Creating an Odds LIne

[classhandicapper]

Here's a theoretical issue.

Assume any kind of speed/power rating and all the horses having an equal probability of running back to their recent race .

1. Horse A = 100; Horse B = 95

2. Horse A = 50; Horse B = 45

In both cases, horse A is better by 5 points. However, in example 1 horse B is 5% worse and in example 2 horse B is 10% worse.

Should the lines in both these example be identical or does horse A in example 2 have a better chance of winning than horse A in example 1 because he is 10% better?

[Overlay]

I would say (as you suggested) that Horse A would have a better chance in the second scenario, since its rating there is slightly proportionately higher by about 1.35% (10/19 (52.63%) versus 20/39 (51.28%)) in comparison to B's. However, in terms of fair toteboard odds, horse A would be 4-5 in either case, and horse B would be 6-5.

[Overlay]

However, in terms of fair toteboard odds, horse A would be 4-5 in either case, and horse B would be 6-5.
Of course, that would be assuming a two-horse match race, with no other horses participating.

Correction: Fair odds on Horse B would be even money (1-1).

[Jeff P]

If the numbers posted are based on speed figs, does the first situation represent a (significantly) higher class of horses than the second?

I ask because R&D suggests that the ability to consistently string like speed figs together does vary a little based on class and age.

EDIT: If the two situations do represent different class levels, and if the horses in one are less consistent than the other: You might want your odds line to account for the increased possibility of a horse running a klunker.


-jp

.

[cj]

This is pretty vague. What does 1 point represent, and what is the scale? 0-100? 40-100? 40-1000?

[sjk]

I see no difference between the two. The time that represents 0 points and the disparity from that point is purely arbitrary.

[Robert Goren]

If the first example is stakes horses and the second example is maiden claimers, the 5 point difference in first is worth than 5 point difference in the second. The slower horses get the more inconsistent they get. Not that any of them are all that consistent to start out with. A few years ago I did some reasearch on Beyers numbers. A horse running back in 30 days in the same class had an SD of a little less than 6 in my sample.

[pondman]

Assume any kind of speed/power rating and all the horses having an equal probability of running back to their recent race .


There's a lot built into that statement (exluding class), including the age of the horse. Young horses, including maidens wouldn't fit. They're likely to improve. It would have to be a very narrow condition.

I don't know how a 10% margin could be establish on a horse with the given ratings in the #2 example. My own personal line rarely get's above 40% on any given horse. That's asking too much based on the given speed rating. If it had 10%, margin over time, we probably could all figure out a way to retire and make a living. Horses with speed rating of 50 don't win at a 10% higher rate than horses with speed ratings of 45. Do they?

I think you are making a mistake of not including as many horses as you can in the aggregate. In other words, you need to create a line somehow base on how many horses under similiar conditions have won in the last year (or whatever length you think is appropriate.)

[maddog42]

Here's a theoretical issue.

Assume any kind of speed/power rating and all the horses having an equal probability of running back to their recent race .

1. Horse A = 100; Horse B = 95

2. Horse A = 50; Horse B = 45

In both cases, horse A is better by 5 points. However, in example 1 horse B is 5% worse and in example 2 horse B is 10% worse.

Should the lines in both these example be identical or does horse A in example 2 have a better chance of winning than horse A in example 1 because he is 10% better?

Handicap 500 races using this method. Note the winning percentages and payoffs. There probably is no other way to be sure. Patterns will emerge, that will surprise you. Lotsa work, but worth it. I have done this a few times and the results nearly always varies.

Make your odds line match the percentages.

[classhandicapper]

If the numbers posted are based on speed figs, does the first situation represent a (significantly) higher class of horses than the second?

I ask because R&D suggests that the ability to consistently string like speed figs together does vary a little based on class and age.

EDIT: If the two situations do represent different class levels, and if the horses in one are less consistent than the other: You might want your odds line to account for the increased possibility of a horse running a klunker.


-jp

.

Assume all else being equal

[classhandicapper]

This is pretty vague. What does 1 point represent, and what is the scale? 0-100? 40-100? 40-1000?

Assume Beyer scale.

[classhandicapper]

Handicap 500 races using this method. Note the winning percentages and payoffs. There probably is no other way to be sure. Patterns will emerge, that will surprise you. Lotsa work, but worth it. I have done this a few times and the results nearly always varies.

Make your odds line match the percentages.

I've tried things like that from time to time, but it's really difficult to isolate the impact of just the rating because there are always issues like class, distance, trainer, recency etc.. impacting the results. I think the sample size would have to huge in order for all that stuff to kind of equal out.

[cj]

Assume Beyer scale.

Then I would say they are virtually the same mathematically. A 5 point edge is a 5 point edge. However, there is a vast discrepancy in class, so I tend to think the 5 point edge is more significant at the higher levels.

I had a tool that would make a line from figures like this, but I lost it and moved to something else. What it did do was use the lowest figure to establish the base, and add on from there. The 45 would be a 5 (the base), the 50 a 10. The 95 would likewise be a 5, and the 100 a 10.

I tinkered with changing the base depending on class. For example, the 45 might remain a 5 base, giving him 5 of 15 points in the mythical match up, or 2 to 1 odds. But, the 95 horse would get a 10 base, the other horse a 15, giving him a 40% chance as opposed to 33%. I went the other way too, giving the lower ratings a higher base figure. Neither improved the results, they BOTH actually decreased the ROI on betting overlays. But, that was several years ago.

[proximity]

I see no difference between the two. The time that represents 0 points and the disparity from that point is purely arbitrary.

this.

[pondman]

A 5 point edge is a 5 point edge. However, there is a vast discrepancy in class, so I tend to think the 5 point edge is more significant at the higher levels.

Are you expressing an edge as a percent chance of winning? 5%.

I'm just throwing out the idea that the scale itself is larger than 100%. I'm not a beyer's historian, but I think the highest recorded beyers was 128. And I've personally bet on a number of horses with - O beyers, whatever that means.

I now guys who have the data on beyers per class x surface, and create their own subjective line. When you do this however, you're going to get small numbers, which need to somhow be normalized into a scale.

But I haven't heard of anyone estimating a line based just on speed ratings. I just have to assume it's not overwhelming successful. Is it?

[Robert Fischer]

If you convert the Speed Figure to a 100pt scale then it is EASY.

1. Horse A = 100; Horse B = 95

horse A is 5% better than horse B.

2. Horse A = 50; Horse B = 45

Horse A is 5% better than horse B.

[classhandicapper]

Then I would say they are virtually the same mathematically. A 5 point edge is a 5 point edge. However, there is a vast discrepancy in class, so I tend to think the 5 point edge is more significant at the higher levels.



That's what I thought also.

The reason I brought it up is that I am creating a spreadsheet tool similar to the one you used to have for your subscribers.

I am using a somewhat complex mathematical formula I got from someone else as a basis for some of the calculations. It's giving me slightly different probabilities depending on where on the scale a horse has the edge (5 points in this example). But it's not a class related thing. So I guess I need to figure out a way to tweak it so 5 points = 5 points no matter where I am on the scale.

Thanks for the input everyone.

[gm10]

That's what I thought also.

The reason I brought it up is that I am creating a spreadsheet tool similar to the one you used to have for your subscribers.

I am using a somewhat complex mathematical formula I got from someone else as a basis for some of the calculations. It's giving me slightly different probabilities depending on where on the scale a horse has the edge (5 points in this example). But it's not a class related thing. So I guess I need to figure out a way to tweak it so 5 points = 5 points no matter where I am on the scale.

Thanks for the input everyone.

You don't have to. It is convention, but it's not necessary. In fact it is pretty misleading. The significance of a length varies enormously, so most "points" do as well.

Depending on your formula, it could be a pointless exercise anyway. If you really want it to mean the same anywhere on the scale, you can just use the difference (50-45) or (100-95) as the input variable.

[pondman]

If you convert the Speed Figure to a 100pt scale then it is EASY.

Horse A is 5% better than horse B.

?????

[pondman]

Depending on your formula, it could be a pointless exercise anyway. If you really want it to mean the same anywhere on the scale, you can just use the difference (50-45) or (100-95) as the input variable.

Are you not considering the aggregate?

A 100 may be rare. A 50 might be the average. I'm not a speed person or a statistics head, but it sounds dangerous to give so much credit to horses near the medium.

[cj]

Depending on your formula, it could be a pointless exercise anyway. If you really want it to mean the same anywhere on the scale, you can just use the difference (50-45) or (100-95) as the input variable.

That was the question though, is the 5 the same at various points on the scale. I'm sure he knows he COULD use it, the question was SHOULD he?

[gm10]

That was the question though, is the 5 the same at various points on the scale. I'm sure he knows he COULD use it, the question was SHOULD he?

It depends on the formula, but imo he probably should.

[Capper Al]

Where's TrifectaMike when you need him? The bigger side of this equation is saying everything else being equal. Comprehensive handicappers look at the pieces and add them with impact values. Once they have an idea of what the impact values are then the math is easy and a scenario where all factors are equal could be entertained.

[Robert Fischer]

?????

if the speed figure is calibrated(?) so that it is on a "100-point scale"... then each point becomes 1% or 1 part of 100.

Then you can use gm10's suggestion of using the difference(subtracting.)

With many of the popular speed figures the scale is not exactly 100. You get 120 speed figures and then you generally have a lower limit as well that is usually above zero (30s?). The popular speed figures are usually "curved".

[Dave Schwartz]

Once they have an idea of what the impact values are then the math is easy and a scenario where all factors are equal could be entertained.

Well, the math to produce something theoretically correct is easy. The problem is that it will not be correct in the real world. It will need tweaking. It always does.

While many people know that the "big boys" (or maybe I should say "the biggest of the big boys) use Probit or Logit regression. Folks, this is nothing like "least squares regression."

What most people do not realize is that it is an "iterative" process. Loosely, select (or determine) the first factor and run it against your database, determining the optimum weight. Then add cycle one factor at a time into the 2nd slot, to determine which one fits best and at what weight.

Once you have the 2nd factor, repeat the process with the 3rd, 4th, 5th factors, all the way to the nth factor. Some of the "biggest of the big" use models with 50 or more factors.

Back to the real world. My point is that to think anyone is going to ask a question like, "How do I make an odds line from speed numbers that look like this?" and get a simple solution that works is probably not looking at reality here.

The fact that one is only using simple data does not make the answer to the question any less complex.

This would be my observed experience.



Dave

[Robert Fischer]

Where's TrifectaMike when you need him? The bigger side of this equation is saying everything else being equal. Comprehensive handicappers look at the pieces and add them with impact values. Once they have an idea of what the impact values are then the math is easy and a scenario where all factors are equal could be entertained.

right, that is the assumption(actually a couple)


everything else is equal = we are talking only about figs here
horses will run back to their figs
Operating on those two unrealistic assumptions, create an odds line, and do it at different places on the number-line.


Opposite of comprehensive, the goal is to focus just on this specific.

[classhandicapper]

right, that is the assumption(actually a couple)



everything else is equal = we are talking only about figs here
horses will run back to their figs
Operating on those two unrealistic assumptions, create an odds line, and do it at different places on the number-line.


Opposite of comprehensive, the goal is to focus just on this specific.




Essentially, before I can get to some of the complexities, I have to make sure the simple assumptions are correct.

The first thing I noticed was that I was getting a 1% +/- difference in the estimated win probabilities depending on where I was on the scale using the algorithm I had. That intuitively seemed wrong.

I found a way to correct it.

[Robert Fischer]

Essentially, before I can get to some of the complexities, I have to make sure the simple assumptions are correct.

The first thing I noticed was that I was getting a 1% +/- difference in the estimated win probabilities depending on where I was on the scale using the algorithm I had. That intuitively seemed wrong.

I found a way to correct it.

often, but not always this can be corrected with the formula.

sounds like you tweaked it

[classhandicapper]

Well, the math to produce something theoretically correct is easy. The problem is that it will not be correct in the real world. It will need tweaking. It always does.

While many people know that the "big boys" (or maybe I should say "the biggest of the big boys) use Probit or Logit regression. Folks, this is nothing like "least squares regression."

What most people do not realize is that it is an "iterative" process. Loosely, select (or determine) the first factor and run it against your database, determining the optimum weight. Then add cycle one factor at a time into the 2nd slot, to determine which one fits best and at what weight.

Once you have the 2nd factor, repeat the process with the 3rd, 4th, 5th factors, all the way to the nth factor. Some of the "biggest of the big" use models with 50 or more factors.

Back to the real world. My point is that to think anyone is going to ask a question like, "How do I make an odds line from speed numbers that look like this?" and get a simple solution that works is probably not looking at reality here.

The fact that one is only using simple data does not make the answer to the question any less complex.

This would be my observed experience.



Dave

I'm not anything close to a regression expert, but I've seen some interesting things looking at the work of others for basketball handicapping.

For example:

If you run a test using only factors A and B, the best weight might be 50% - 50%

If you run a test using factor A, B, and C the best weight might be 20%, 30%, 50%.

All of a sudden the correct relationship between A and B changes a lot when you include C, but you could never know that just running through a database.

Running through the database you'd find the best values of A, B, and C when A and B were equal - which might be good, but would not be the best possible answer.

I think you almost have to use a regression analysis to get the values close to correct. The trial and error approach is a very good one and can find profitable results, but it has issues.

I am nowhere near ready to try to run a regression analysis (even though I probably have some sources in basketball that could help me). My approach for now is going to be to try to mimic my actually current handicapping/thinking as closely as possible so the formula is spitting out the same results as my thinking.

[TrifectaMike]

Well, the math to produce something theoretically correct is easy. The problem is that it will not be correct in the real world. It will need tweaking. It always does.

While many people know that the "big boys" (or maybe I should say "the biggest of the big boys) use Probit or Logit regression. Folks, this is nothing like "least squares regression."

What most people do not realize is that it is an "iterative" process. Loosely, select (or determine) the first factor and run it against your database, determining the optimum weight. Then add cycle one factor at a time into the 2nd slot, to determine which one fits best and at what weight.

Once you have the 2nd factor, repeat the process with the 3rd, 4th, 5th factors, all the way to the nth factor. Some of the "biggest of the big" use models with 50 or more factors.

Or start with 50 and reduce 1 factor at a time.

Back to the real world. My point is that to think anyone is going to ask a question like, "How do I make an odds line from speed numbers that look like this?" and get a simple solution that works is probably not looking at reality here.

You can achieve a fairly good odds line using speed ratings.

However, it may not be as simple as one thinks. One can use a multinomial regression equation to predict a speed rating... using past speed ratings as a predictors. You'll need to handle special case when a horse does not finish ( which means rating = 0, you'll need a Bernoulli probaility to handle this case).

If you assume the predicted speed ratings have a joint distribution of SR_1, .... SR_n is multivariate normal, you can perform a simulation to arrive at your probabilities.

In order to perform the simulation you will have to determine a mean vector and covariance matrix for the multivariate normal.

Since the speed ratings are not independent, you'll have to employ some further analysis to determine the off diagonal terms in the covariance matrix. ( This can be done by determining the sample ( pairs of horse that ran in the same race) intraclass correlation.

Next, you'll have to generate speed rating samples as a multivariate normal with mean (speed rating from the regression equation) and the covariance as described above ( specifically the diagonals are set to the square of standard error of regression and the off diagonal terms a function of the intraclass correlation and the square of standard error of regression).

Still not done! The multivariate samples are indirectly derived by generating IID standard Normal samples and using a Closkey factorization ( lower triangular) on the covariance.

This is essentially what would be required at arrive a "good" odds line using ONLY speed ratings.

The fact that one is only using simple data does not make the answer to the question any less complex.

This would be my observed experience.



Dave

As Dave said, "simple data does not make the answer to the question any less complex.".

Mike (Dr Beav)

[classhandicapper]

Just to be clear, I am not trying to create an odds line based ON SPEED FIGURES ALONE. I'm about as comprehensive a handicapper as one can imagine. If anything, I look at too many things.

As a first step I was trying to create a tool that would more or less mimic my thinking if all I had to look at was speed figures. But I ran into the issue I described. I was able to tweak that and get it work.

Today I have added class and it seems to be approximating my thinking fairly well. As time passes I will continue to add things and refine it.

[pondman]

Today I have added class and it seems to be approximating my thinking fairly well. As time passes I will continue to add things and refine it.

I've tried adding speed to a class method. The combination of speed and class doesn't mix well for me. The result: the odds are going to be to low. I couldn't get the same play as if I just went with class, and ignored at least the race's pp and all ratings. I need to see the where and when of races , but I ignore the performance. At that point my ROI became stellar. It's become a regular routine of cashing +$3,000 tickets.

[Dave Schwartz]

I've tried adding speed to a class method. The combination of speed and class doesn't mix well for me. The result: the odds are going to be to low.

IMHO, this is why most odds line systems fail: the handicapper does not LIKE the answer he gets!

See, the odds are SUPPOSED to be low because - guess what? - The favorite(s) are the best horse(s) in most races!

[Robert Fischer]

I take note anytime a horse is changing class.


Is this a step up in class today?
Is this a first time starter?/trainer?/class?/distance?
Does this step up fit a common trainer angle ? (ie. 1st after claim up in class)
Does this trainer angle tip-off trainer intent ?
Is this horse obvious ?? or is it an unknown ??
Does the trainer win a high% of these types of races in general?? With this angle??
...

[pondman]

See, the odds are SUPPOSED to be low because - guess what? - The favorite(s) are the best horse(s) in most races!

I'd go with 2/3 of the races on the high end are won by the crowd's top 4 choices, which includes most horse at less than 4-1. But I'd continue with: the crowd (in combination with encouragement by the M/L ) makes big mistakes in 30% of the races, and ignore the best horse in a race. Most of these mistakes are the result of a not knowing local industry practices, over reliance on speed ratings. and handicapping generalization (For example don't bet horses after a layoff.) I'd rather be throwin $300 on a $67 horses a few times a week, and hitting a fair number a month, than worrying about a method which requires betting favorites. I wouldn't be doing this everyday If I had to bet favorites. This is all in context of the higher end with pool large enough to support the bets.

[pondman]

I take note anytime a horse is changing class.

Is this a step up in class today?
Is this a first time starter?/trainer?/class?/distance?
Does this step up fit a common trainer angle ? (ie. 1st after claim up in class)
Does this trainer angle tip-off trainer intent ?
Is this horse obvious ?? or is it an unknown ??
Does the trainer win a high% of these types of races in general?? With this angle??
...

#1 What about: is this billionaire spending money wildly, racing this horse in something soft?

In context to the OP's question, if you add a high speed ratings to this, you'll get nothing back on the bet. If it looks like crap, and it's ridden by a low % or unknown jockey, the crowd will ignores it. And what happens? You'll buy yourself a new car.

If the trainer rules works for you great. I don't see the trainer being the go to-- it's the connection itself. The trainer is just piggy backing.

[Robert Fischer]

I'd go with 2/3 of the races on the high end are won by the crowd's top 4 choices, which includes most horse at less than 4-1. But I'd continue with: the crowd (in combination with encouragement by the M/L ) makes big mistakes in 30% of the races, and ignore the best horse in a race. Most of these mistakes are the result of a not knowing local industry practices, over reliance on speed ratings. and handicapping generalization (For example don't bet horses after a layoff.) I'd rather be throwin $300 on a $67 horses a few times a week, and hitting a fair number a month, than worrying about a method which requires betting favorites. I wouldn't be doing this everyday If I had to bet favorites. This is all in context of the higher end with pool large enough to support the bets.

You must have some cojones ! :D

[Dave Schwartz]

I'd go with 2/3 of the races on the high end are won by the crowd's top 4 choices

Top 3 win around 75%, top 5 win 90%.

[HUSKER55]

hey robert, tailored boxers :D

[Robert Fischer]

#1 What about: is this billionaire spending money wildly, racing this horse in something soft?

In context to the OP's question, if you add a high speed ratings to this, you'll get nothing back on the bet. If it looks like crap, and it's ridden by a low % or unknown jockey, the crowd will ignores it. And what happens? You'll buy yourself a new car.

If the trainer rules works for you great. I don't see the trainer being the go to-- it's the connection itself. The trainer is just piggy backing.

You make good points here.

these type of trainer rules work better for me if instead of trying to add them to the speed figures(or however i got my contenders from), I look at the few horses that fit these rules and see if there is any value.

For horses that are not changing in class today, and are running for the same connections should already be well accounted for with the speed figure.

[Capper Al]

Before PC computer and sophisticate algorithms, some paper and pencil handicappers created a simple odds line. They take the number of contenders and assign that to their top pick. Their second choice would be # contenders+1. The third choice would be # of contenders +2, etc. For example, in a 12 horse field if they had identified 4 contenders their top choice would be 4/1, second choice 5/1, third choice 6/1, and fourth and last choice would be 7/1. The reasoning works like counting sides of a die. If you are correct about there being four contenders then the odds should be 3/1 but one is add for the unknown horse that could win and isn't a contender so you get back to 4/1, the original number of contenders. It's a simple paper and pencil method, but it works if you can handicap well and have discipline.

I can hear the math boys. Some contenders are more likely to win than others and statistical weights are a must for figure this. I don't disagree, but it's good enough for horse racing. Whatever over estimates of odds will force one to wait out for higher odds like they should be doing anyway. It works.

[Capper Al]

IMHO, this is why most odds line systems fail: the handicapper does not LIKE the answer he gets!

See, the odds are SUPPOSED to be low because - guess what? - The favorite(s) are the best horse(s) in most races!

For example -- Once a horse is pegged at 8/1, and it's on the tote-board at 6/1, quite often a voice comes from nowhers and says, "It's close enough. Make the play."

[MPRanger]

Here's a theoretical issue.

Assume any kind of speed/power rating and all the horses having an equal probability of running back to their recent race .

1. Horse A = 100; Horse B = 95

2. Horse A = 50; Horse B = 45

In both cases, horse A is better by 5 points. However, in example 1 horse B is 5% worse and in example 2 horse B is 10% worse.

Should the lines in both these example be identical or does horse A in example 2 have a better chance of winning than horse A in example 1 because he is 10% better?

Are these two horses the only contenders? The odds to win of all the others = 0? You are asking about the value between only the two.

I can't really answer your question as asked. Maybe I don't understand. Here's the way I make an oddsline;

I start off by giving the non-contenders 15-20% of 100 total points. I just Zen out and decide whether it's 15 or 20. No real rule. I just go by feel. No one can teach this but you will develop the skill with practise. Every race is different.

Then I divide the remaining 80 points for example evenly among the contenders. Then I Zen out again to move points around among the contenders til it seems right.

I don't actually track my results as a pro would so I can't say exactly how well I do it. But I am completly confident that I can evaluate a field in the races I like.

I know that's vague but that's all I can offer. And I'm just an amatuer anyway. But I do it every single time. I used to use it for finding overlays in the win pool but it bores me to play that way. Lately I have been experimenting with market efficiency where my selections must be in line with the toteboard and only playing to win when my picks are in the top three on the board. Like I say, I'm not keeping records but I cash a lot more win and place tickets. My goal is not to grow a bankroll but to win every time I go to the track.

Also, making an oddsline is very helpful on saving money when buying races with the .10 super in my experience.

[Dave Schwartz]

Ranger,

That is a start on a very powerful approach, one that is amazingly close to what I use.

:ThmbUp: :ThmbUp:


Originally Posted by Dave Schwartz
IMHO, this is why most odds line systems fail: the handicapper does not LIKE the answer he gets!

See, the odds are SUPPOSED to be low because - guess what? - The favorite(s) are the best horse(s) in most races!


From Capper Al:
For example -- Once a horse is pegged at 8/1, and it's on the tote-board at 6/1, quite often a voice comes from nowhers and says, "It's close enough. Make the play."


Al, actually, what I meant was that most handicapper's don't like the fact that their top horses are consistently the low-priced horses in the race. They seem to have a tendency to say, "Well, this can't work for me because I am just not going to bet that many favorites."

The purpose of making a line is - first - to get a realistic view of the race. If a player doesn't actually WANT that, how can he expect to be competitive?



Dave

[Robert Fischer]

Before PC computer and sophisticate algorithms, some paper and pencil handicappers created a simple odds line. They take the number of contenders and assign that to their top pick. Their second choice would be # contenders+1. The third choice would be # of contenders +2, etc. For example, in a 12 horse field if they had identified 4 contenders their top choice would be 4/1, second choice 5/1, third choice 6/1, and fourth and last choice would be 7/1. The reasoning works like counting sides of a die. If you are correct about there being four contenders then the odds should be 3/1 but one is add for the unknown horse that could win and isn't a contender so you get back to 4/1, the original number of contenders. It's a simple paper and pencil method, but it works if you can handicap well and have discipline.

I can hear the math boys. Some contenders are more likely to win than others and statistical weights are a must for figure this. I don't disagree, but it's good enough for horse racing. Whatever over estimates of odds will force one to wait out for higher odds like they should be doing anyway. It works.

This is fun.

Obviously their will be complaints, and that this will often be "too picky" for some. MOST people would reduce their losses if they were "too picky"(see bold^^).
Nice stakes race BCDAY DerbyDay strategy for beginners or novices trying for a fun day at the races.

Other than pulling this out on the big days (competitive, full fields, good odds), I would also suggest that beginners bet EVERY contender who happens to be over the odds line. You may be lucky to find one in most races (in which they should pass), but if they find multiple qualifiers playing each one would increase their hit%. For the type of player that would benefit from this Paper&Pencil Odds Line, it would help them stay alive for the day and should their contenders be solid - then we are talking about the difference between a few bucks in a race they would have won anyway, or staying alive in a race they would have only playing the top choice.

Should be a book of these things, as a complete novice could try several P&P methods and have a rewarding day at the track. :ThmbUp:

[Capper Al]

This is fun.

Obviously their will be complaints, and that this will often be "too picky" for some. MOST people would reduce their losses if they were "too picky"(see bold^^).
Nice stakes race BCDAY DerbyDay strategy for beginners or novices trying for a fun day at the races.

Other than pulling this out on the big days (competitive, full fields, good odds), I would also suggest that beginners bet EVERY contender who happens to be over the odds line. You may be lucky to find one in most races (in which they should pass), but if they find multiple qualifiers playing each one would increase their hit%. For the type of player that would benefit from this Paper&Pencil Odds Line, it would help them stay alive for the day and should their contenders be solid - then we are talking about the difference between a few bucks in a race they would have won anyway, or staying alive in a race they would have only playing the top choice.

Should be a book of these things, as a complete novice could try several P&P methods and have a rewarding day at the track. :ThmbUp:

Robert,

I agree. It is good for anyone who can pick contenders with an 80% or better hit ratio. A novice might have a little difficulty hitting that mark. Still it is better to have a wagering method with some sense to it than not to have one.

[MPRanger]

[QUOTE=Dave Schwartz]Ranger,

That is a start on a very powerful approach, one that is amazingly close to what I use.

:ThmbUp: :ThmbUp:

Hello Dave, if was a betting man, I would have wagered on that!:)

[cj]

Ranger,

That is a start on a very powerful approach, one that is amazingly close to what I use.

:ThmbUp: :ThmbUp:

Hello Dave, if was a betting man, I would have wagered on that!:)

This sounds a lot like Steve Fierro's method from his book The Four Quarters of Horse Investing.

[Dave Schwartz]

What I am doing is pretty far off from there but long ago I gave credit to Fiero for his excellent work. Lots of problems with his original approach but it was the forerunner of where I am today.

LOL - A really interesting twist that I put on it at the end. Something would would surprise a lot of people. If there is a lot of interest, perhaps I will do a workshop on it this summer.


Dave

[Capper Al]

What I am doing is pretty far off from there but long ago I gave credit to Fiero for his excellent work. Lots of problems with his original approach but it was the forerunner of where I am today.

LOL - A really interesting twist that I put on it at the end. Something would would surprise a lot of people. If there is a lot of interest, perhaps I will do a workshop on it this summer.


Dave

I'd be interested in a workshop on creating an odds line.

[Robert Goren]

Robert,

I agree. It is good for anyone who can pick contenders with an 80% or better hit ratio. A novice might have a little difficulty hitting that mark. Still it is better to have a wagering method with some sense to it than not to have one.You still have to find a way to bet the contenders so you make a profit no matter who wins. That means you are going to have some pretty decent odds on a horse that has at least fair chance of winning.
A better case can be made dutching exactas and doubles where the likihood of having more than one good bet goes up quite a bit.

[Capper Al]

You still have to find a way to bet the contenders so you make a profit no matter who wins. That means you are going to have some pretty decent odds on a horse that has at least fair chance of winning.
A better case can be made dutching exactas and doubles where the likihood of having more than one good bet goes up quite a bit.

You're right. That's why my paper and pencil (P&P) method in post #41 of this thread is good. For example, in an 8 horse field with 3 contenders with real computed odds of 1/2, 2/1, and 6/1-- the P&P method would insist on 3/1, 4/1, and 5/1. Mathematically, too high on the 1/2, and 2/1 contenders. Too low on the 6/1 horse. But the point of view isn't on the horse with the P&P method, but on how the handicapper sees them. When in combat, I'll go with the P&P because of this. As proven with my prediction of Tampa Bay's race 9 yesterday being a chaos race, this system will also point out a chaos race. The statistical approach could predict chaos races also when it takes more than half the field to achieve an 80% confidence level. Look Mom, no software needed!