Odds again

[Handiman]

I know there is nothing new under the sun, but I'm having a tough go. I have my ratings pretty much right where I want them. But, the numbers are all fairly close, so I can't seem to figure out how to make a good odds line. That's not the same as I don't know how to make an oddsline.

I'm just having a heck of time developing a good one. I did not do the MC Sim all that well. For lack of anything else, I may have to go back and use Mitchel's advice and formula.

Any input would be appreciated.

Handi :confused:

:bang:

[cj]

Can you post a sample of your ratings for one race? How many ratings per horse are you using? Is there a length or time scale to your ratings?

[formula_2002]

I know there is nothing new under the sun, but I'm having a tough go. I have my ratings pretty much right where I want them. But, the numbers are all fairly close, so I can't seem to figure out how to make a good odds line. That's not the same as I don't know how to make an oddsline.

I'm just having a heck of time developing a good one. I did not do the MC Sim all that well. For lack of anything else, I may have to go back and use Mitchel's advice and formula.

Any input would be appreciated.

Handi :confused:

:bang:

Suggestion, if the numbers are close, play anything>=5-1
If you have some past races, post the date, track and race # and rating(after scratches). If I can match my data base results with your races, I'll do some checking for you. I have great number of track for this past month.
Closely related figures offer a smaller deviation and may (if accurate) just be chaotic enough to show good results on longer odds. (>=5-1 seem to work for my stuff).
Ratings dont mean much without emperical review.

[Handiman]

Ok....Here's Aqueduct 11-24-06

First number is Saddle cloth number
Second is M/L

disregard the 3rd number
Last number is rating.


AQU "20061124" Race 1 "Clm30000NW2L" 8F D
4 5 BREDWINNER 190
8 4.5 24 THATS OUR POINT 181
3 3 STORY LINE 178
1 8 BIG CHIEF BUBBA 175
6 20 KILLER SPEED 166
2 12 15 MEESES TWO PIECES 162


AQU "20061124" Race 2 "MC50000" 5.5F D
2 3.5 15 AD ME IN 168
10 4 LEADFOOT LIZZIE 153
3 8 KULIA 152
6 6 26 AD MUSTER 150
4 3 FLIGHTOFTHEPELICAN 148
8 15 EXPECT DRAMA 128
7 20 DEVILISH MAX 52
5 10 19 GILDED MEMORY 0
1 15 I'M TOP SHELF 0

AQU "20061124" Race 3 "MC25000" 8F D
1 1.6 18 OUT ON TOP 161
6 10 DEACON DAN 145
1A 1.6 18 LOUIE'S VERDICT 140
3 5 MASTERMIND 137
5 12 17 NA'LU 133
2 30 A HELLUVA PITCH 123

AQU "20061124" Race 4 "MC50000" 5.5F D
10 6 KARAKORMKIPNITREAL 143
5 1.6 ZITTA 136
7 20 LOVACIOUS 117
2 12 SIEMPRE VERDE 115
3 15 DARLING DANCER 81
8 20 CALAMITY JENN 52
9 15 BROOKLYN BETTY 47
6 6 JUST ZIP IT 0
1 4 LOTTIE ZIP 0

AQU "20061124" Race 5 "Clm25000" 6.5F D
7 3 BIG BLIND 170
1A 2 15 MISS PARK PLACE 168
8 3.5 ZELNA J 165
3 5 16 SIGN POST 161
4 6 DARBY HAVEN 161
6 20 VESUVIUS 155
2 30 ALLISON'S APPROVAL 149
5 15 LOVE MY GAL 148

AQU "20061124" Race 6 "Mdn42000" 9F D
3 3 18 ALLEY SINGER 162
5 5 CRUSADER'S CODE 160
7 5 MY SONG TO YOU 151
2 4 CARRS AND STRIPES 149
8 8 THE KING'S RIGHT 146
9 12 JENNYS ON RADIO 144
2B 4 WORTH A SHOT 137
4 15 SWEET HITCH HIKER 130
6 15 SMALL POTATOES 96
1A 6 PLAYINGINTHEBAND 0

AQU "20061124" Race 7 "Clm15000NW2L" 8F D
6 2 ZARA'S STORM 165
8 8 17 KOBA 165
5 10 SIX GUN 164
3 2.5 LUCKY STRAIGHT 158
9 15 CHIEF OSCEOLA 148
2 12 LITTLE MOISHE 143
7 20 SPEEDY FREDDIE 142
10 20 PATTON'S HEART 139
1 5 DIDTHETEST 115

AQU "20061124" Race 8 "TopFlgtH-G2" 8F D
5 2.5 24 MO CUISHLE 222
9 4 26 CARMANDIA 220
3 3 MALIBU MINT 215

4 8 17 MISS SHOP 212
8 15 19 RAHYS' APPEAL 206
1 6 17 YOLANDA B. TOO 200

AQU "20061124" Race 9 "OC30000NW2X" 7F D

5 8 WHO WHAT WIN 204
10 4 HE'S A LUMBERJACK 201
7 10 D MONEY 200
9 5 24 SURFING IAN 197
2 3 15 WHISTLIN' 193
12 30 HIMMARSHEE 189
1 20 BOYSIK 180
11 20 PRINCE OF PEACE 179
4 30 JUST IN FUN 164
8 20 GOOD GOING DARL 135
13 50 OF ALL TIMES 5


Thanks,
Handi :)

[cj]

I'm about done for the night, but I'll post a method tomorrow that should work with these pretty well. You will also be able to tinker with it to get it to your liking.

[formula_2002]

I'm throwing out the maidens.
Their deviation were , 40,13,37,22.

Non maiden deviation were 10,7,14,8,21
By my estimation, these may be on the high side.

As it turns out, the 2nd lowest deviation (8) race winner won at 23-1.
as I said
Closely related figures offer a smaller deviation and may (if accurate) just be chaotic enough to show good results on longer odds. (>=5-1 seem to work for my stuff).

Pure luck on my part :jump:

Prime power figures can get down to a deviation of 3!! and seem to show good returns when >=5-1.
Long term result are still open to question.

When you have 20,000 or so horse figures (after scratches) and have the results, you can start to determine the accuracy of your figures.
Accurate figures make accurate odds line, but determining the accuracy of the figures requires emperical data and good analysis.


PS. remember, I'm not a player..the data keeps me away from betting ;)

[Handiman]

Thanks CJ. Form..How did you come up with the dev numbers? Did you do devs for every race? It looks like it, but just want to be sure.


I wrote some code that lets me develop a dev each race, based on variance.

Anyway, thanks for the help you guys.

Thanks to Dave S. too for some guidance. This is truly a great board.

Handi :)

[PriceAnProbability]

When you have 20,000 or so horse figures (after scratches) and have the results, you can start to determine the accuracy of your figures.

And while you are accumualting that data (which you then have to test going forward), how much are you paying in living expenses? Then you have to hope that the system holds up again after that. Very expensive research.

Betting into a rating system is the fastest way to test it, even if the bets are small. Your pocket will tell you pretty quickly where you stand.

[PriceAnProbability]

I know there is nothing new under the sun, but I'm having a tough go. I have my ratings pretty much right where I want them. But, the numbers are all fairly close, so I can't seem to figure out how to make a good odds line. That's not the same as I don't know how to make an oddsline.

I'm just having a heck of time developing a good one. I did not do the MC Sim all that well. For lack of anything else, I may have to go back and use Mitchel's advice and formula.

Any input would be appreciated.

Handi :confused:

:bang:

If you have your ratings where you want them, why aren't your results following suit?

I don't have a book deal, so I have to rely on betting for money from my method, which is why I keep it to myself. There just isn't enough money in giving this stuff out on a message board, so your quest might be hampered by that. Most pros who post here never part with what makes them pros.

[Handiman]

Because I don't have a way of knowing value. I need a way to decipher whether to bet top horse, 3rd horse, 2nd horse, maybe even 4th horse. If I had 1,000 races and could log at what rate my top 4 horses were hitting, then I'd have a better idea of what their odds should be...unfortunately I don't have the database to work with, so my quest continues.


Handi

[PriceAnProbability]

Because I don't have a way of knowing value. I need a way to decipher whether to bet top horse, 3rd horse, 2nd horse, maybe even 4th horse. If I had 1,000 races and could log at what rate my top 4 horses were hitting, then I'd have a better idea of what their odds should be...unfortunately I don't have the database to work with, so my quest continues.


Handi

Your problem isn't the database, it's that your ratings aren't profitable. If they were, you wouldn't need a database but just to cash in on them. You're not cashing in, which means you need to tweak your ratings.

If you want a simple approach, rank your horses in order of rating, compare it to the rank in order of morning line, and the first time you have a horse ranked higher than the program, put it on top.

[formula_2002]

And while you are accumualting that data (which you then have to test going forward), how much are you paying in living expenses? Then you have to hope that the system holds up again after that. Very expensive research.

Betting into a rating system is the fastest way to test it, even if the bets are small. Your pocket will tell you pretty quickly where you stand.

pnp, "Very expensive researh"
MUCH LESS THAN MYBETTING. :)

[formula_2002]

.How did you come up with the dev numbers?

For each race.
One thing I did not mention, not only do I skip Maidens, but I would suggest studying only races in which every horse has a rating.

[Overlay]

I know there is nothing new under the sun, but I'm having a tough go. I have my ratings pretty much right where I want them. But, the numbers are all fairly close, so I can't seem to figure out how to make a good odds line.

You might also try looking at the probabilities associated with the relative positions of horses in their fields based on their ratings (such as first, second, third, front half, rear half).

[PriceAnProbability]

For each race.
One thing I did not mention, not only do I skip Maidens, but I would suggest studying only races in which every horse has a rating.

You can rate first-timers just as easily as those who have run before if you understand a few things.

What are they? Yeah, let me give up the stuff that let's me single a 3-1 shot in a 12-horse field of debut juveniles at Saratoga...

I hate threads like this because I know the answer to every last question he is asking, but publishing any of it would kill prices for all of us.

[kingfin66]

You can rate first-timers just as easily as those who have run before if you understand a few things.

What are they? Yeah, let me give up the stuff that let's me single a 3-1 shot in a 12-horse field of debut juveniles at Saratoga...

I hate threads like this because I know the answer to every last question he is asking, but publishing any of it would kill prices for all of us.

Pace Advantage, I have been meaning to ask you; where did you find this PriceAnProbability character? :ThmbDown:

[PaceAdvantage]

Pace Advantage, I have been meaning to ask you; where did you find this PriceAnProbability character? :ThmbDown:

Me? I don't find anyone....they just show up....

[Tom]

You can rate first-timers just as easily as those who have run before if you understand a few things.

What are they? Yeah, let me give up the stuff that let's me single a 3-1 shot in a 12-horse field of debut juveniles at Saratoga...

I hate threads like this because I know the answer to every last question he is asking, but publishing any of it would kill prices for all of us.

That's funny, I was just thinking how you have not been in tune with one post yet in this thread. :lol:
You don't hold a clue as to what he is talking about.

[PaceAdvantage]

I guess I've missed it with this poster.....I probably skimmed when I should have read more thoroughly....what have I missed?

[cj]

AQU "20061124" Race 8 "TopFlgtH-G2" 8F D
5 2.5 24 MO CUISHLE 222
9 4 26 CARMANDIA 220
3 3 MALIBU MINT 215
4 8 17 MISS SHOP 212
8 15 19 RAHYS' APPEAL 206
1 6 17 YOLANDA B. TOO 200

OK, here goes. What I am going to propose can be easily be programmed into a spreadsheet. You could use a Monte Carlo sim if you want, which I could also do, but let's start here.

Looking at your figures, I'm guessing a cutoff of about 16 below the top is a contender. I haven't done a line yet for this race, so I'm shooting from the hip here. Anyone more than 16 below the top is eliminated. In this race, that is only #1 Yolanda B.

Judging by your figures as well, I'm going to use a base of 5 points. You could tinker with the 16 and 5, but they are pretty solid starting points. To get a new rating for each contender, rate the lowest a 5, then build the rest from there. Rahy's Appeal is lowest, so she becomes a 5. The rest will be the difference between that horse and Rahy's Appeal, +5. Here is what we have:

5 MO CUISHLE 222 - 206 = 16 + 5 = 21
9 CARMANDIA 220 - 206 = 14 + 5 = 19
3 MALIBU MINT 215 - 206 = 9 + 5 = 14
4 MISS SHOP 212 - 206 = 6 + 5 = 11
8 RAHYS' APPEAL 206 - 206 = 0 + 5 = 5

Add up the new ratings to get a total: 21 + 19 + 14 + 11 + 5 = 70.

I then figure out the percentage I think my contenders will win. I split the difference between the percentage of horse I have tabbed against field size and being perfect, 100%. In this case, I have 5 of the 6 horses, so I have a natural chance of 83%. I add 100% and divide by two, to get approximately 92%. If I had only 5 horses in a 10 horse field, I would take 50% + 100% and divide by two, and assume one of my contenders will win 75% of the time.

I take this percentage, and divide the total above by it. In this case, it would be 70 / .92 = 76. I then divide each horse's total into this new total to get the expected winning percentage:

5 MO CUISHLE 21 / 76 = .28
9 CARMANDIA 19 / 76 = .25
3 MALIBU MINT 14 / 76 = .18
4 MISS SHOP 11 / 76 = .14
8 RAHYS' APPEAL 5 / 76 = .07

You then convert this expected win % to fair odds by taking 1 / win% - 1:

5 MO CUISHLE 1 / .28 - 1 = 2.6 to 1
9 CARMANDIA 1 / .25 - 1 = 3 to 1
3 MALIBU MINT 1 / .18 - 1 = 4.6 to 1
4 MISS SHOP 1 / .14 = 6.1 to 1
8 RAHYS' APPEAL 1 / .07 = 13.3 to 1

So, those are fair odds. You can build in an overlaid percentage if you like to make a betting line instead of a fair odds line. Some people want 20%, some like 50%, some 100%, etc.

Anyway, that should get you started. Of course, the ratings have to be solid as stated by others.

[cj]

I hate threads like this because I know the answer to every last question he is asking, but publishing any of it would kill prices for all of us.

Then why are you here? There is nothing wrong with helping those that ask for it. Why do people help others to learn anything? It is human nature. Everytime you educate anyone about anything, they could wind up being your competitor later on in that very field.

[Tom]

CJ - the odds thing, nice! :ThmbUp:

[kingfin66]

I guess I've missed it with this poster.....I probably skimmed when I should have read more thoroughly....what have I missed?

Mostly, just high levels of arrogance and condenscension.

[Handiman]

CJ. Thanks a million. I do understand the ratings have to be solid. I believe everyone is on that path, some just farther along than others.

As for the guy that knows all the answers to my questions, please explain to me why the sun continues to burn without oxygen. That's something I've always wondered about, but was too lazy to look up or research. You should be able to answer this as it in no way should affect your handicapping.

Handi

[PriceAnProbability]

That's funny, I was just thinking how you have not been in tune with one post yet in this thread. :lol:
You don't hold a clue as to what he is talking about.

I know what he's talking about: he wants to learn to set a value line from his ratings. He said his results aren't there yet. I said they weren't because his ratings aren't profitable and that if I told everyone how to make an accurate odds line I'd be kiling the prices for a lot of profitable players in the process.

[PriceAnProbability]

Then why are you here? There is nothing wrong with helping those that ask for it. Why do people help others to learn anything? It is human nature. Everytime you educate anyone about anything, they could wind up being your competitor later on in that very field.

If this board didn't have "authorized advertisers" I might be quicker to agree with you. If sponsors are elevated in status by virtue of being sponsors then it becomes a propaganda board rather than a discussion board.

Why am I here? To pluck all the profitable knowledge I can from the boards, such as when one guy posted that filles do better with second lasix than first (made a lot so far with that), and with many other people who think they can just give away stuff without it ever biting them back in the pools.

To the extent I can help others without giving away my edge, I do. Value lines are just my bread and butter so I have to bite my tongue.

Now you might ask why I don't just profit at the windows with all this knowledge? I do. There's just a lot of time to kill during the day.

[PriceAnProbability]

Then why are you here? There is nothing wrong with helping those that ask for it. Why do people help others to learn anything? It is human nature. Everytime you educate anyone about anything, they could wind up being your competitor later on in that very field.

CJ, if I showed the public how to generate an accurate morning line based on nothing other than past speed figures and about a dozen other factors, I will *guarantee* you that the prices on *your* horses as well as mine will erode. What happens is everyone doesn't have to listen, but those who do will continue accumulating money until the value is gone. I happen to like my $1.06 ROI or whatever and want to hang onto it for a few years until I can really cash in. Took me a decade to answer the questions this guy is asking. While you guys are contemplating this, I find the question about as challenging as how to add two and two.

Why would you make a line on a partial field? You think horses who are more than 16 points (Beyer) weaker than the top figure never win or never offer value? I know one situation where you'd get creamed if you applied that theory.

There is a fixed relationship between speed figures and a value line. A horse you think will run a 78 today should never be lower in price than one you think will run an 80, and that figure differential can be converted into an oddsline for every horse. A larger differential in figures will lead to a larger differential in price.

Now if you can figure out how to translate figures into a morning line, then figure out what figure to assign the horse, you're good to go.

[Tom]

This is what Kingfin meant, PA. :D

Another expert here t dazzle us.:sleeping:

[Robert Fischer]

5 MO CUISHLE 222 - 206 = 16 + 5 = 21
9 CARMANDIA 220 - 206 = 14 + 5 = 19
3 MALIBU MINT 215 - 206 = 9 + 5 = 14
4 MISS SHOP 212 - 206 = 6 + 5 = 11
8 RAHYS' APPEAL 206 - 206 = 0 + 5 = 5


Maybe off topic, this race was a big near-miss for me. Also a lesson.
Actually was looking at this with the old man (who has some talent in handicapping), as this was right after Thanksgiving. Mo Cuishle was undefeated in three starts, with three easy victories. However, the past competition(class level) was weaker than this current field in this race. Also, I believe at a weaker track. A quick glance at the raw times(taboo?, maybe.., I happen to use raw times as a moderately important reference to speed figs), - the raw times were nothing impressive.
Key horses in this race(no order)-
yolanda b. too - speed, cuts back to her distance
miss shop - very classy, distance here a bit short
malibu mint - quality horse ran against boys in big race previous
rahys appeal- in great form(turf form)

Mo cuishle was a classic example of a horse, who the public would like to be a super-horse. Was she in for a rude awakening here in the class hike, or was this really a super horse? Not to many super-horses out there, and the above negatives(class,track,raw times,)...
The horse was in for a rude awakening, and was a throw-out for exactas.

value race recognized- :ThmbUp:
result - :ThmbDown:
- Excluded the turf horse Rahys Appeal, in spite of the fact, that there was absolutely no need to exclude horses from an exacta wager. It was a short field, with a huge underlay.
Lesson learned.

[Robert Fischer]

maybe you can be profitable with an acurate odds "range"?

[Handiman]

Price,


If you do in fact have things wired as you say, more power to you. I can almost assure you that any info dropped on this board, is not going to affect the odds on your horses. Racing being the "Sport of Kings" has been around for centuries, as it developed when Regional Kings got together and wanted to prove the best of their herd was the best in all the land.

Horses have been running against each other for centuries, and no one has wired the game to the extent they can effect the odds on an exact and continous basis. If they had, winners would not finish offering double digit payoffs. There are just about as many ways to handicap as there are individuals handicapping.

I write homemade software for people who want to easily test or handicap in a manner that is particular to their own desires, and uses formulas and algorithims of their making. I want to be able to include an odds line for them in their programs based on their ratings.

I am in no way, any threat to the commercial producers. That's one reason Dave Schwartz has been kind enough to speak with me and offer advice. CJ also recognizes that I pose no threat to his software venture.

But keep your family secrets and continue to cash in in anonymity. Meanwhile I will keep trying to help people and hope fellow members here will help me to do that.

Regards,
Handi

[NoDayJob]

:D PAP sounds like our old friend----- Lousycapper. I wonder. What ever happened to Grampa???? Being that you are older, are ya any wiser???? I doubt it!

[PriceAnProbability]

Price,
If you do in fact have things wired as you say, more power to you. I can almost assure you that any info dropped on this board, is not going to affect the odds on your horses.

And I can assure you that it will. Only takes a few guys to do it.


Racing being the "Sport of Kings" has been around for centuries, as it developed when Regional Kings got together and wanted to prove the best of their herd was the best in all the land. Horses have been running against each other for centuries, and no one has wired the game to the extent they can effect the odds on an exact and continous basis. If they had, winners would not finish offering double digit payoffs. There are just about as many ways to handicap as there are individuals handicapping.

What happened to Beyer after Picking Winners was published?

I have a very solid method predicting what speed figure a horse will earn in its next race, and I can take any set of expected figures and convert it into an accurate value line. There is some expertise required with regard to trainers and pedigree, but for the most part the calculations are mechanical. One reason I surf this board is to see how close to the truth the other players are, and I'm happy to learn that they are missing a lot of stuff that would sound very basic once it was pointed out to them.


I write homemade software for people who want to easily test or handicap in a manner that is particular to their own desires, and uses formulas and algorithims of their making. I want to be able to include an odds line for them in their programs based on their ratings.

To do that you have to consider literally *every* relevant factor.


I am in no way, any threat to the commercial producers. That's one reason Dave Schwartz has been kind enough to speak with me and offer advice. CJ also recognizes that I pose no threat to his software venture.

And how much has Mr. Schwartz profited from betting with his method in the past year? I know enough about speed-figure making to know CJ turns a profit unless he's the worst money-manager in history, and have learned a few useful things from him (mostly related to pace).

But keep your family secrets and continue to cash in in anonymity. Meanwhile I will keep trying to help people and hope fellow members here will help me to do that.

To the extent something doesn't threaten my power rating recipe, I'm more than happy to share my knowledge.

[Dave Schwartz]

I did just fine this year - thanks for asking.


Regards,
Dave Schwartz