Anyone coded the MATCHUP?

[rsetup]

Automating the MATCHUP would probably prove quite lucrative as one could play any number of tracks and pick the most advantageous plays. Has anyone programmed the method?

[Robert Fischer]

is this 'Matchup' as in Doc Sartin / Hat Bradshaw??

I'm not familiar with the Matchup method, but I'm sure some here are.

[DeltaLover]

What is this?

[DeltaLover]

I see.. Thx for the info.

I do not think that there is a lot of value here though..

[rsetup]

I see.. Thx for the info.

I do not think that there is a lot of value here though..

But there's value in statistical analysis by post position to determine surface biases?

[DeltaLover]

But there's value in statistical analysis by post position to determine surface biases?

Most likely not.. I am not a huge believer neither in surface nor in post position biases..

[rsetup]

Most likely not.. I am not a huge believer neither in surface nor in post position biases..

Weren't you doing this on another thread? :bang:

Had me wondering.

[DeltaLover]

Weren't you doing this on another thread? :bang:

Had me wondering.

I've got involved expressing my scepticism about biases.

The game is way deeper than biases, the MATCHUP or whatever else deterministic illusion. I think that the most important tool by far, is the creation of superior speed and pace figures which is also the most challenging and complicated part of the handicapping process.

People tend to underestimate the complexity of track variant estimation shifting their interest in not important factors and approaches (like the MATCHUP for example).

[traynor]

One of the most basic strategies Sartin used went over the heads of most, including some who were considered "experts" in the applications. The various software apps served the role of "lucky charms." That is, they provided an excuse for winning for those who failed to understand that the primary purpose was to train pattern recognition skills in selecting "true contenders," "representative pace lines," and other key elements. Those who understood what they were doing often did well. Those who attempted to (or continued to) mechanically apply "concepts" usually failed--and heaped scorn and blame on Sartin's methods.

Some of the most adept at applying Sartin's techniques pretty much ignored the "recommendations" made by the software. If the software didn't predict the outcome the user thought would occur, the user (often) tweaked the input to make it conform--by selecting other pace lines, more or fewer or different "true contenders," or some other process. The bottom line was that some of the most successful Sartin users essentially had eyeballed the race, determined the entry(ies) they thought would prevail, BEFORE entering any data into the apps.

Hand entry of data went a long way toward boosting the analytical skills of the more successful Sartin users. It was such a pain that the decision process for a given race was often completed before the start of (minimalist) data entry.

Automating data entry diminished the skill level of users, rather than increasing it. The more that was left to the application to "decide," the less able the user(s) became to make the (very critical) preliminary decisions that enabled accurate prediction. The more explicitly users attempted to establish "rules" and "procedures" for applying Sartin's methods, the less well those methods performed. No great surprise there. Many tend to overlook the fact that Sartin was a psychologist, not a "handicapping guru." His methods were based on sound psychological approaches that worked (often very well) to train pattern recognition skills that enabled users to make better choices in their wagering endeavors. Those who expected the software apps to make all the decisions generally failed in those endeavors.

[traynor]

A simpler answer. Successful use of the MatchUp requires/required a high level of pattern recognition skill in pre-race, pre-computer decisions. Those who understand/understood this (usually) do/did well. Those who attempt/attempted to code those processes into apps do/did (much) less well. Coding the MatchUp works/worked about as well as explicitly coding the selection of "the most predictive pace line" for each (selected) "true contender." It doesn't/didn't.

[rsetup]

A simpler answer. Successful use of the MatchUp requires/required a high level of pattern recognition skill in pre-race, pre-computer decisions. Those who understand/understood this (usually) do/did well. Those who attempt/attempted to code those processes into apps do/did (much) less well. Coding the MatchUp works/worked about as well as explicitly coding the selection of "the most predictive pace line" for each (selected) "true contender." It doesn't/didn't.

I don't know with certainty whether the matchup can be effectively coded or not. I'll probably have a clearer sense if Richie shows up. Having observed some of its proponents work races over on Pace n Cap a number of years ago, I'm inclined to think that it can be. It'd require a lot of logic, of course. But it'd probably be the lesser of 2 evils. The next option would be to train a machine learning model on it. This should handle the pattern recognition fine

[cj]

A simpler answer. Successful use of the MatchUp requires/required a high level of pattern recognition skill in pre-race, pre-computer decisions. Those who understand/understood this (usually) do/did well. Those who attempt/attempted to code those processes into apps do/did (much) less well. Coding the MatchUp works/worked about as well as explicitly coding the selection of "the most predictive pace line" for each (selected) "true contender." It doesn't/didn't.

Having read the book, it can be coded without question to my mind. Whenever I hear something can't be I think it is either intentionally vague or just not well defined.

[Tom]

Two programs I am aware of - A Odds and The Match Up (Bradshaw).
The later was much more in depth.

[DeltaLover]

I don't know with certainty whether the matchup can be effectively coded or not. I'll probably have a clearer sense if Richie shows up. Having observed some of its proponents work races over on Pace n Cap a number of years ago, I'm inclined to think that it can be. It'd require a lot of logic, of course. But it'd probably be the lesser of 2 evils. The next option would be to train a machine learning model on it. This should handle the pattern recognition fine

Actually the logic must be the unknown of the model.

[JJMartin]

I could code all the rules in my program in Excel but it would probably take a month. If the required data is present in the pp's, it can be done.

[JJMartin]

Having read the book, it can be coded without question to my mind. Whenever I hear something can't be I think it is either intentionally vague or just not well defined.

I would agree...
The way I look at it, when applying a method manually, you would expect to make the same exact decisions consistently in regards to all data derived from past performances. So if you looked at the same race twice, you would make the same decision twice, if not then something is wrong with the decision making process. As long as you document every possible scenario that can be extracted from the pp's and your response to the conditions never changes, it can be coded. The program simply does what you would normally do if it were done manually. Of course the program would not be able to account for any race-day anomalies that would negatively impact any selections made. For example, a horse whose form is questionable based on it's physical appearance or some other variable which requires subjective interpretation. The rules in the coding would not be based on anything subjective and therefore should always result in a predictable output 100% of the time. Any "intuitive" or otherwise adjustments to the software's recommendations is not a reason to fault the program. Since horse racing is rather complex, it is likely that there are going to be certain races that contain variables (that could be coded but weren't) that haven't been taken into account because they were unforeseen.

[traynor]

Having read the book, it can be coded without question to my mind. Whenever I hear something can't be I think it is either intentionally vague or just not well defined.
Or both.