Accurate Odds Line

[arkansasman]

Not that I know of one, but what would you pay for an accurate odds line that rivals the public's odds?

[xfile]

Not that I know of one, but what would you pay for an accurate odds line that rivals the public's odds?

Why pay for one?? Learn how to make your own. :cool:

[Capper Al]

Not that I know of one, but what would you pay for an accurate odds line that rivals the public's odds?

I wrote a spreadsheet that translated an odds line into true probabilities. When running a small sample of morning line odds through the sheet, I was amazed that after one takes out about a 23% mark-up that it reflected the final odds very closely. This supports what many modern odds line makers say; the modern odds maker doesn't try to predict the races anymore but the public instead. And why not? The public has consistently been one of the best handicappers over all times and all tracks.

[mcikey01]

Not that I know of one, but what would you pay for an accurate odds line that rivals the public's odds?

Could you be a little more specific? Its sounds like your asking what one would pay for information with long-run predictive accuarcy, that, if utilized at its optimal investment value, would have you lose money consistently over the long run? :confused:

[tahoesid]

As long as I remember I always thought the morning oddlines were a reflection of the way the public was going to bet, not the odds that the handicapper thought the horses should actually be. It's more of a
"public" handicapping than a horse handicapping.

[sjk]

I think he means a line which is equally accurate as the public line but sufficiently different that it presents frequent overlay opportunities.

I'm not looking to buy or sell but I would think the value would be in the 6 figure range.

[xfile]

For me a re-sort of the ML works well. :cool:

[boxcar]

As long as I remember I always thought the morning oddlines were a reflection of the way the public was going to bet, not the odds that the handicapper thought the horses should actually be. It's more of a
"public" handicapping than a horse handicapping.

Actually, the ML-maker attempts to synthesize the two concepts.

Boxcar

[Lucky Maria]

For me a re-sort of the ML works well. :cool:


I agree. After a lot of frustration I found that by sticking to certain tracks I could simply tweak the morning line and come up with a very solid line.

Someone posted a while back that 50% of the races are won by horses 4-1 or less on the morning line. I use that as one of my parameters to see if a track's line tends to be accurate. It's working well for me.

Maria

[Vegas711]

An accurate odds line in my opinion is a myth. Every race is different , it is inpossible to put an exact number on the probability that a horse has a blank % of a chance that he will win. It is all guessing, sorry but that is the reality of the game.

[Capper Al]

I agree. After a lot of frustration I found that by sticking to certain tracks I could simply tweak the morning line and come up with a very solid line.

Someone posted a while back that 50% of the races are won by horses 4-1 or less on the morning line. I use that as one of my parameters to see if a track's line tends to be accurate. It's working well for me.

Maria

How well was the ML of 4/1 or less tested? It does sound right.

Al

[jeebus1083]

I find this site useful. Saves a lot of work.

http://www.horseracinggold.com/MorningLineTool.htm

[RaceIsClosed]

An accurate odds line in my opinion is a myth. Every race is different , it is inpossible to put an exact number on the probability that a horse has a blank % of a chance that he will win. It is all guessing, sorry but that is the reality of the game.

"I don't have to outrun the cheetah, just you."

[Jeff P]

An accurate odds line in my opinion is a myth. Every race is different , it is inpossible to put an exact number on the probability that a horse has a blank % of a chance that he will win. It is all guessing, sorry but that is the reality of the game. Oh Really??

Because I have this little algorithm. Ok, it's really not so little. In fact, it's quite long. It performs the task of weighing hundreds of factors through thousands of decision points. Eventually, maybe a millisecond or two into a race, it generates an odds line. Not just an odds line but an accurate odds line. At least I THINK it's an accurate odds line. Because every time I check it against a decent sized data sample I see that reality, or rather the probability distribution produced by actual races, falls right in line with the probability distribution produced by the algorithm, and all within a very tiny margin of error.

For example, in my database, looking at all starters at 19 tracks for the first three quarters of 2005, covering about 19,000 races, the odds line probability distribution looks like this:

Data Window Settings:
999 Divisor
Surface: (ALL*) Distance: (All*) (From Index File: D:\2005\Q4_2005\pl_Complete_History.txt)


Data Summary Win Place Show
Mutuel Totals 234017.70 232646.90 227901.90
Bet -307492.00-307492.00-307492.00
Gain -73474.30 -74845.10 -79590.10

Wins 19409 38596 56968
Plays 153746 153746 153746
PCT .1262 .2510 .3705

ROI 0.7611 0.7566 0.7412
Avg Mut 12.06 6.03 4.00


By: Assigned Probability Number

>=Min <Max Gain Bet Roi Wins Plays Pct Impact
-999.00 0.00 0.00 0.00 0.0000 0 0 .0000 0.0000
0.00 0.05 -19666.10 43522.00 0.5481 877 21761 .0403 0.3192
0.05 0.10 -32181.40 112396.00 0.7137 4344 56198 .0773 0.6123
0.10 0.15 -12412.40 68292.00 0.8182 4522 34146 .1324 1.0490
0.15 0.20 -5613.60 40698.00 0.8621 3734 20349 .1835 1.4536
0.20 0.25 -2541.60 22190.00 0.8855 2602 11095 .2345 1.8577
0.25 0.30 -393.50 10822.00 0.9636 1623 5411 .2999 2.3760
0.30 0.35 -497.30 5626.00 0.9116 941 2813 .3345 2.6498
0.35 0.40 -204.70 3754.00 0.9455 716 1877 .3815 3.0217
0.40 0.45 36.20 190.00 1.1905 49 95 .5158 4.0858
0.45 0.50 0.00 0.00 0.0000 0 0 .0000 0.0000
0.50 0.55 0.10 2.00 1.0500 1 1 1.0000 7.9214
0.55 0.60 0.00 0.00 0.0000 0 0 .0000 0.0000
0.60 0.65 0.00 0.00 0.0000 0 0 .0000 0.0000
0.65 0.70 0.00 0.00 0.0000 0 0 .0000 0.0000
0.70 0.75 0.00 0.00 0.0000 0 0 .0000 0.0000
0.75 0.80 0.00 0.00 0.0000 0 0 .0000 0.0000
0.80 0.85 0.00 0.00 0.0000 0 0 .0000 0.0000
0.85 0.90 0.00 0.00 0.0000 0 0 .0000 0.0000
0.90 999999.00 0.00 0.00 0.0000 0 0 .0000 0.0000

How to read the above table: The columns headed min and max describe min and max probability ranges assigned by the algorithm before results of the races became known. The columns headed wins plays and pct show results of actual races run within each min and max probability range after the results became known.

The ones that the algorithm says should win between 0 and 5 percent of the time had an actual win rate of between 0 and 5 percent.

The ones that the algorithm says should win between 5 and 10 percent of the time won between 5 and 10 percent of the time... and so on... right on down the line... in every category... except for the last one... the one with the tiny data sample of only 95 starters.

Question: Do you still think generating an accurate odds line is a myth?

-jp

.

[toetoe]

Jeff,

I was thrown at first, as the bets are positive, and the results negative. I see it now in terms of absolute value, same diff. Some computer exigency, I imagine. Very interesting that you never have a horse with greater than 55% winning probability. That seems low, but it guarantees no chasing 4/5 shots or lower.

[boxcar]

Jeff,Very interesting that you never have a horse with greater than 55% winning probability.

That is a very interesting stat in more ways than one!

Perhaps it's coincidental, but many moons ago I determined "intuitively" that no horse deserved a better than 50% chance of winning a race ("50%" being a very important number in this game). By limiting my win bets (in the vast majority of cases) to horses whose PTO were >=3-1, I assured myself that I would have a sufficient enough "price edge" to beat the game pretty good.

This "50%" fig carried right over to my multi-horse wagering and the minimum ROI that I would accept. My minimum was 100% or double my money. This meant that in a dual-selection scenario, I could still back a horse at odds as low as 3-1 and if he paid exactly $8.00, my minimum would be realized. Of course, there are two ways to compute that return: The ususal way is to compute it on the basis of "markup" (in sales, an amount added to the cost price to determine the selling price). But another way of doing is to markup the "selling price" to get a true return. In a two-horse win pari-mutuel framework, one investment unit would be $4.00 for both horses in the race, therefore, $4.00/.50 = 8.00 (or 3-1) minimum price/odds required. In a three-horse win scenario, the formula would be one $6.00 betting unit for all three horses, therefore, $6.00/.50 = $12.00 (or 5-1) miniimum price/odds required.

It was truly uncanny at just how big of a role this "magic number" frequently played in my handicapping -- even in computing a horse's consistency, etc.

Boxcar

[Murph]

Why pay for one?? Learn how to make your own. :cool:

You HAVE MADE your own odds line as soon as you bet against the favorite, or with the favorite infact. Just understand that the point of odds lines is when YOU bet, OR decide NOT to bet.

Paying for an ODDS line is out of the question.

Does that make ANY sense to ANY handicappers?
Murph

[boxcar]

You HAVE MADE your own odds line as soon as you bet against the favorite, or with the favorite infact. Just understand that the point of odds lines is when YOU bet, OR decide NOT to bet.

Paying for an ODDS line is out of the question.

Does that make ANY sense to ANY handicappers?
Murph

Yeah, makes sense to me, especially since I think the public's line is pretty darn good. And best of all -- it's FREE!. :D

Boxcar

[Jeff P]

The really intersting thing to me odds lines is this:

For my own oddsline, I use an algorithm that weighs a few hundred factors and runs them through a few thousand decision points. Took me years to develop and I'm quite proud of it.

You'd think that might be pretty good, right? Wrong. My odds line pales by comparison to the public's odds line.

If you think about it, the public's odds line is actually created by a type of super-computer. The public's odds line is constantly updated as people collectively make bets. There's what? Maybe 150,000 hard core active horseplayers across North America each day? (Someone please correct me and supply the real number here if you know it.) And all of them are doing something slightly ever so different in their approach than what the guy next to them is doing. I'm going to guess that each of those players is using maybe a dozen different factors with maybe 25-30 unique decision points in their handicapping as each of them decides what to bet. Multiply it out. 25 decision points per player. 150,000 players. 25 x 150,000 = 3.75 million decision points.

So basically, even though I might think I use a pretty sophisticated algorithm to create a pretty good line, the public's line uses about a thousand times more decision points than I do. It's no wonder the public is such a difficult opponent to overcome.

Food for thought.

-jp

.

[GameTheory]

The really intersting thing to me odds lines is this:

For my own oddsline, I use an algorithm that weighs a few hundred factors and runs them through a few thousand decision points. Took me years to develop and I'm quite proud of it.

You'd think that might be pretty good, right? Wrong. My odds line pales by comparison to the public's odds line.

If you think about it, the public's odds line is actually created by a type of super-computer. The public's odds line is constantly updated as people collectively make bets. There's what? Maybe 150,000 hard core active horseplayers across North America each day? (Someone please correct me and supply the real number here if you know it.) And all of them are doing something slightly ever so different in their approach than what the guy next to them is doing. I'm going to guess that each of those players is using maybe a dozen different factors with maybe 25-30 unique decision points in their handicapping as each of them decides what to bet. Multiply it out. 25 decision points per player. 150,000 players. 25 x 150,000 = 3.75 million decision points.

So basically, even though I might think I use a pretty sophisticated algorithm to create a pretty good line, the public's line uses about a thousand times more decision points than I do. It's no wonder the public is such a difficult opponent to overcome.

Food for thought.

You certainly don't have 150k players betting every card, else the pools would be much larger. So for a typical individual race, you have to lower your estimate considerably.

The key is not to "beat" or "overcome" the public, but simply be different while remaining unbiased...

[toetoe]

I'm no mathematician, but I think a sort of regression to the mean is happening in any pool of decent size. Also, as a firm believer in Schumacher's idea that everything has an optimal size, I wonder whether there's an optimal range for pool size. Hard to imagine one being too big. Anyway, here's to Olympic-sized pools. :D

[Jeff P]

GT,

You wrote:
You certainly don't have 150k players betting every card, else the pools would be much larger. So for a typical individual race, you have to lower your estimate considerably.Probably true. Didn't think it through in those terms. A typical Saturday win pool at a NYRA or SOCAL track might have what? 250k or so in it? So what's the actual head count of players looking at that race? 40,000? If so, 40,000 pairs of eyes times 25 decision points equates to about 1.1 million collective decision points being reached as that race is bet. Even if I'm still overestimating the head count and/or the number of decision points reached per player it still comes out to way more decision points than any algorithm I plan to code out in my lifetime.



also:
The key is not to "beat" or "overcome" the public, but simply be different while remaining unbiased...I'm in total agreement here.

-jp

.

[Handiman]

Maybe I'm missing something here. But I don't believe the crowds odds line is all that good. If it were true, there would be alot less longshots and mid-odds horses finishing in the money and winning races. The reason that the crowd puts favorites in the winner circle is because the horse's PP's are so glaringly crystal clear. That being said, they still only get the right horse across the line first 33% of the time.

Then someone will blurt out, the top three favs win 67% of the races. Well, if you consider how many races have only 6 runner's or maybe 7, that isn't all that surprising. I"m a believer that making your own line is a good thing, and can put you ahead of the game. Boxcar...I like your thoughts about the 50% probability and demanding that your plays go off at 3-1 or better. One thing that kills players, is lack of will to stick to their plan.

I have found that when you make a line, and you have a horse as your top pick which is so much better than the other runners, it's best to skip the race if that horse is an underlay, rather than bet the other horses just because they seem to be overlay. An overlay only exists if the horse truly has a chance to win.

I could be wrong......But it works for me.

Handi

[Vegas711]

I go back to the old Sartin days, I believe that you either play your top pick or pass the race. If i am getting only 2-1 on my top pick who I think is going to win, I am not going to bet another horse just becouse he has value.

Sartin once said value is the money you get when you cash a winning ticket.
THERE IS NO VALUE IN BUYING A LOSING TICKET!!!.


As for the smart aleck response I recieved for my honest answer, I may not be a cheetah but I also am not a dead carcass. I doubt very very much that you made a damn dime from my play becouse the past 5 years I lost a total of $125.00 , that is less than $25 per year. That is with NO handicapping before I go to the races and no more than 5 minutes handicapping before I place a bet. One winning day of $125 and I am back to even steven.
Enjoy the slim pickens.

[Vegas711]

"I don't have to outrun the cheetah, just you."


This is an Example of someone who is RUDE, if you disagree with a post the PROPER thing to do is offer a counter opinion, not an attack on someone.

People who attack others expecially when they do not know anything about them show their true ignorance. Handicapping one on one I can hold my own with anyone if I wish to put in the time. Years ago when i used to put in 8 to 10 hours a day studying this game I made decent money( 2 to 3000 a month), but nothing close to the hourly average I make with my own business.Horse players I guesse if you look at all the time they put in maybe make 15 to $20 per hour which is OK but you can do a lot better else where.

This Game in my opinion has 2 directions:

1. You go ALL out make it a full time job and hopefully reap the benefits.

2. You make it a part time hobby with the first priority to minnimize your expences and second priority to seek a small profit.


In 20 years of experience I never meet anyone who makes large profits without putting in a large amount of time. Making large amounts of money has a time cost.

[PaceAdvantage]

I didn't look at it as an attack. I looked at it sort of as a Chinese proverb....methinks you are reading too much into this...

Also, I disagree with you about your theory on value betting, but that's just me.

[dav4463]

[QUOTE=Vegas711]I go back to the old Sartin days, I believe that you either play your top pick or pass the race. If i am getting only 2-1 on my top pick who I think is going to win, I am not going to bet another horse just becouse he has value.

Sartin once said value is the money you get when you cash a winning ticket.
THERE IS NO VALUE IN BUYING A LOSING TICKET!!!.

[dav4463]

Couldn't get the quote to work!

Just had to disagree with the top choice only concept. If my records show that my second choice wins just as often or often enough at a higher price, it would be crazy not to run with it. Do you keep records on your second, third, or fourth choice?

Most of the time, I don't have a strong enough opinion to key one horse as my choice. When I do have just one choice that I really like, more often than not my second, third, or fourth choice will win and at much better odds. I have found that it is better (at least for me) to do less handicapping and more value searching because I am not able to narrow a race down to one choice.

[socantra]

I go back to the old Sartin days, I believe that you either play your top pick or pass the race. If i am getting only 2-1 on my top pick who I think is going to win, I am not going to bet another horse just becouse he has value.

Sartin once said value is the money you get when you cash a winning ticket.
THERE IS NO VALUE IN BUYING A LOSING TICKET!!!.

You have to really go back to find a time when Howard Sartin advocated one horse betting. He did at one time advocate betting your top TWO choices, but in later years become a staunch proponent of a form of value betting.

As with most things, his value betting was contrarian and not like that of the "value boys", who he continued to deride.

socantra...

[garyoz]

I go back to the old Sartin days, I believe that you either play your top pick or pass the race. If i am getting only 2-1 on my top pick who I think is going to win, I am not going to bet another horse just becouse he has value.

Sartin once said value is the money you get when you cash a winning ticket.
THERE IS NO VALUE IN BUYING A LOSING TICKET!!!.

This is a pretty retro perspective. Each race represents an array of statistical probabilities relative to outcomes. Handicapping processes represent an attempt to order those statistical probabilities. To become fixed on one outcome (e.g., a single horse) has the opportunity cost of missing value opportunities and almost certainly getting eaten alive by the takeout. In the long run, there is much better value in making the overlay bet than taking underlays. I'll take ROI over win percentage anyday.

[Tom]

Yeah, makes sense to me, especially since I think the public's line is pretty darn good. And best of all -- it's FREE!. :D

Boxcar

"The Wisdom of Crowds"

http://www.cbsnews.com/stories/1998/07/09/sunday/main13562.shtml

Interesting piece on TV this morning about the power of the crowd in making estimates. They used horse racing as one example. Looks like a book I want to read.

[Secretariat]

In the long run, there is much better value in making the overlay bet than taking underlays. I'll take ROI over win percentage anyday.

In a perfect world, I would agree, but if your linemaking capabilties tend to create too high an odds for winning horses then underlays might actually be better.

For example, if one used the Top ML choice, and looked at whether underlays or overlays are better picks with it, one would often find at many tracks that underlays outperform overlays, and in certain subsets noticeably such as Starter races, and at tracks where Allowance races are predictable.

The problem I have with lines is they are contingent on you being correct about a large group of horses. It's kind of like picking the exact order of finish in each race with all the horses.

While I appreciate the hypothesis of "value" betting witjh overlays, "value" betting is contingent on a line makers actual ability, and how deep one goes down that line. For example, you may find your top two horses win 53% of races, but this is based on long term statistics. (You may go broke while waiting for the long term.) I prefer sticking to my top two picks, and setting a line based off this 53% However, using this I've found some situations where underlays win more often than with my crude linemaking attempts. I've tried tweaking the lines of those races, but have sometimes given up, and went with the flow, playing underlays for those special situations. Knowing when you're not effective in assessing lines is as much about "value" betting as always playing overlays. If the "value" is in the underlay, don't stick dogmatically to what theory tells me I should do.

[JackS]

I think the public is probably right about half the time. Their average of 33% represents their hit rate. the other 17% represents the unfortunate events that predict that the "best" horse cannot win in every race or even every other race.
Since everyone is aware of the "false favorite", why must this lone horse be singled out for false favorite status when the 2nd ,3rd 4th favorites are too often accepted at board odds?
If you can discount one of these other horses, you may have a fair or even more than fair bet on the public favorite.