|1|4.30| 9-2|0.39
Mar092014-Stga-R03-Trot-PP5|1|5.20| 9-2|0.47
Mar092014-Stga-R04-Trot-PP2|1|4.80| 9-2|0.44
Mar092014-Stga-R05-Pace-PP2|1|5.00| 5-2|0.71
Mar092014-Stga-R06-Trot-PP4|1|18.40| 9-2|1.67
Mar092014-Stga-R07-Pace-PP3|1|2.10| 8-5|0.40
Mar092014-Stga-R08-Pace-PP9|1|4.20| 2-1|0.70
Mar092014-Stga-R09-Pace-PP6|1|4.60| 5-2|0.66
Mar092014-Stga-R10-Pace-PP4|1|2.70| 4-1|0.27
Mar092014-Stga-R11-Pace-PP4|1|3.80| 9-2|0.35
Mar092014-Stga-R12-Trot-PP2|1|10.00| 5-1|0.83

The last figure is the ratio of win mutuel to the morning line odds of the winner.

[traynor]

In 11016 Pace Races:
Average Odds for First were 3.68
Combined Odds for First were 10.25
66.34 % Lower 10.43 % Equal 23.23 % Higher 55.55 % Over 2-1 This 22.00 % Under 1-1 This
Purse 29.01 % Lower 54.99 % Equal 15.99 % Higher

53.01 % Lower 17.46 % Equal 29.54 % Higher 78.15 % Over 2-1 This
Average Odds for Second were 5.73
Combined Odds for Second were 13.166.98 % Under 1-1 This
Purse 24.84 % Lower 58.49 % Equal 16.67 % Higher

Average Odds for Third were 7.32
Combined Odds for Third were 15.26
40.54 % Lower 26.03 % Equal 33.43 % Higher 88.16 % Over 2-1 This 3.25 % Under 1-1 This
Purse 22.57 % Lower 59.37 % Equal 18.06 % Higher

If you are not cashing enough exacta and trifecta wagers, it may be because you are betting too many short priced entries for second and third. Reality suggests the most advantageous approach in exotics may be a reverse dutch, rather than a straight dutch or tight "best to 2nd best and 3rd best."

The entries that place are different than the entries that win, and are less frequently the "2nd best horse in the race" than the "best place horse in the race."

[thaskalos]

I would be very surprised to see a close correlation between the ML and the final odds. Based on my somewhat limited harness-playing experience...the ML is generally TERRIBLE at its assigned task.

[mrroyboy]

ML is terrible partially because the odds makers are limited in what they can do. Too many low ML's is not good for the handle so you will get 2-1 ml on horses which are clearly 3-5

[traynor]

Originally Posted by thaskalos

I would be very surprised to see a close correlation between the ML and the final odds. Based on my somewhat limited harness-playing experience...the ML is generally TERRIBLE at its assigned task.

You may be right. I just started monitoring it. I think the odds are highly useful at a number of tracks, but this is the first time I have seriously looked at morning lines in relation to the final odds. (Sartin's admonition in the dinosaur days that whatever% of winners were in the top 5 whatever morning line not considered).

When I have accumulated enough new races to suggest trends, I will post the information. Right now it is still in the "dump into a file for later use" category for mining when enough (new) data (in an easily mined format) has been collected.

[thaskalos]

Originally Posted by mrroyboy

ML is terrible partially because the odds makers are limited in what they can do. Too many low ML's is not good for the handle so you will get 2-1 ml on horses which are clearly 3-5

It works the other way too.

Horses who are clearly 30/1 shots are listed in the ML at 8/1.

Maybe the line-maker doesn't want to embarrass the connections...

[Dave Schwartz]

In order to look for correlation between ML and final odds, you must first adjust the ML to equal the same number of booking points as the tote pool usually shows at the track.

This also adjusts for scratches.


Also, developing a small regression algorithm to make it more accurate is relatively simple.

If you just try to measure the correlation using the published ML the end result will be disappointing.

[Sea Biscuit]

Originally Posted by traynor

Has anyone done a serious comparison and contrast between the morning line and race results?

I think you are confusing the ML as the track handicappers picks most likely to win the race. Not True. A Morning line is simply a track handicapper's prediction on how the public will wager on a race and NOT his picks which he thinks has the best chance to win the race.

Please read this article by Ed Burgart of how the ML is formulated

http://www.aqha.com/Racing/Content-P...ning-Line.aspx

Ken Middleton Sr is the Morning line maker for Woodbine Standardbreds. He also doubles up as a handicapper and offers his picks in the program for the public at large. You will find that the MLs and his picks are not the same. I am giving a link for Woodbine program for Mar 10. Check it out yourself

http://www.woodbineentertainment.com...10,%202014.pdf

Interestingly in Race 8 he places #7 as the ML favorite and in his picks he does not put this horse in his 1st four picks. The horse wins and pays $7.40. Go figure this one out.