Paceline Selection Algorithms

[Jeff P]

I've been following the back and forth over Paceline Selection between Light, Doug, HeadHawg, and Handi in the Handifast thread... Some good points raised on both sides...

Pacelines... To pick or not to pick?

On the one hand you can create algorithms to evaluate, score, and weight every running line... and come up with pretty good compound numbers that test out nicely roi-wise.

On the other hand there is something to be said for allowing the expert handicapper to pick his (or her) own lines.

In my own R&D I've always leaned more towards creating compound numbers... as I've never been able to (meaning me personally) pick lines in a way that outperforms (using flat win bet roi in large data samples as the benchmark for comparing two different approaches) what I've been able to achieve using a compound number approach. I've written a few paceline selection algorithms myself... but let's save that topic for later... I'm hoping to get some input first.

There are a lot of intelligent posters on this board. I'm curious about your thoughts...

Pacelines... To Pick... or Not To Pick?



-jp

.

[Red Knave]

Jeff
This seems to me to be one of several areas of handicapping where art is still more useful than science. And, same as you, I know I'm no artist.
For those who can visualize an entire race, picking a representative paceline seems obvious. For me, not so much. Dick Schmidt, when he edited the Sartin Follow Up, said he always used the last line except in the case of a nuclear attack (I'm paraphrasing, but words to that effect). Jim Bradshaw just intuited the correct line because he knew what was going to happen during the running of the race.
In my handicapping, lazy and tentative as it is, I prefer NOT to pick pacelines and use software that gives me a composite view of each horse that I can use to compare them one to one as well as one to the entire field. I feel a lot more comfortable with that than if I had just one or two hopefully representative lines.

[Light]

Jeff

This is a vast subject,but let me tell you why their paceline method is flawed using the 3 basic scenarios often encountered.

1) Many horses drop in class. Used to run for graded stakes and are now in claiming ranks. Does it make sense to you to use a race a few races back where the horse ran a 90 vs stake competition and now cant reach an 80 with claimers? Not to me. The owner and trainer have already said to the readers of the pp's,"my horse cannot do that 90 anymore and he cant even do an 80 vs claimers so I'm dropping him some more". Yet people who average x number of pacelines are ignoring this really obvious form factor and using the horse's 90 when he used to have 4 legs and now uses a walker.

2) Nobody has ever come up with a turf to dirt equivalent chart. Everyone knows its apples and oranges. But those who average x number of pacelines back ignore the fallacy of using incongrous surface change figs.

3) Bris distance figs are not the same for a sprinter and a router. A 100 in a sprint and a 100 in a route are not equal. Which is better depends on todays distance. You really need to make an asterisk on distance switchers and apply a totally seperate criteria to see if a sprinter can stretch out or a router can sprint.

[Warren Henry]

Jeff

This is a vast subject,but let me tell you why their paceline method is flawed using the 3 basic scenarios often encountered.

1) Many horses drop in class. Used to run for graded stakes and are now in claiming ranks. Does it make sense to you to use a race a few races back where the horse ran a 90 vs stake competition and now cant reach an 80 with claimers? Not to me. The owner and trainer have already said to the readers of the pp's,"my horse cannot do that 90 anymore and he cant even do an 80 vs claimers so I'm dropping him some more". Yet people who average x number of pacelines are ignoring this really obvious form factor and using the horse's 90 when he used to have 4 legs and now uses a walker.

2) Nobody has ever come up with a turf to dirt equivalent chart. Everyone knows its apples and oranges. But those who average x number of pacelines back ignore the fallacy of using incongrous surface change figs.

3) Bris distance figs are not the same for a sprinter and a router. A 100 in a sprint and a 100 in a route are not equal. Which is better depends on todays distance. You really need to make an asterisk on distance switchers and apply a totally seperate criteria to see if a sprinter can stretch out or a router can sprint.
So, you have identified three areas of concern regarding paceline selection, but I don't see any suggestion for overcoming the problem. Enlighten us.

[sjk]

I have had good luck with compound numbers (although I have no actual number because it is all internal to the program processing). Makes good sense to me to use all of the relevant information you can.

[raybo]

AllData PPs will have an auto-paceline select method, with 14 selection criteria defaults that the user has options for each one. It also has these 14 criteria grouped into 6 categories which the user ranks (1 to 6) in order of importance. The program will look for perfect matches first then proceed to drop the least important category and re-search for matches, then the 2nd least important, etc., etc..

The user also has the option of overriding any or all of the auto-selected pacelines by checking a cell next to his preferred paceline.

The final selections will then appear in a summary with only the selected paceline for each horse

[Tom]

I use a couple of the auto-modes in HTR, but I always look at them and change what I don't agree with.

[BillW]

On the one hand you can create algorithms to evaluate, score, and weight every running line... and come up with pretty good compound numbers that test out nicely roi-wise.

.

I generate/evaluate both :) .

[douglasw32]

I think if you are going to use pacelines, become an artist.
It is far to subjective for a computer.

[Light]

Red Knave and douglasw32 both refer to selecting your own paceline as an art. I would say that's true in only a small number of cases. In the vast majority of cases its a no brainer which paceline to pick.

[raybo]

I think if you are going to use pacelines, become an artist.
It is far to subjective for a computer.

I agree that the auto-selection process should be verified by the user and overridden when it is obviously in error. If one uses a thorough approach, in programming the auto-selection criteria and decision matrix, most of the time it will yield the same paceline the user would have chosen manually. Auto-select saves handicapping time and helps prevent user errors by ensuring the fundamental selection criteria and method remains consistent.

[raybo]

Red Knave and douglasw32 both refer to selecting your own paceline as an art. I would say that's true in only a small number of cases. In the vast majority of cases its a no brainer which paceline to pick.

The only time it's a "no-brainer" is when there is only one paceline, or none. However, after that, the brain better be working overtime.

[douglasw32]

I agree that the auto-selection process should be verified by the user and overridden when it is obviously in error. If one uses a thorough approach, in programming the auto-selection criteria and decision matrix, most of the time it will yield the same paceline the user would have chosen manually. Auto-select saves handicapping time and helps prevent user errors by ensuring the fundamental selection criteria and method remains consistent.

Yes that is a very good idea, make sure with the computer the human does not mess it up but then as the human know when to tell the computer to take a hike.

[headhawg]

The only time it's a "no-brainer" is when there is only one paceline, or none. However, after that, the brain better be working overtime.That's the same way that I think about it. You could use the "good enough" method which basically what Mitchell said about using the last paceline 90% of the time. It's kind of like using TPR or Bris figs. They just need to be good enough because we're not trying to land a shuttle on mars. But the implication with that approach is that there is something else in one's handicapping that supplements the paceline. That is, the paceline is not the be all, end all part of handicapping, but merely one step in a longer -- and hopefully more revealing -- process.

On the other hand, if one feels that picking pacelines is the way, the truth, and the light, then you better be damn good at it. And I am not convinced that there is an auto-selection method out there that is decidedly better than a compound method that Jeff P and sjk may use. Obviously without seeing the algorithms I am in no position to evaluate them. They could be complex, but that doesn't necessarily make them useful over and above using the last line which would be simpler. Or it may not even be better than the averaging method that Handifast uses.

And as I have mentioned in the other thread I don't have anything against paceline selection either manually or automagically (my own software does it). I just don't think that a program has to do it that way, just like I don't think that a program that doesn't include trainer stats as a significant selection factor is fatally flawed. It's utter nonsense to think that there is only one way to design a handicapping program.

[LottaKash]

For me, I will almost always use the most recent paceline, provided that there is a genuine excuse for a loss or a poor performance off of that line, than I will go back as far as 4-back if the first 3-are excusable...after that, a crystal ball may be needed...(I don't use software, tho)

Can a software program know when a running-line may be excused ?

best.

[headhawg]

Can a software program know when a running-line may be excused ?Yes, as long as the programmer can write the decision rules that the program uses to select the paceline. If it's fairly simple like Pizzolla's LASST method then not really a problem. (Although defining what constitutes a "trouble" line is a bit challenging.) However if you're really going to have a top-notch auto-selection method there is a much more complex set of rules to code.

[raybo]

For me, I will almost always use the most recent paceline, provided that there is a genuine excuse for a loss or a poor performance off of that line, than I will go back as far as 4-back if the first 3-are excusable...after that, a crystal ball may be needed...(I don't use software, tho)

Can a software program know when a running-line may be excused ?

best.

Yes, what headhawg said!

Having a program/spreadsheet "read" the trouble lines is, indeed, complex but can be done well enough to give your auto-selection method a good base, regarding trouble. However, it is no replacement for reading them yourself, watching race videos, etc.. Thus, the need for manual paceline, individual, over-ride capability.

Also, no matter how good your paceline selection method is, there will often be times when no representative paceline exists, which means the handicapper must have another means of comparing those horses to the rest of the field. This is really the "art" portion of the problem.

[highnote]

2) Nobody has ever come up with a turf to dirt equivalent chart. Everyone knows its apples and oranges.

The pace and speed figures of Cary Fotias from Equiform are the best I've used for comparing Turf and Dirt.

3) Bris distance figs are not the same for a sprinter and a router. A 100 in a sprint and a 100 in a route are not equal. Which is better depends on todays distance. You really need to make an asterisk on distance switchers and apply a totally seperate criteria to see if a sprinter can stretch out or a router can sprint.

I know you're talking about BRIS, but for your point 3) Equiform pace and speed figures can be used to compare a sprinter with a router. I know for a fact that he uses some pretty sophisticated statistical techniques to make distance comparisons.

[douglasw32]

I have a new program based on some of the handifast work I use more often than not...

If I were to tackle the paceline idea, what in your opinions would be the algo

Like 1st have it look for distance or surface or both, then recency then only wins etc.

in what order would you have the computer start being objective.

I think that I would like to average the 6, have the computer toss anything it felt it needed to, re average, then combine the two it should lower some and raise others ???

any ideas...

[raybo]

I have a new program based on some of the handifast work I use more often than not...

If I were to tackle the paceline idea, what in your opinions would be the algo

Like 1st have it look for distance or surface or both, then recency then only wins etc.

in what order would you have the computer start being objective.

I think that I would like to average the 6, have the computer toss anything it felt it needed to, re average, then combine the two it should lower some and raise others ???

any ideas...

Here's my default criteria rank, from most important to least important:

1. Surface (including surface condition)
2. Distance (within 1f)
3. Recency (for last race and also how far back for previous races)
4. Class (type,NW,% +-price/purse,state bred)
5. Age/Sex
6. Track

Look for a perfect match 1st, then drop #6 and look for perfect match again, then drop #5, then #4, then #3, then #2, and finally #1.

[Zoot]

Another procedure would be to assign points to each paceline selection factor and come up with a total score fore each paceline.

For example :

if surf = today’s surf then 6 points else ? pts
+ if distance = today’s dist then 3 pts, else if within 1 furlong then 2 pts, else …
+ if days since <= x then x points, else …
+ if same jockey as today then x points
+ etc …

Then add the points for each paceline and rank them. Then you can use the top ranked, or top x ranked for multiple pacelines, or all pacelines that meet a minimum score … you can slice it and dice it any way you want.

From a programming point of view this makes it a little easier if you want to experiment with different weightings for the various factors.

[Dave Schwartz]

I hope I am not intruding here but why not build a user-programmable "paceline selection system?"

We've had one in our software for years. It is not as difficult a process as one might think.

A system has a set of rules. The rules are numbered in logical order (i.e. they fire from top to bottom).

Each rule has three phases:
1. The elimination phase (which removes pacelines)
and
2. The ordering phase which sorts the pacelines that are not elim'd.
and
3. The selection phase (which actually weights pacelines).


If you would care to hear more, I'll provide a deeper explanation.


regards,
Dave Schwartz

[raybo]

Another procedure would be to assign points to each paceline selection factor and come up with a total score fore each paceline.

For example :

if surf = today’s surf then 6 points else ? pts
+ if distance = today’s dist then 3 pts, else if within 1 furlong then 2 pts, else …
+ if days since <= x then x points, else …
+ if same jockey as today then x points
+ etc …

Then add the points for each paceline and rank them. Then you can use the top ranked, or top x ranked for multiple pacelines, or all pacelines that meet a minimum score … you can slice it and dice it any way you want.

From a programming point of view this makes it a little easier if you want to experiment with different weightings for the various factors.

I had thought of doing "AllData" like that but thought that it would be better to look for exact matches, with the user's most important factors remaining "full value", if at all possible.

If you weight the factors, in pacelines, you could end up with higher points, for some pacelines, that were actually too weak in the most important factors. I decided that the user's important factors should remain valid, during the selection process, as long as possible by dropping the least important factors first, working upward from the least important to the most important.

[raybo]

I hope I am not intruding here but why not build a user-programmable "paceline selection system?"

We've had one in our software for years. It is not as difficult a process as one might think.

A system has a set of rules. The rules are numbered in logical order (i.e. they fire from top to bottom).

Each rule has three phases:
1. The elimination phase (which removes pacelines)
and
2. The ordering phase which sorts the pacelines that are not elim'd.
and
3. The selection phase (which actually weights pacelines).


If you would care to hear more, I'll provide a deeper explanation.


regards,
Dave Schwartz

"AllData" will be user programmable, as the user can select from 2 or more options for each of the 14 criteria, then he/she must rank these criteria (which are grouped into 6 subject related categories) from 1 to 6, regarding their preference concerning the importance of each category.

Regarding recency the user can specify number of days since last race, number of days concerning how far back to look for previous pacelines, or he/she can specify number of pacelines to go back. Later I will add the ability to average or find the median for multiple pacelines (again, the user will be able to specify how many pacelines or how far back to go for those pacelines and whether or not to exclude the high or low or both). The user will have to specify which factor(s) to average/find the median.

[douglasw32]

Another procedure would be to assign points to each paceline selection factor and come up with a total score fore each paceline.

For example :

if surf = today’s surf then 6 points else ? pts
+ if distance = today’s dist then 3 pts, else if within 1 furlong then 2 pts, else …
+ if days since <= x then x points, else …
+ if same jockey as today then x points
+ etc …

Then add the points for each paceline and rank them. Then you can use the top ranked, or top x ranked for multiple pacelines, or all pacelines that meet a minimum score … you can slice it and dice it any way you want.

From a programming point of view this makes it a little easier if you want to experiment with different weightings for the various factors.

Cool Thanks everyone, I think I may try ZOOT's ideas on Raybo criteria for my lil home grown.

Dave please explain, it may help someone, I think on the homegrown I have I want it to all be done without my help ;) in the end.

[Dave Schwartz]

Here is an example of the built-in paceline selection system for sprints.

http://www.horsestreet.com/BBSImages/PLineSys01.png

There are 9 rules. My belief is that the most representative paceline is a close finish (within 1 length) at similar distance and surface. Thus, the first rule is:

http://www.horsestreet.com/BBSImages/PLineSys02.png

http://www.horsestreet.com/BBSImages/PLineSys03.png

If you follow this example, you can image that the second rule loosens the "days" restrictions. Of course, we will never get to the second rule if the first rule does all we need.


http://www.horsestreet.com/BBSImages/PLineSys04.png

[ranchwest]

Jeff

This is a vast subject,but let me tell you why their paceline method is flawed using the 3 basic scenarios often encountered.

1) Many horses drop in class. Used to run for graded stakes and are now in claiming ranks. Does it make sense to you to use a race a few races back where the horse ran a 90 vs stake competition and now cant reach an 80 with claimers? Not to me. The owner and trainer have already said to the readers of the pp's,"my horse cannot do that 90 anymore and he cant even do an 80 vs claimers so I'm dropping him some more". Yet people who average x number of pacelines are ignoring this really obvious form factor and using the horse's 90 when he used to have 4 legs and now uses a walker.

2) Nobody has ever come up with a turf to dirt equivalent chart. Everyone knows its apples and oranges. But those who average x number of pacelines back ignore the fallacy of using incongrous surface change figs.

3) Bris distance figs are not the same for a sprinter and a router. A 100 in a sprint and a 100 in a route are not equal. Which is better depends on todays distance. You really need to make an asterisk on distance switchers and apply a totally seperate criteria to see if a sprinter can stretch out or a router can sprint.

Okay, so you've bounced the question to another thread. The question remains. Are you going to discuss the specifics of paceline selection or are you going to continue to obfuscate with how not to select pacelines? I suggest that you not dance the sidestep on this.

[douglasw32]

Dave that is kick ass !

[Light]

Okay, so you've bounced the question to another thread. The question remains. Are you going to discuss the specifics of paceline selection or are you going to continue to obfuscate with how not to select pacelines? I suggest that you not dance the sidestep on this.

I wrote what you quoted as a premise to my way of thinking. If you dont agree with them or at least are not open to what I wrote,then there is no point in me getting carpal tunnel syndrome trying to explain it to you. As you can see all that time on the Handi thread was fruitless except to inspire Jeff to start another thread. I'd rather not make the same mistake if you are not open to new ideas. I would like to hear your method as well. I'll just start with some stuff and see your reaction.

There has been other threads about pl selections. I think the majority here believe in just looking at the last 3 to select from. That's basically where I start but there are exceptions if we get that far. I want to know the best the horse can do. So I take the horses best race of the last 3. This puts all the horses on an equal level,no biasis, and it removes some of the "judgement or "art" form people think they need to pick a PL. It becomes mechanical and programable.

Second, I look at obvious things like I want a pl with a similar surface and distance.

Third and this is a big one because this is where I differ from alot of others,it has to be an appropriate class in claiming ranks. Non claiming are different. In claiming, if the horse's last 3 are like

1) speed fig 78 for 10K
2) speed fig 84 for 10k
3) speed fig 98 for 16K

And todays race is 10K, I am NOT going to take his 98 even though I take the best of the last 3. There are exceptions. But genearlly that is not the appropriate PL because the horse has been certified by his trainer that he is no longer a 16K claimer. That leaves 2 other PL's. I will give the horse the benefit of the doubt and use 2 back. Now if today's class is 25K,what paceline do you think I will use?

This is kind of sensitive stuff. But not many will change their style and most think I suck so I think I'm safe. Now you tell me what you do when you pick a PL.

[raybo]

I wrote what you quoted as a premise to my way of thinking. If you dont agree with them or at least are not open to what I wrote,then there is no point in me getting carpal tunnel syndrome trying to explain it to you. As you can see all that time on the Handi thread was fruitless except to inspire Jeff to start another thread. I'd rather not make the same mistake if you are not open to new ideas. I would like to hear your method as well. I'll just start with some stuff and see your reaction.

There has been other threads about pl selections. I think the majority here believe in just looking at the last 3 to select from. That's basically where I start but there are exceptions if we get that far. I want to know the best the horse can do. So I take the horses best race of the last 3. This puts all the horses on an equal level,no biasis, and it removes some of the "judgement or "art" form people think they need to pick a PL. It becomes mechanical and programable.

Second, I look at obvious things like I want a pl with a similar surface and distance.

Third and this is a big one because this is where I differ from alot of others,it has to be an appropriate class in claiming ranks. Non claiming are different. In claiming, if the horse's last 3 are like

1) speed fig 78 for 10K
2) speed fig 84 for 10k
3) speed fig 98 for 16K

And todays race is 10K, I am NOT going to take his 98 even though I take the best of the last 3. There are exceptions. But genearlly that is not the appropriate PL because the horse has been certified by his trainer that he is no longer a 16K claimer. That leaves 2 other PL's. I will give the horse the benefit of the doubt and use 2 back. Now if today's class is 25K,what paceline do you think I will use?

This is kind of sensitive stuff. But not many will change their style and most think I suck so I think I'm safe. Now you tell me what you do when you pick a PL.

IMO, lots of handicappers use your method, looking at the last 3 or 4 pacelines and finding the best race at or near the distance, on a similar surface and at a similar class.

One of the problems, I have experienced, with this approach comes in when the horse has had a layoff, especially a long one. The last 3 or 4 races may not be adequate for determining a representative paceline. Then you have to start going backwards in the older pacelines to try to find a paceline.

When programming a paceline selection method, I find it better to look at all pacelines first, looking for all representative pacelines (many times there are more than one that match your selection criteria). The method should either mark each qualified paceline or physically break them out from the rest of the pacelines (my method does both, eventually, marking all the qualified pacelines so the user can see which ones they are and when they occurred, in relation to the other pacelines, which enables one to see patterns prior to and after the qualified pacelines (including workouts and medication/equipment/jockey/trainer changes), making it easier to understand trainer methods and motivations/intentions and also helps to determine current form, due to trainer patterns, and associating them with today's race). Breaking them out into another view, disassociates them from the rest of the horse's races, making it impossible to put them in the perspective one has available before breaking them out.

An automated paceline selection method can then be programmed to key on these qualified pacelines, looking both prior to and after them, to further narrow down which one(s) is(are) more representative, regarding trainer intent and form. The method can then mark that(those) paceline(s) as it's selection(s). All this can be viewed within the same view as the PPs, so that one sees which paceline(s) are qualified and which the computer is using for the grade/ranking/etc..

Within the same view, preferably, one will have the ability to override the computer's selection(s) by marking his/her preferred paceline(s).

Once this is accomplished, then the final selected paceline(s) can be broken out from the PPs into a "summary" view, of all the horses in the race, containing only the selected paceline data, for each horse, the user wants to see, where horse to horse and horse to field comparisons can be accomplished more easily.

I hope I've made some sense here.

[ranchwest]

I wrote what you quoted as a premise to my way of thinking. If you dont agree with them or at least are not open to what I wrote,then there is no point in me getting carpal tunnel syndrome trying to explain it to you. As you can see all that time on the Handi thread was fruitless except to inspire Jeff to start another thread. I'd rather not make the same mistake if you are not open to new ideas. I would like to hear your method as well. I'll just start with some stuff and see your reaction.

There has been other threads about pl selections. I think the majority here believe in just looking at the last 3 to select from. That's basically where I start but there are exceptions if we get that far. I want to know the best the horse can do. So I take the horses best race of the last 3. This puts all the horses on an equal level,no biasis, and it removes some of the "judgement or "art" form people think they need to pick a PL. It becomes mechanical and programable.

Second, I look at obvious things like I want a pl with a similar surface and distance.

Third and this is a big one because this is where I differ from alot of others,it has to be an appropriate class in claiming ranks. Non claiming are different. In claiming, if the horse's last 3 are like

1) speed fig 78 for 10K
2) speed fig 84 for 10k
3) speed fig 98 for 16K

And todays race is 10K, I am NOT going to take his 98 even though I take the best of the last 3. There are exceptions. But genearlly that is not the appropriate PL because the horse has been certified by his trainer that he is no longer a 16K claimer. That leaves 2 other PL's. I will give the horse the benefit of the doubt and use 2 back. Now if today's class is 25K,what paceline do you think I will use?

This is kind of sensitive stuff. But not many will change their style and most think I suck so I think I'm safe. Now you tell me what you do when you pick a PL.

I'm somewhere in the vicinity of you and Zoot.

As Zoot does, I use a weighted method. I do include the class level, as you do.

However, I go back up to 10 races for the paceline race. Generally, needing to go back a lot of races when I give weight to recency suggests to me that I should be wary of that horse and I have some criteria related to that paceline as to whether the horse should be eliminated.

I factor in:

Class
Distance
Finish
Surface
Recency

That's about it.

[Red Knave]

Third ... it has to be an appropriate class in claiming ranks. ... In claiming, if the horse's last 3 are like

1) speed fig 78 for 10K
2) speed fig 84 for 10k
3) speed fig 98 for 16K

And todays race is 10K, I am NOT going to take his 98 even though I take the best of the last 3.
This is the kind of thing I was referring to in my original post. Whether you agree or not, I think that this is the 'art' part of paceline picking.
As well as your decision, in this example, to not use the 3rd back, I bet that you also use some non-mechanical way to determine whether to use the last line or the 2nd last; you make a decision that requires experience and judgement. To me, that's an art.

[headhawg]

As well as your decision, in this example, to not use the 3rd back, I bet that you also use some non-mechanical way to determine whether to use the last line or the 2nd last; you make a decision that requires experience and judgement. To me, that's an art.I couldn't agree with you more. I could easily make the argument that the 3rd race back the horse was loose on the lead and that the same pace scenario will happen again today. It would have nothing to do with trying to predict what's in the trainer's head -- that the horse is not worth $16K anymore and therefore that paceline is not a valid one.

I think that creating an automatic paceline selection algorithm that is really, really good is damn hard mostly because of the judgment factor involved. The ideas suggested here were very solid ones but (and I am not trying to insult anyone here) are any of them earth-shattering? (Now, maybe the super-secret algorithms that everyone has are but those didn't get posted. :) ) That was my point in the other thread -- can one honestly say that over the course of time that picking a single paceline is better than an averaging method as logical as it might sound? Just because every program seems to do it doesn't automatically make it the best method.

[raybo]

I couldn't agree with you more. I could easily make the argument that the 3rd race back the horse was loose on the lead and that the same pace scenario will happen again today. It would have nothing to do with trying to predict what's in the trainer's head -- that the horse is not worth $16K anymore and therefore that paceline is not a valid one.

I think that creating an automatic paceline selection algorithm that is really, really good is damn hard mostly because of the judgment factor involved. The ideas suggested here were very solid ones but (and I am not trying to insult anyone here) are any of them earth-shattering? (Now, maybe the super-secret algorithms that everyone has are but those didn't get posted. :) ) That was my point in the other thread -- can one honestly say that over the course of time that picking a single paceline is better than an averaging method as logical as it might sound? Just because every program seems to do it doesn't automatically make it the best method.

Totally agree!

All I'm saying is, if you're going to go to the trouble and headaches to program an auto-selection method, at least have it looking at all the available data, not just the last 3 or 4 races. And, have it as complete as possible so that all the user has to do is check a couple of things and either accept what the program says or reject it in favor of another paceline(pacelines, if you want to do some averaging or finding a median, etc.).

[Light]

I could easily make the argument that the 3rd race back the horse was loose on the lead and that the same pace scenario will happen again today. It would have nothing to do with trying to predict what's in the trainer's head -- that the horse is not worth $16K anymore and therefore that paceline is not a valid one.


The problem with that philosophy is if the trainer thought the horse could still wire a field of 16K claimers,(if that's what he did 3 back),he wouldn't put him in for 10K. You might have $20 at stake in the 10K race. But the owner and trainer are already conceeding a $6k loss. That speaks volumes.

My paceline method is based as closely as possible to what the trainer is telling me in the pp's,and I'm not going to second guess the trainer since he is the best person qualified about where to place his horse.Yes he is wrong at times,but mostly right. That's his job.

[headhawg]

Look, I think that interpreting trainer intent is as important of a factor as the next 'capper but if you're going to say that it definitely outweighs pace/race shape that we'll just have to agree to disagree. There are many $10K races that are stronger than those of the $16K variety. And I am glad that your computer program can read the trainer's mind. I would still prefer to know how the number was earned -- choosing a paceline should be contextual. If you can't do that then the method is flawed. And if a class match was what you based your decision on then why bother listing the speed figs?

I also find it very interesting that you waited for one of my posts before you responded in this thread again even though you have visited this thread since your response to ranchwest. If you're trying to prove that you're a better programmer or handicapper or that you're smarter than me you have a LOOOONG way to go.

[DRIVEWAY]

The problem with that philosophy is if the trainer thought the horse could still wire a field of 16K claimers,(if that's what he did 3 back),he wouldn't put him in for 10K. You might have $20 at stake in the 10K race. But the owner and trainer are already conceeding a $6k loss. That speaks volumes.

My paceline method is based as closely as possible to what the trainer is telling me in the pp's,and I'm not going to second guess the trainer since he is the best person qualified about where to place his horse.Yes he is wrong at times,but mostly right. That's his job.

Many times a trainer will drop a horse prior to a logical break in training. The difference in claiming price can easily be made up by not having to pay the bills for 3-4 months.

Pick up a check in the race, get the claiming tag and save on bills for 3-4 months. Not a bad strategy. If we only knew in advance!!

[Light]

Look, I think that interpreting trainer intent is as important of a factor as the next 'capper but if you're going to say that it definitely outweighs pace/race shape that we'll just have to agree to disagree.

The PL you pick will carry the pace/race shape for the horse in question.So if you pick an obsolete PL,you also get obsolete pace figs.

There are many $10K races that are stronger than those of the $16K variety.

Yeah,its called reading the conditions.

And I am glad that your computer program can read the trainer's mind.

I prefer to pick my own PL and have the computer ask me which one I want to use.

I also find it very interesting that you waited for one of my posts before you responded

Yeah, like I knew you would respond and was just waiting for that opportunity. :rolleyes: You think alot of yourself don't you. Like your statement here:

If you're trying to prove that you're a better programmer or handicapper or that you're smarter than me you have a LOOOONG way to go.

Give me a break.

[Light]

Many times a trainer will drop a horse prior to a logical break in training. The difference in claiming price can easily be made up by not having to pay the bills for 3-4 months.

Even if the horse is idle,I believe its around $1500-$2000 a month for room and board,medical, recreational and training. With a cheap claimer,3-4 months of not racing will have his bills outweigh his claiming price.

[headhawg]

The problem with that philosophy is if the trainer thought the horse could still wire a field of 16K claimers,(if that's what he did 3 back),he wouldn't put him in for 10K. You might have $20 at stake in the 10K race. But the owner and trainer are already conceeding a $6k loss. That speaks volumes.
Even if the horse is idle,I believe its around $1500-$2000 a month for room and board,medical, recreational and training. With a cheap claimer,3-4 months of not racing will have his bills outweigh his claiming price.So let me get this straight. Even though the pace scenario says use the $16K race you're going to use the $10K because the trainer thinks/knows that the horse is not worth 16K now. Yet, if the horse gets claimed for 10K plus no bills for three or four months.... Let me do the math. $1500 to $2000 X 4 is...I don't know...about $6000 to $8000? Plus the claiming price plus the winnings.... Yep. Go with the 2nd race back because trainer intent is easier to predict than pace shape, mathematically speaking. :rolleyes:

[Light]

I have no idea what you are saying.

[headhawg]

I'm saying that your response to Driveway supported his assertion which basically refuted your conclusion that the connections were conceding a $6K loss.

Never mind, though. Keep doing what you're doing.

[headhawg]

The PL you pick will carry the pace/race shape for the horse in question.So if you pick an obsolete PL,you also get obsolete pace figs.Then why not just use the latest paceline then?

Yeah,its called reading the conditions.They don't necessarily tell the whole story. Ever hear of a key race for example?
I prefer to pick my own PL and have the computer ask me which one I want to use. Why does your program need to confirm a paceline that you've already selected? :confused:
Yeah, like I knew you would respond and was just waiting for that opportunity. :rolleyes: You think alot of yourself don't you. Just making an observation. Nope, my ego isn't that big. I actually try to be helpful where and when I can.

[Light]

I'm saying that your response to Driveway supported his assertion which basically refuted your conclusion that the connections were conceding a $6K loss.


Well his premise is wrong. You dont get free room and board at a racetrack for your horse.

[Tom]

There is nor right or wrong. You will hit some hand he will hit others. Both ideas work some times. Your assertion that your way is the only one that works is absurd.

[headhawg]

Well his premise is wrong. You dont get free room and board at a racetrack for your horse.Reread Driveway's post. I'm pretty sure that's not what he wrote nor what he implied. Then, when you have a moment of clarity reread my post stating that your response appeared to substantiate his.

And...what Tom said.

[Jeff P]

A while back I was putting together material for a handicapping seminar... The presentation itself is structured in such a way that it could be done as either a "light" 45-60 minute piece as part of a HANA event... or as a "heavy" full on data intensive 2-3 day JCapper seminar.

In my opinion, today's successful player needs to have reliable repeatable methods for:

1. Determining probability.

2. Determining value.

3. Play or pass decision making.

4. Backing an opinion to maximize returns.

5. Bet sizing/growing a bankroll.

6. Record keeping.

7. R&D/constantly working to improve each area of your game.

In addition, the ability to find better pricing helps.




So where does selecting pacelines fit into this?

I've spent a lot of R&D time over the years subjecting just about every method of selecting pacelines I could imagine to hard database testing.

Based on my own R&D using large data samples, I came to the conclusion that paceline selection isn't nearly as important as a lot of players think it is.

When I compared win rate and flat win bet roi of the BEST performing paceline selection methods I was able to produce vs. the worst performing paceline selection methods I was able to produce... the results kind of surprised me.

Provided that the same set of rules for picking lines was uniformly applied to every horse in every race... differences between the best and worst performing sets of paceline rules were surprisingly small.

Remember... this is a parimutuel game.

It makes perfect intuitive sense to use the most logical lines... recent ones that best match today's conditions.

But if everyone is drawn to the same lines... What do you think happens to the win mutuels of the horses possessing the lines that everyone is drawn to?

Maybe there are artists out there who are somehow drawn to lines that everyone else overlooks. I'm not saying it can't be done.

What I AM saying is that I spent considerable time evaluating paceline selection and came to the conclusion that selecting pacelines does not offer a magic bullet... at least not for me. For me, paceline selection is just one very tiny piece of the puzzle... one that can be incorporated into a much larger comprehensive selection process.

I actually use algorithms that rate and score every running line... and them roll the weighted scores together into multiple compound numbers that represent "ability from speed figs."

The factors used to represent "ability from speed figs" in turn become just one component (about 1/7th) of an overall selection process.

All that number crunching... Sounds complicated doesn't it?

Wait -- it gets worse.

The selection process itself is just one piece of the first required skill... Determining Probability.

There's actually a purpose behind all of the number crunching.

Simplicity.

In the end, algorithms in the software boil everything down to 1's and 0's... a precise Yes No decision about each horse:

1 = Yes = Good Bet = Play

0 = No = Bad Bet = Pass

That takes care of skills #1, #2, and #3 from the above list... freeing me up on race day to give focus to the other skills on my list.

Back to the point I'm trying to make.

When paceline selection is used as just one piece of a much larger comprehensive selection process, results from large data samples show that paceline selection doesn't have the ability to materially affect the quality of my Yes = Good Bet and No = Bad Bet decisions.


-jp

.

[DanG]

Based on my own R&D using large data samples, I came to the conclusion that paceline selection isn't nearly as important as a lot of players think it is.

Provided that the same set of rules for picking lines was uniformly applied to every horse in every race... differences between the best and worst performing sets of paceline rules were surprisingly small.

(#47) Not a good post…a GREAT post imo. Quoted a small section in the interest of bandwidth.

I couldn’t agree more in terms of the #1 priority concerning pace line selection is consistency. In terms of volume betting; ironically I would back the player who was consistently “wrong” (if there is such a thing) over the player who flew by the seat of their wallet picking lines. Paceline selection more then most areas of capping lends itself to being virtually 100% rule based.

Once your method of PL selection is established (and it may be very complex); it allows for further analysis….form, trip, competition level etc…but the initial pace line process that sets your software / paper and pencil in motion should be like a drum machine with one setting. It should provide the same beat…over and over.

[ranchwest]

(#47) Not a good post…a GREAT post imo. Quoted a small section in the interest of bandwidth.

I couldn’t agree more in terms of the #1 priority concerning pace line selection is consistency. In terms of volume betting; ironically I would back the player who was consistently “wrong” (if there is such a thing) over the player who flew by the seat of their wallet picking lines. Paceline selection more then most areas of capping lends itself to being virtually 100% rule based.

Once your method of PL selection is established (and it may be very complex); it allows for further analysis….form, trip, competition level etc…but the initial pace line process that sets your software / paper and pencil in motion should be like a drum machine with one setting. It should provide the same beat…over and over.

In this vein, I find that my handicapping adapts to the paceline selection I have. So, I think you're right. Being consistent in paceline selection lends itself to molding the rest of handicapping around the paceline method.

[headhawg]

Excellent post Jeff, and a little bit surprising especially the part about minimal differences between the best and worst performing paceline selection methods. I especially liked your concluding paragraph which is what I -- and likely many others -- had previously thought. It's nice to see an analysis done to confirm that though. Thanks.

[46zilzal]

Why use ONE paceline and have possible sample error? Horses that are good enough to consider today would have shown the same ability historically.

Pizzolla's form cycle window has shown this repeatedly

[headhawg]

Why use ONE paceline and have possible sample error? Horses that are good enough to consider today would have shown the same ability historically.

Pizzolla's form cycle window has shown this repeatedlyUm...isn't the form cycle window intended to give rules (or perhaps guidelines is more correct) to find a single paceline to use? Are you saying that MP now prefers a multiple paceline/averaging(?) approach?

And I'm no statistics guru but there's no guarantee that using multiple lines is going to decrease that sampling error.

[Tom]

I think paceline is the most critical part of my handicapping. Using the wrong lines for pace based handicapping is disasterous. I can't agree with averaging line either - you end with something the horse never ran in reality.

[harntrox]

I compare the 2 most different approaches and get definite results:

Optimistic: use the horses last win within 120 days as the paceline
or
Pessimistic: use the last race (at any distance) that wasnt a win

No matter how different the paceline race was to todays conditions, you make the square peg fit in the round hole by normalizing everything to the current conditions. I believe this was the Brohammer approach, and is why some call this skill an art.

Then you are comparing an actual race that was run and basically linearly interpolating the positions of the fractions of the pace - (with all entries being modified consistently) - to fit todays race.

The other reason i like to use the 'optimistic' line is (imho) it shows the intent of the trainer. In that he obviously knows thats the best the horse can do, and if its entered into a race of this class, the trainer thinks he can win. The usual jockey, trainer patterns can then be used to decide if the horse can come back, usually at long odds if its been a few races since a win.

[ranchwest]

I compare the 2 most different approaches and get definite results:

Optimistic: use the horses last win within 120 days as the paceline
or
Pessimistic: use the last race (at any distance) that wasnt a win

No matter how different the paceline race was to todays conditions, you make the square peg fit in the round hole by normalizing everything to the current conditions. I believe this was the Brohammer approach, and is why some call this skill an art.

Then you are comparing an actual race that was run and basically linearly interpolating the positions of the fractions of the pace - (with all entries being modified consistently) - to fit todays race.

The other reason i like to use the 'optimistic' line is (imho) it shows the intent of the trainer. In that he obviously knows thats the best the horse can do, and if its entered into a race of this class, the trainer thinks he can win. The usual jockey, trainer patterns can then be used to decide if the horse can come back, usually at long odds if its been a few races since a win.

Whew! I don't make that sort of assumption about a trainer! Trainers don't always place their horses to win, at least not from what I've seen.

Trying to compare different surfaces, for instance, is apples and oranges. For instance, horses without turf conformation are not likely to show their dirt ability on turf.

[raybo]

In this vein, I find that my handicapping adapts to the paceline selection I have. So, I think you're right. Being consistent in paceline selection lends itself to molding the rest of handicapping around the paceline method.

This is exactly why I have an automated paceline selection system, consistency.

However, this is only the first step in finalizing the paceline selection. For instance, when the automated process selects an obviously bogus paceline, bells and whistles should go off in your head. If your automated process has been thoroughly analyzed and correctly programmed and it comes up with a bad paceline then you know immediately that, either there are no qualified pacelines (a good thing to know about the animal as it relates to his/her qualifications to run in today's race), or something you may not readily see, without digging deep, caused your thoroughly programmed process to pick that line for a reason (again, maybe you need to dig deep to see why, it may be exactly the paceline you should be using, for a variety of possible reasons).

The automated selection process only sets the base or the bar, you either accept it or raise or lower it.

But, the process forces consistency and lends itself well to providing a system of checks and balances, regarding your overall approach, if nothing else.

[46zilzal]

Um...isn't the form cycle window intended to give rules (or perhaps guidelines is more correct) to find a single paceline to use? Are you saying that MP now prefers a multiple paceline/averaging(?) approach?

And I'm no statistics guru but there's no guarantee that using multiple lines is going to decrease that sampling error.
RULES? who follows rules?

[headhawg]

Computers do, 46. Computers. They don't do what you want them to do they only do what you tell them to do. They need rules. This thread is entitled Paceline Selection Algorithms. Try to stay on topic.

I also noticed that you didn't answer my reply. Are you going to bring up median energy here too? Never mind. I don't need to read your response.

[Warren Henry]

What is the best pace line to use, or which is the best way to calculate the actual ratings, or which method produces the BEST speed rating. Would it be blasphemy to suggest that IT DOESN'T MATTER. (Jeff may have been saying that). Tom speaks the truth when he says that some work one time and others work another. What does matter is which ones produce the BEST ROI.

I personally never used pace in picking my contenders. I only used pace to make sure that my overlay selection based on other factors wasn't going to be significantly disadvantaged by a likely pace scenario in the upcoming race. That I could usually do at a glance.

Today, everyone is caught up in pace and speed numbers, but there may be better value elsewhere. I am absolutely sure that on certain days (based on weather conditions), I can go to the track and use the old 6-2-1 purse analysis method and make more money than the best speed and pace players. That old method worked great in its day. The only reason it fell out of favor back then was that everyone ended up using it and the selections became the chalk.

All you have to do to make money at this game is have sound/consistent methods to select legitimate contenders that do not use the exact same stuff everyone else is using.

[Light]

Here's an example of what I'm talking about. Haw 12/20 /09 Race #8 1 1/16

http://i333.photobucket.com/albums/m376/Light1011/Horseracing/droplet.gif

This horse went off at 4/5 and finished 5th of 8. Why? Because of poor paceline selection by the public. Every other horse in this field had at best low 70's speed ratings. The public incorrectly judges this horse like many do here without regard to present form or class as towering over this field with consistently high 70's speed ratings.

My paceline selection says this horse gets a 71 speed rating which fits in the middle of the pack with the rest of this field's speed ratings and is where she finished.You guys here would have disregarded the class factor, averaged her last several races and had her,like the public, towering over the field. The trainer dropped her to this new low level and she should have romped. Instead she ran the worst speed rating of the last 5 showing with no trouble in the race. This is not the same horse as she was when she won for 10K. Therefore using anything but that last paceline to evaluate this horse on a numbers basis would be wrong. And that's my only beef with paceline picking. I dont care if you average or not . Take 1 race or 10. But you must take races that are relevant to where the horse is at today.

This does not take skill nor an art. It's common sense. If a horse is running at a lower class,you cannot take his races from higher classes for paceline computations without probable error. Otherwise,if you think horses can run those figures from those higher class races you should claim the horses yourself and jump them up in class. Apparently no trainer or owner was dumb enough to claim Droplet with her "towering figs". :rolleyes: This is the folly of mindless paceline averaging.

[Dave Schwartz]

Howard Sartin used to say that there were three things you had to do to win:

1. Pick the right contenders.
2. Pick the right pacelines.
3. Interpret the readouts (i.e. model) correctly.


I have found this to be wise council.

IMHO the most representative paceline is the race where a horse won or lost by a nose. If he was any better he would have won (or won by a larger margin). This race represents just how good the horse really is when he runs.

That "when he runs" part is important. Will he run well today? That should be a function of your contender selection.


Of course, Howard also said that 90% of the time you should use one of the last two lines, and I totally disagree with that. That is not really "pace" handicapping; that's "recent form" handicapping because the big recent race will almost always be your play(s).

Just my opinion.

Regards,
Dave Schwartz

[Jeff P]

Would it be blasphemy to suggest that IT DOESN'T MATTER. (Jeff may have been saying that).Warren, that's exactly what I was saying.

Keep records. Analyze your records and understand the strengths and weaknesses of your own selection process. Apply your selection process in a consistent manner and make informed play or pass decisions based on your records... Do that, and paceline selection becomes just one tiny piece of a much larger workable process.

The very best players I know, whether they realize it or not - do a reasonably good job of the following:
1. Determining probability.

2. Determining value.

3. Play or pass decision making.

4. Backing an opinion to maximize returns.

5. Bet sizing/growing a bankroll.

6. Record keeping.

7. R&D/constantly working to improve each area of your game.

In addition, the ability to find better pricing helps.



-jp

.

[headhawg]

This does not take skill nor an art. It's common sense. If a horse is running at a lower class,you cannot take his races from higher classes for paceline computations without probable error. Otherwise,if you think horses can run those figures from those higher class races you should claim the horses yourself and jump them up in class. Apparently no trainer or owner was dumb enough to claim Droplet with her "towering figs". :rolleyes: This is the folly of mindless paceline averaging.First of all thanks for choosing this race to make your point. Using Bris files, HHX v2, and the default weights, Handifast listed the top 4 with Fair Odds as follows:

#2 Droplet 1.8
#3 Miss Josefine 5.1
#5 Indian Love Cry 5.5
#7 Duchess Eowyn 6.4

The 2 and 3 both went off below the fair odds which left only two possible plays -- the 5 and 7. Normally when I find two possible overlays in the Top 4 I look at some specific ratings. As I believe in winning connections the trainer/jockey combo on the 5 was better than the 7. I also am a big fan of the FAST number and again, the 5 rated higher than the 7. So my play was the 5. Nice winner at $15.00.

Now you are going to call this redboarding. Here's the difference -- Light picked this race and anyone who has HHX v2 can get the same results for real, not with some imaginary program that he has. (He posts a screen shot of the Bris PP generator??) Yeah this paceline averaging thing doesn't work. :rolleyes: And the users and programmers have repeatedly posted that it is not a black box, just a tool. You have to find out what works for you.

Light has now in three threads made attacks and insults about Handifast and its programmers because he thinks he and his method is superior. They are not. Jeff has posted that in his research he found that there was minimal differences in paceline selection. And I would like to post what I really think of him but the smart members here already know and I don't want PA to close this thread as there is very good information here.

And Light, if you use "common sense" in handicapping you're destined to always lose but that's ok by me.

[46zilzal]

Howard Sartin used to say that there were three things you had to do to win:

1. Pick the right contenders.
2. Pick the right pacelines.
3. Interpret the readouts (i.e. model) correctly.


Note the PLURAL above

[Tom]

HH, I can agree with averaging lines in the manner you do here - and then looking at individual lines as you suggest. For the pace evaluation, I want single lines, but the averaging serves a different purpose.

Looks like Handi is pretty decent program! :ThmbUp:

[Tom]

Light, in your example, what is the distance and surface of today's race?
That would determine my PL selection more than anything.

[Light]

Using Bris files, HHX v2, and the default weights, Handifast listed the top 4 with Fair Odds as follows:

#2 Droplet 1.8
#3 Miss Josefine 5.1
#5 Indian Love Cry 5.5
#7 Duchess Eowyn 6.4

The 2 and 3 both went off below the fair odds which left only two possible plays -- the 5 and 7.

We already have Hcap's little study of your fair odds method and it only broke even.

Secondly,your fair odds are not accurate and I have no idea of their rhyme or reason but paceline selection obviously affects them. If you used my paceline method,Droplet's fair odds would have been much higher. The winner's fair odds would have been much lower as this horse had what I consider the strongest angle in horseracing which I will not divulge.

Third,using fair odds is useless for exotic players who will be using races in serial bets before the money comes down on them.

Fourth,I dont know anyone who could be so disciplined using fair odds that they would consistently bet this way. Furthermore we all know a large amount of money comes in as the horses are loading into the gate,so this form of betting is impractical in many cases since you can never be sure your 3-1 horse didn't plummet to 2-1 when the gates opened with a 5/2 fair odds.



Light has now in three threads made attacks and insults about Handifast and its programmers because he thinks he and his method is superior.

You consider a difference of opinion an attack? Do I say you are attacking me because you disagree with me? No. Grow up.

Jeff has posted that in his research he found that there was minimal differences in paceline selection.

I beg to differ. My program has 2 paceline options and one totally outperforms the other and I already demonstrated how my PL selection has outperformed yours in one random weekend @ Aqueduct. I'm sure there are PL methods where one picks one Pl and the other averages Pl's and there is minimum differences. But they are probably both sound methods that discard obvious PL errors. I think Jeff might have left that little detail out of his statement.

[Light]

Light, in your example, what is the distance and surface of today's race?


It's the same as Droplet's last race.

[Jeff P]

But they are probably both sound methods that discard obvious PL errors. I think Jeff might have left that little detail out of his statement.
Maybe I should clarify....

1. The numbers (pace figs/velocity/incremental velocity for E1, E2, F3, Final Time, etc.) derived from the selected paceline(s) represent a very tiny part of my overall selection process. For players using info derived from pacelines as part of a larger comprehensive approach, paceline selection has very little effect on overall results. Consistency in line selection is important. Large data samples bear this out.

2. If info derived from the selected line plays a major role in the player's selection process -- That's different... especially if the player is manually selecting lines.

That said, if you are getting good results using an approach similar to what I described in #2 above... my hat is off to you -- and by all means keep doing it.

You are doing something I have never been able to do -- which is the reason I migrated to approach #1.

Light... If you keep records and carefully analyze them and apply your selection process in a consistent manner and make informed play or pass decisions based on your records... you just might be surprised how little difference there is in the "long run" (over 10's of thousands of plays) from one paceline selection method to another.


-jp

.

[Jeff P]

Ran out of "edit time" on the above post...

Start with this: Light... If you keep records and carefully analyze them and apply your selection process in a consistent manner and make informed play or pass decisions based on your records... you just might be surprised how little difference there is in the "long run" (over 10's of thousands of plays) from one paceline selection method to another.
And add this...

What I'm really trying to say is that the selection process is just a tiny part of the overall required skill set. In my opinion, a good play or pass decision making process is FAR more important than the selection process.

Knowing when to pull the trigger -- and when to sit on your hands is far more important in determining whether a player wins or loses in the long run than selecting pacelines.

-jp

.

[DRIVEWAY]

We already have Hcap's little study of your fair odds method and it only broke even.

Secondly,your fair odds are not accurate and I have no idea of their rhyme or reason but paceline selection obviously affects them. If you used my paceline method,Droplet's fair odds would have been much higher. The winner's fair odds would have been much lower as this horse had what I consider the strongest angle in horseracing which I will not divulge.



The horse's pattern is as follows.
Same Jockey, Trainer, Distance and Track.
Two back was 5000nw2 then minimum/mandatory class rise to 5000nw3 and today the same 5000nw3.
Two back post 5(Good Post) last race post 10(Bad Post) and today post 5(Good Post).
Results are win, lose, win.
My guess at Strongest Angle In Racing is Good-Bad-Good post when all other factors are essentially the same.
I'm probably wrong but maybe someone else can figure it out.

[Tom]

It's the same as Droplet's last race.
That would be the only one I would use, then. I never mix dirt and artificial and prefer to not use off track whenever possible.

[headhawg]

We already have Hcap's little study of your fair odds method and it only broke even.And we have zero tests of yours except what you tell us. And I believe everything you say as gospel. :rolleyes:
Secondly,your fair odds are not accurate and I have no idea of their rhyme or reason but paceline selection obviously affects them. If you used my paceline method,Droplet's fair odds would have been much higher. The winner's fair odds would have been much lower as this horse had what I consider the strongest angle in horseracing which I will not divulge.You've said some dumb things already but this one is killer -- how can you say that another person's fair odds are not accurate? That is beyond lame. And Ooh. Your secret angle. Wow I'm impressed. Meaningless drivel again. :bang:

Third,using fair odds is useless for exotic players who will be using races in serial bets before the money comes down on them.And now your indicting the fair odds method which is a whole other debate.

Fourth,I dont know anyone who could be so disciplined using fair odds that they would consistently bet this way. Furthermore we all know a large amount of money comes in as the horses are loading into the gate,so this form of betting is impractical in many cases since you can never be sure your 3-1 horse didn't plummet to 2-1 when the gates opened with a 5/2 fair odds.I didn't figure you to be disciplined. Anyway, I've read and understood money management recommendations from Mitchell and Meadow. It would be built into my plays.
You consider a difference of opinion an attack? Do I say you are attacking me because you disagree with me? No. Grow up.Do I need to quote all your posts from the three threads? This is way beyond a difference of opinion. To wit, how come you are not incessantly pounding on Jeff P for his research and conclusions? Or sjk? Or others who disagree with your phenomenal paceline selection method? It's become personal and you made it that way.

So I present to you a challenge which has been a long time coming. Handifast's picks vs yours. Name the track, dates, and we'll use the Top 4 picks. Posted in advance like the TLG vs cmoore thread.

[ranchwest]

The horse's pattern is as follows.
Same Jockey, Trainer, Distance and Track.
Two back was 5000nw2 then minimum/mandatory class rise to 5000nw3 and today the same 5000nw3.
Two back post 5(Good Post) last race post 10(Bad Post) and today post 5(Good Post).
Results are win, lose, win.
My guess at Strongest Angle In Racing is Good-Bad-Good post when all other factors are essentially the same.
I'm probably wrong but maybe someone else can figure it out.

You are assuming facts not in evidence. We don't really know whether we're looking at the "Strongest Angle in Racing" or not.

[headhawg]

No one has mentioned the obvious. When I started playing the horses in the early 80"s, it was very very difficult for a 10K claimer to go up against 16K claimers. Or a 16K claimer to go into allowance company. There was a definite division in quality of competition back then amongst the classes. Now horses intertact in the class heirarchy with greater ease. At cheaper tracks even sharp 4K claimers regularly romp in allowance company. Something totally unheard of when I started. The "class" factor has changed in meaning and complexity and is not as clearly defined as when Quinn wrote "class of the field."

The problem with that philosophy is if the trainer thought the horse could still wire a field of 16K claimers,(if that's what he did 3 back),he wouldn't put him in for 10K. You might have $20 at stake in the 10K race. But the owner and trainer are already conceeding a $6k loss. That speaks volumes.
With regard to your paceline selection algorithm, please reconcile these two apparently contradictory views.

[douglasw32]

Here ... http://localpcpros.com/tam122309.pdf

Is one of those whacky average paceline summary's

BUT... I took the advice here, and since I never use any of my programs as a "black box"

I did what I normally do and considered the top 4 averages as "Win Contenders"

Then I filtered the bris PP generator to show me only "Like" races, same surface, distance, fast tracks, etc

Then took the most recent as the "Paceline" double checked it could hold up against the assumed "pace set up"

And came up with a single final selection.

I am not sure if anyone will be interested in following along but it was completely mechanical.

So we will see how I do.

Highlighted are the top 4 (including ties) with the final choice based on a single, comparative, recent paceline highlighted all the way across.

The readout is what I have evolved the "learnings" of handifast into.

[Light]

Maybe I should clarify....

1. The numbers (pace figs/velocity/incremental velocity for E1, E2, F3, Final Time, etc.) derived from the selected paceline(s) represent a very tiny part of my overall selection process.

I could see that,but I think the vast majority of players base their decisions and selections on pacelines and the data within them.

Light... you just might be surprised how little difference there is in the "long run" (over 10's of thousands of plays) from one paceline selection method to another.

Certainly time and thousands of races can cause some differences to become smaller simply because of being in the context of massive numbers.But each day we play we are not in that context. We are in the context of a small sample with a greater impact.

The greater impact comes from seeing one PL method that ranks the winning horse 2nd over another that ranks him 5th. This certainly affects your selection process. It may not show up in your study of thousands of races of how the top pick in a race did, but it will show up in the wallet of a user who has better overall information due to a better PL selection method.

[Light]

The horse's pattern is as follows.
Same Jockey, Trainer, Distance and Track.
Two back was 5000nw2 then minimum/mandatory class rise to 5000nw3 and today the same 5000nw3.
Two back post 5(Good Post) last race post 10(Bad Post) and today post 5(Good Post).
Results are win, lose, win.
My guess at Strongest Angle In Racing is Good-Bad-Good post when all other factors are essentially the same.
I'm probably wrong but maybe someone else can figure it out.

You'll never guess it.

[Light]

You are assuming facts not in evidence. We don't really know whether we're looking at the "Strongest Angle in Racing" or not.

You're right. Its debatable whether it's the strongest angle in horseracing, but it certainly is mine.

[ranchwest]

You'll never guess it.

Ah, THE biggest shocker of 2009.

[headhawg]

Certainly time and thousands of races can cause some differences to become smaller simply because of being in the context of massive numbers.But each day we play we are not in that context. We are in the context of a small sample with a greater impact.

The greater impact comes from seeing one PL method that ranks the winning horse 2nd over another that ranks him 5th. This certainly affects your selection process. It may not show up in your study of thousands of races of how the top pick in a race did, but it will show up in the wallet of a user who has better overall information due to a better PL selection method. Are you serious with this rambling? But of course -- don't let solid research over a large sample get in the way of the Light philosophy. Do you actually type this stuff with a straight face??

[Light]

With regard to your paceline selection algorithm, please reconcile these two apparently contradictory views.

Just because I said "Now horses intertact in the class heirarchy with greater ease" doesn't mean you can use a higher class PL without error like in Droplet's case. I was only inferring that the quality of racing has gone down and horses can climb the class ladder with less resistance. I wasn't inferring that its OK to override the wisdom of where the trainer is placing his horses,especially when going down the class ladder like Droplet.

[Light]

Do you actually type this stuff with a straight face??

Sort of.

[Light]

Ah, THE biggest shocker of 2009.

Alright,I'll sell you my secret for just $19.99. And if you order now I'll throw in a complete set of Ginsu knives. But wait. That's not all...

[ranchwest]

Alright,I'll sell you my secret for just $19.99. And if you order now I'll throw in a complete set of Ginsu knives. But wait. That's not all...

Every record ever made.

[DRIVEWAY]

You'll never guess it.

Then perhaps a sporting man such as yourself could give us a clue or two.

[douglasw32]

http://localpcpros.com/tam122309rev1.pdf (http://localpcpros.com/tam122309.pdf)

Updated.

[Gallop58]

Tangential post here on the original subject of pacelines, art vs science....

I always liked this quote and thought it might be an appropriate way of framing the paceline perspectives...

"Engineers want to be creative everywhere. The trick is pushing them to be creative in the art part and making the rote part so routine that they don't get creative there at all."
- Jon Rubenstein

Is a handicapper an engineer or an artist? To each his own. What works is real.

Good Racing.

[Jeff P]

The greater impact comes from seeing one PL method that ranks the winning horse 2nd over another that ranks him 5th. This certainly affects your selection process. It may not show up in your study of thousands of races of how the top pick in a race did, but it will show up in the wallet of a user who has better overall information due to a better PL selection method.Light, before I respond to this I want to say, please understand that I am not picking on you. If what you are doing is working for you and you are happy with your results -- by all means keep doing it!

That said, if a player really does have "better overall information due to a better PL selection method"... or any better method for that matter...

I'll make the argument that evidence of having a better method absolutely WILL show up when win rate and flat win bet roi produced by different methods is compared in large data samples.

-jp

.

[Light]

I'll make the argument that evidence of having a better method absolutely WILL show up when win rate and flat win bet roi produced by different methods is compared in large data samples.



I agree with the above. But the reason I disagreed with your statement about minimal differences in PL methods is because you said it as a blanket statement without reference to the criteria used in the study.

[PaceAdvantage]

So I present to you a challenge which has been a long time coming. Handifast's picks vs yours. Name the track, dates, and we'll use the Top 4 picks. Posted in advance like the TLG vs cmoore thread.Now THIS sounds like fun...something I would enjoy watching....

Come on Light...you're not going to pass on TWO straight challenges, are you?

[Speed Figure]

Now THIS sounds like fun...something I would enjoy watching....

Come on Light...you're not going to pass on TWO straight challenges, are you?
It will NEVER HAPPEN!

[Jeff P]

I agree with the above. But the reason I disagreed with your statement about minimal differences in PL methods is because you said it as a blanket statement without reference to the criteria used in the study.
The criteria for line selection rules that I tested in my R&D involved every possible combination that I could dream up using the following variables:

surface, distance, number of races back, number of days back, and loose class definitions

Believe me, going in I had hoped to discover a method or at least a thought process for line selection that would lead to improved roi. Testing the above criteria in large samples was a dry well... at least as far as finding significant differences. Small differences... 1/2 to 3/4 point or so in roi - yes. Material differences... 3 to 4 or more roi points - certainly not.

The improved roi eventually came... but it wasn't paceline selection driven.


-jp

.

[46zilzal]

Horses should be evaluated as PERFORMERS (akin to what Timeform ratings do) and not PERFORMANCES to get rid of sample error.

All the time I see other Sartinistas base their entire evaluation of a PERFORMER on a single PERFORMANCE and then wonder why the horse does nothing.

What can the PERFORMER do when challenged by a particular pace? You have MULTIPLE examples to see how this animal did against similar paces. Using more that one example tells you what the horse can, could have done, or did to elucidate form cycle and overall direction of ability.

Settling on a single pace line to establish what the horse can/may do is akin to deciding what a pitcher or quarterback CAN POSSIBLY do based upon watching a single game with all of it's inherent specificity.

[Vinnie]

Now THIS sounds like fun...something I would enjoy watching....

Come on Light...you're not going to pass on TWO straight challenges, are you?


PA, Speed, and HH! I get the strong feeling (for some odd reason) that such an event will never transpire. I sense the white flag being placed on the stick and it is about to start waving. :)

[ranchwest]

PA, Speed, and HH! I get the strong feeling (for some odd reason) that such an event will never transpire. I sense the white flag being placed on the stick and it is about to start waving. :)

They seem to be content flowing through their own competitions with changes by the minute.

[douglasw32]

If anyone has the time, I really would love to learn this aspect of handicapping, perhaps I picked a bad day, bad track, or the wrong pacelines.

But.. in the example PDF I posted of software that only acerages pace and speed, combines it with current form rules and position to the horses ran against and the ratings they produced.

I went a step further and looked at a different source (bris produced my ratings) I used the Post Time daily speed figures, looked at the most recent race that was the same distance, same surface, same condition, and if there was an obvious reason to toss it went back one more (rarely)

And I has the most disapointing day I can remember....

I am not sure but I noticed a number of times through the day had I simply stuck with the best figure (based on an average of lines) it would have out performed adding in the single comparative paceline as a decision making factor.

If someone could look at the pdf, find pacelines and see if they come up with a different single horse per race than I did , that worked out, I would be very interested in continuing to use such a decision model.

until then I will stick to taking my power rating at face value and deciding if it is at odds by running style with the way the race looks to shape up then go to the 2nd best total rating.

Please compare my work and see if I did something wrong that can prove using a single line out performed the averaged one.

As for the challeng, I do not see how it would prove anything even though it would be amusing, it would be like a handicapping tournament where with rules one lucky pick could simply launch someone so far ahead they never can catch up.

There would need to be solid, fair, provable rules set forth.

But if so I would like to take on my earlier work that has been tweaked along with Light.

so we would actually have three methods (two based on average lines), three handicappers, it would only be fair if anyone else joined in using paceline selection and see if what Jeff found holds true n a one day "Throw Down"

[acorn54]

one day's results is meaningless. heck i have 3 month periods where what has worked to show a good profit for the past 3 years fails to in a 3 month span.

[Vinnie]

Ranch:

Merry Christmas to you and yours.

Meant nothing by my post but to state that of course you would need a rather large sample of races to determine anything meaningful. However, when someone continually puts in their 2 cents and extols the virtues of some phantom program that is evidently a world beater in and of itself and never for a moment puts anything evidentiary forth, it just tends to make one wonder?

Have a very Merry Christmas and Happy Holiday season. :)

[headhawg]

Not to take this thread too far off the path, but I made the challenge because I was in a pissy mood. One day is not going to prove anything but I was thinking more long term anyway. My main objective was to make a point -- that there's more to handicapping software than picking pacelines. I have no way of knowing if changing the paceline choices in Handifast would make it better or worse without rewriting the code and testing it. Would it be worthwhile to do so? Probably, but that is a lot of work for what I believe --especially after reading Jeff's posts -- won't improve Handifast by a significant amount if at all.

The other point I was trying to make is the old "put your money where your mouth is" kind, but I'm pretty much over that now. (Although as PA pointed out it would be fun to see how the challenge turned out. We could call it the Head-Light or the Light-Head(ed) Challenge or something. :D )

And it's been great being associated with Handifast but this is Doug's baby first, and then Handi's. I'm just the babysitter here; I made a few bug fixes and added some nice features, but the only significant change to the numbers was that I changed the FAST fig a little. Hopefully users will continue to post suggestions and stories of success in the Handifast thread.

Best of luck and Happy Holidays to all.

[douglasw32]

It Would still be fun =)~

And you underestimate the change you made, I still have every version and I compare them often when something odd strikes,mutual wise.

It is no longer mine, it is everyone who contributed, code wise or opinion wise AND USE it themselves.

Flawed, not flawed, a different way to skin a cat and on and on and on.

If you use it and like it, great, if you helped add to it Awesome, If you coded it and took nothing in return though....you move to a new level !

[LottaKash]

We could call it the Head-Light or the Light-Head(ed) Challenge or something. :D )


Best of luck and Happy Holidays to all.



Go for it !....It would be fun, I think....And "fun" is all it would be, I would guess...:jump:

Merry Christmans & Happy Holidays everyone....:jump: :jump: :jump:

[Light]

You dont need a contest with me. I already have a little data on how my program does with its paceline selection. I can give you the results,but what would this paceline contest be based on. Top pick? Top 2 or 3,or top 4 like hcap recently did?

If you did like hcap's test,my last data for 2009 @ Aqu inner starting last January to end of March showed I handicapped 282 races. My PL method had the winner in the top 4 235 times using the top speed fig by rank as a result of my PL method. That's 83%. I think that's pretty good. Maybe JeffP or Head or Doug have some data they'd like to share that does better.

[acorn54]

in my experience one meet's worth of data is not conclusive. i have found that i need at least 18 month's worth of data for results to be reliably carried forward.

[Jake]

I was really hesitant to post on this, because I felt that Jeff had covered it extremely well. But with the dueling banjos thread drift, let me add some personal experience with paceline selection research.

A couple of years ago, my research partner and I completed a massive project where we torn apart 6+years of paceline data covering all the tracks. It pushed database limitations at the time, at least what we were doing. We looked, like Jeff has done, at every possible paceline selection method that we could program, including shifting the primary selection factor (speed, class, pace). What we found was that there was only a few percentage points win difference between the various selection methods using the same primary factor. So, using best of last 3 speed was within 2 or 3 percentage points of using best 2 of 3 speed or best of last 2 or whatever. Distance and surface did affect rank order here, but again only in a minor way. What did matter were ROI percentages relative to betting preferences.

Collectively, all our efforts indicated you should look to match paceline selection methods based on ROI (not win rates) to your preferred handicapping style. If you are primarily a longshot player, you should prefer a single paceline method like best of last 3 or best of last 4 or best ever, or a multiple method like best 2 ever (no recency restriction). You might also choose a strategy of dropping the top 2 ML horses and then applying your longshot paceline method. The key here is that the paceline method should remain favorable to your handicapping wagering preference (playing prices). If you prefer to hammer chalk, switch to a multiple method like best 2 of 3 or best 2 of 4 or best 3 of 4 which will have a slightly better win rate with roughly the same ROI.

A combination approach works best for exotics. I use this with Pick 3’s and exactas/doubles. The approach is a mixed paceline strategy. A pick 3 becomes playable when I have a single leg based on a longshot paceline method combined with multiple paceline methods for either of the two remaining legs, sometimes like this: 1X5X5. For exactas, my key is always a longshot horse with a longshot paceline selection method over multiple paceline selected horses. Same with doubles. This type of exotic approach won't maximize win percentage, but aims at ROI returns.

In our research, my partner and I also looks at composite factors from different pacelines, so we might have a composite pace figures from the last 5 pacelines, which might differ from a composite late figures from different pacelines in the last 5. All kinds of combinations. Again, Jeff and Dave Schwartz have done this kind of thing, as well as others posting on this thread with the software programs under discussions. This works just as well as the pace line approaches.

That’s why I started this with saying that Jeff had nailed the essential parts of this discussion with his earlier post. Just pick a pace line method or a composite method, research it strengths and weaknesses, make sure it match your wagering style, and then stick with it, betting only when you see value. It’s doesn’t have to be more complicated than that.

Jake

[Tom]

We used to have a guy at HTR - used the last line. Period.
He modeled it, knew what to expect, and made money.

[Light]

Another result of my PL method was that of the 282 races,exactly half of the winners,(141) were rated in the top 2 speed rank as a result of my paceline method. The return was $1166. The investment betting the top 2 blind is $1128.

My point here is not ego enchancement.After all its a tiny profit for alot of bets. It's that choosing a single paceline is as viable if not more viable than using multiple pacelines.

[Tom]

OK, then your top 2 vs Handi top 2.

[Jake]

We used to have a guy at HTR - used the last line. Period.
He modeled it, knew what to expect, and made money.

Thanks, Tom. Perfect example. My only caveat here is that last line, which is a fairly common recommended method, is highly overrated, and gains you nothing in terms of ROI, generally speaking. Last race paceline is best used as a form indicator and as an early pace indicator, not as a speed number. Last race paceline combinations are excellent contender grouping methods as well, but that's a different issue. For handicapping, God is in the details, not the specific paceline.

Jake

[douglasw32]

I played every track in north america once for about 18 months, Fast Dirt tracks, Sprints only, for winners over 2years old.

picked the winner 97% of the time on win bets alone and had a +2.07 ROI

Of course I can not "Prove it"

Never used a paceline selection algorithm to do so.

[Light]

picked the winner 97% of the time on win bets alone and had a +2.07 ROI



That means your payoffs were around $2.20. Can't see you finding many horses in 18 months that paid that low on win bets.

[Light]

OK, then your top 2 vs Handi top 2.

Last tiny test I posted of a recent weekend @Aqu had Handi 7 for 17 in top 2 with a $67.70 return. Mine was 10 for 17 in top 2 with a $76.70 return. Its over. :D

[Speed Figure]

Last tiny test I posted of a recent weekend @Aqu had Handi 7 for 17 in top 2 with a $67.70 return. Mine was 10 for 17 in top 2 with a $76.70 return. Its over. :D
Your really like the biggest joke on this board. You don't have the BALLS to do the test for everyone to see!

[PaceAdvantage]

Any test should be for at least 500 races...that's my baseline as some know already...:lol:

I haven't had anyone last that long...yet....:lol: :lol:

[raybo]

I've seen some pretty egotistical people on this site but that guy takes the cake! I can't believe he was allowed to destroy this thread like he did. Maybe it's time for a change of venue.

[headhawg]

I've seen some pretty egotistical people on this site but that guy takes the cake! I can't believe he was allowed to destroy this thread like he did.Yeah, Light is kind of like Charlie Brown's Pigpen -- he leaves a lot of dust, dirt, and who knows what other kind of crap everywhere he goes.

I'm done. Everyone here knows what he is. That's good enough for me. I wish there was something else beside the Ignore button but that will have to do. Hey PA -- try to put in the "Erase all memory of" button when you get a chance, ok? :D

[douglasw32]

That means your payoffs were around $2.20. Can't see you finding many horses in 18 months that paid that low on win bets.

No one got me being a "Smart A$$" or just an A$$ ?
My point was lets see some proof not just someones "word"

I can attach the screenshot from youbet...

Anything under a mile on the dirt allowance,claiming or stake to win or place since 2005

Bets 432, win 32%, ROI 11%, $3,484.50 Wagered, 3,876.65 Returned, Profit $392.15

All distances , all surfaces, 1036 bets, $7,186.50 wagered, loss of $224.00 , <-3%> ROI

Never used a paceline selection method until the other day.

Maybe I can improve on my record ?

[douglasw32]

I can attach the screenshot from youbet...



http://localpcpros.com/Screenshot.png

[Dave Schwartz]

played every track in north america once for about 18 months, Fast Dirt tracks, Sprints only, for winners over 2years old.

picked the winner 97% of the time on win bets alone and had a +2.07 ROI


Douglas,

I think what confused people was the above quote. What did that mean?


Dave

[headhawg]

It appeared to me that he was posting...um...fictitious data in response to a certain poster who revels in posting unsubstantiated results of some magical software program.

Doug has a funny side, don't ya know. :)

[Dave Schwartz]

ahh, I thought it was just a mistyped statement.


Dave

[Light]

What exactly am I trying to prove in a PL picking contest? I don't even know.Have I posted something too good to be true that I need to prove it? I think what I posted so far is pretty modest. So far I dont have any hard data from either Doug,Head or JeffP of how their paceline methods tested out.What is their win percentages and ROI? You would think they would have displayed that already since they disagree with my method.And they say I'm copping out. HO HO HO.

[headhawg]

Because I haven't put you on Ignore yet (this being Christmas and all) I'll respond. As I've said before, apparently you don't actually read what people post.
For the record I am definitely not against the manual or automatic selection of pacelines. (In fact my own handicapping program has a fairly sophisticated algorithm for selecting them.) My objection was to the insulting fashion in which the "suggestion" was posted. And then repeated ad nauseam.

But as I replied to Handiman (when I wasn't aware he was being facetious) I am not sure how much there is to be gained. In fact, it's quite possible that Handifast will lose its effectiveness. I certainly can see it as an option, but I think there are features to be added that are of a higher priority. That will certainly be up to Handi to decide what he's going to code.Nowhere have I ever posted that paceline selection and specifically your paceline selection is flawed. My objection to you and your posts is twofold: you post stats that are unbelievable because you yourself are unbelievable, and the denigrating manner in which you started posting and continue to post in this thread and in two Handifast threads. Some examples for those of you not scoring at home:
Just because you have the inability to program this very important feature into your program does not mean the program is somehow going to overcome it.
If a handicapping program cant do as well as Mr. B. it belongs in the trash... You guys still have a tremendous flaw in your program and you think that my pointing it out was rude.You can shape it all you want,but you will never get around it. I'll shut up now.
The flaw is that because the programmer (Doug),cannot program a user selected paceline(s) option,the user is limited to his limitation...
This has more to do with just not agreeing with my paceline selction method. It has to do with a program that is limited by its makers lack of skill.
This is why my paceline selection method is NOT flawed and yours is. I have an option,you don't. Mine is user based,yours is not. You are subject to the program's inabilities and therefore the stupidity of a program with blinkers on. That's why I would never use an program like this even for free. I know it will cost me much more at the windows than its price tag. This isn't a minor ommision that Doug cant program. This a major flaw. Nobody handicaps like that because everyone knows that's stupid.Need I go on?? There's more if you need to revisit them.

And nowhere did I brag that my selection algorithm was better than yours or anybody else's. There is no test to show you as my program is 6-9 months away from being completed. From this point forward I don't give a rat's ass what you post. You have been found out for what you are, and maybe others knew that a long time ago and I was just late to the party.

[douglasw32]

Douglas,

I think what confused people was the above quote. What did that mean?


Dave

Correct ...

It appeared to me that he was posting...um...fictitious data in response to a certain poster who revels in posting unsubstantiated results of some magical software program.

Doug has a funny side, don't ya know. images/UBGX/E1.gif


It did not come across in type, sorry =( was tongue and cheek picking on light, no harm meant.

[douglasw32]

So far I dont have any hard data from either Doug,Head or JeffP of how their paceline methods tested out.What is their win percentages and ROI? You would think they would have displayed that already since they disagree with my method.And they say I'm copping out. HO HO HO.

I recently posted my record, I never said I disagreed with your way, I am actually intrigued enough to try it, you ad HH points out are the one with the negativity and disagreement in your posts, and yet to see any of your hard data about the so called flaw in our program ?

Anything under a mile on the dirt allowance,claiming or stake to win or place since 2005

Bets 432, win 32%, ROI 11%, $3,484.50 Wagered, 3,876.65 Returned, Profit $392.15

All distances , all surfaces, 1036 bets, $7,186.50 wagered, loss of $224.00 , <-3%> ROI

To stay true to the topic, I SAY AGAIN I HAVE Never used a paceline selection method until the other day. However this was my record using an AVERAGED, "FLAWED" paceline selection method or two.

Maybe I can improve on my record ? If someone could use the posted tampa PDF to tell me where I may have gone wrong with how I understand it to be capable of "Working Better".

SCREENSHOT from youbet.
http://localpcpros.com/Screenshot.png

One thing the screenshot should show everyone is if your way works doing one thing (In my case, dirt sprints) quit playing everything under the sun and you may turn a profit, all be it in my case small.

But man I blow a ton of oney on weekends playing whatever I find.. Arabians??? when did I find a way to bet arabians? and twice at that.

Or how about the 29 TRIFECTAS with a negative 100% ROI ?

[Light]

Anything under a mile on the dirt allowance,claiming or stake to win or place since 2005

Bets 432, win 32%, ROI 11%, $3,484.50 Wagered, 3,876.65 Returned, Profit $392.15



I am not concerned with your handicapping ability.You could be the best or worst,I don't care. I was simply talking about where your PL method ranks the winner before further adjustments either human or computer. This isn't about you or Head or Jeffp or me. Try to stay objective. This is a discussion,(argument if you will) about if there is a better paceline method.

I assume you have many race cards on file that you put though your program in the past,before you even offered it here for free. I was interested in what the win percentage and ROI of your program has been in having the winner ranked in the top 2 Fast rating. I am not interested in what you actually bet. I am talking mythically if you bet the top 2 Fast ratings.

[Dave Schwartz]

Doug,

Sorry I didn't get it.


Makes sense now.


Dave

[douglasw32]

Oh mythically... Okay I misunderstood.

So if I run say 500 races through the program and look at the win % and roi of the top 2 the way it stands (Averaging the last 6).

Then recode it to look at the most recent, (same distance (Routes or sprints) and same surface) ONLY and check that win % and ROI.

That would then settle the original argument on another thread of a comprehensive rating being flawed compared to a single paceline method ?

I can write the program both ways, In a matter of minutes, I can run about 200 race cards (mainly NYRA) into it, I do not however have results charts to "Check" what the win % and ROI differences would be.

I think I would find what JEFF pointed out way earlier, that it insignificantly rearranges them in rank ?

If someone wants to send me the data files and has the charts to compare I will write the 2 versions and they can crunch the numbers, anyone?

I would be interested in checking it for my own purposes.

But again that would be seeking a black box, the program I offered for free is simply a tool to toss contenders into your vision you may otherwise overlook, and it does a damn good job of it.

It is all I use, it is how I bet, I put it out there for free to have exactly what has happened happen...growth.

If this idea is a solid way to improve the program I am all in... Not many people believe however that it possibly can be.

The youbet record was to display not how great a handicapper I am or am not, I use the "TOOL" that people now call HandiFast to make decisions, and in sprint races on dirt, for winners over 3 yo. I do better than I do any other way.

Someone else could use the same tool and be great on turf or AW... who knows, but I do not to this day see how it can be considered a "Flaw" to use a comprehensive rating that works to a MYTHICAL one that no one, even some very impressive individuals on this board (IMHO) that have more years and better resources than lil ol me, have NOT BEEN ABLE TO NAIL DOWN.

It is neither a debate or an argument, it is simply two sides of the same coin.

If our program, picked the top 4 by using numerology and the birth month of the horse, I would then agree with you it may have a flaw.

Now back to this thread ------------

Does anyone have results and matching bris single file DRF data, if so I can do a study on the single paceline versus the comprehensive approach OR better yet the two of them combined into a single ranking.

Of course we should have some tight restrictions on the races used and I vote for

LESS THAN 1 MILE, FAST DIRT, NON MAIDEN, 3 YEARS OLD AND UP, CLAIMING, STAKES and ALLOWANCE.

[douglasw32]

Okay as pointed out , I would not want someone to share the data files and tick off equibase.

I would of course, send the 2 versions, one composite paceline, one automatic paceline selection as mentioned above finding the most recent that qualified.

to the person having the needed data and results to crunch the data.

I would set the programs up to only show qualified races for the study mentioned earlier.

;)

[sjk]

Doug,

I have results for the races. If you want to send me a table of horses with the dates of the races in question I could certainly tell you what the win percent and roi would have been.

One caveat is that I have the results as they finished. Also don't have an easy means of dealing with situations where the other half of an entry won so you might avoid races with DQs and entries.

I would have thought the purpose of paceline selection would be to accurately determine midrace position so if I were testing my own methods I would find the percent of horses on the lead at the half to be more germane.

SK

[douglasw32]

Someone layout what they want the programs to do.

I will use mine the tampa post I made, not the one handi (as not to disrupt his hard work) has worked on but it is the same source code.

I still average the last 6 races period.

I can code another one to not do that and select a paceline, anyone want to agree (LOL) on how to let the computer chose it.

then I can run a bunch through it and we can look at the top 2 of each program and see if one is better than the other?

[headhawg]

Doug,

If you believe Jeff P's research (which I do) then just create a paceline selection method that suits your needs or whatever you think is best. Contrary to what some people think, it doesn't appear that it's going to matter all that much. It's kind of like setting weights -- you can keep searching for the perfect settings but if you keep changing them you're just chasing (essentially backfitting). Paceline selection is just one small step of the handicapping process.

HH

[douglasw32]

Probably, but if anyone cares, I have found it HANDI no pun intended to try it out, take the top 4 or anyone ranked number 1 in ANY of the programs separations...as contenders (I use the defaults)

Then I open up the bris custom PP generator and filter, i.e. at mnr tonight, I filtered for distance within 1 furlong, 12 races, same surface, OFF tracks.

I then compared the top ones and looked for the most recent paceline, unless it lloked like crap compared to the next one back and I could excuse it (some art involved there) that is why I still have to think it has no place in the FREE HANDIFAST program (Different thread), since doing it automatically would not have helped much.

But it does seem to have merit when you are using the program I use as 1 of many tools.

[keenang]

Doug:
What is the difference if any between the Fast program and the Handi program.

Thanks and Happy New Year
Geno :)

[douglasw32]

The Fast program is what I wrote, the handi is what the other two guys coded based on numerous requests.

The figures are all based on the same source code but completely different in so many ways it would be very hard to accurately list them.

They both use a comprehensive paceline selection method for the final figure