Pattern Recognition

[bob60566]

Originally Posted by Gamblor

So I guess by that you mean the Beyer "workout within race" where they make a move from back, then drop back again late, that kind of thing?

The 2nd one - even race same position whole race... do you mean even pace of race and they stay in the same spot? What's positive about that pattern? IE if a horse in an even race sits 4th throughout, is that good? Could you explain that one to me?

Yes
Yes it indicates sharpness with trainer intention bottom line every running line tells story.
The workout in the race is the first two calls in a sprint then drops back.

[Gamblor]

Originally Posted by bob60566

Yes
Yes it indicates sharpness with trainer intention bottom line every running line tells story.
The workout in the race is the first two calls in a sprint then drops back.

Oh yes, the Beyer one was the midrace move, change of pace or something he called it?

[teddy]

Where would you start...deciding if a race was playable first then try to set out to use a process to select the best choice based on last min odds?

[bob60566]

Are you talking pattern recognition

[traynor]

Originally Posted by DJofSD

Looks like the link is now out of date. I see an article about the double-double star discovered to have a planet -- I doubt that was the intended article.

In any event, a part of what Gladwell talks about in "Blink" is the subconscious being able to recognize patterns well before you are consciously aware. This is what I refer to as intuition in my own handicapping experiences.

If you read the article, you will see that the planet was discovered using pattern recognition. It has a lot of applications--thoroughbred racing is only one of them.

[traynor]

Originally Posted by Sinner369

traynor if you are stuck for ideas, steps, informal rules, patterns, etc.......why don't you outline them here and perhaps we can help you generate new thoughts, ideas........... or even email us privately!

I appreciate the offer. I will certainly take advantage of it when I get to that point. Thanks!

[traynor]

Originally Posted by Capper Al

I agree, but not to knock computer applications in general. Once a human figures it out then a computer can do it better. The problem as I see it is how the computer is used when developing. The statistician's approach is putting the cart before the horse. Finding out that something occurs 1 in 4 times doesn't directly fit into the analysis of the game. The analysis is still handicapping like the old paper and pencil guy used to do it, but just more rapidly and accurately with computers. As Brad Free kept hammering away in his book Handicapping 101, it's the fundamentals stupid. After the application is built on handicapping logic then the statistics and ROI matter.

Yes. Computers do well at the grunt work, and making sure nothing is overlooked. However--much like people--they need explicit instructions about what to do in order to get it done right. That is where the real work comes in.

Unfortunately, most race analysis apps do little more than crunch numbers and display them in a different format. That is, a limited number of data points are massaged, and the results displayed as something different. No matter how the same basic set of numbers is massaged, the results are generally the same between apps. The format might make them seem different, but the base data points are the same.

The pattern recognition should be going in, not trying to divine which of a dozen or more different "ratings" should be used for the current race. That type of decision is what computers are supposed to do, based on the pattern recognition templates defined in the algorithms.

[traynor]

Originally Posted by Aner

Sometimes humans are too good at pattern recognition. Do all fleecy clouds look like an animal? Are those really images of Jesus or Virgin Mary in the pizza, tree, rock, window, grilled cheese sandwich etc.? Do daily track variants really gyrate greatly from day to day sans weather changes?

I use pattern recognition all the time, but it's a little like the strike rate of idea people. If 10% of their ideas pan out, they are ecstatic.

Pattern recognition has little to do with subjective interpretations of cloud formations. It is one of those areas that has a very specific meaning that may not be the same as the colloquial one. That is, what most people call pattern recognition, and what engineers, scientists, and researchers call pattern recognition, can differ substantially.

[traynor]

Originally Posted by Gamblor

If patterns exist, they can be defined. So, to the pattern analysts here, what kinds of patterns do you see occurring when you "do the form"?

That is what I am trying to put together. Ideas are appreciated.

[Capper Al]

Originally Posted by traynor

That is what I am trying to put together. Ideas are appreciated.

Like I mentioned earlier, patterns are what the old paper and pencil handicappers excelled in. For a starting point, go over the basics as mentioned in Brad Frees book and handicap from there on. We start with the fundamentals and then move on to verify our results with stats. Not stats as a method of finding a way to handicap.

[DJofSD]

Originally Posted by traynor

Pattern recognition has little to do with subjective interpretations of cloud formations. It is one of those areas that has a very specific meaning that may not be the same as the colloquial one. That is, what most people call pattern recognition, and what engineers, scientists, and researchers call pattern recognition, can differ substantially.

Yes, they can differ substantially. That is why these discussions never really amount to anything -- we can even agree on the definitions.

How pattern recognition is performed is open to debate too. Not everything has to be done using a computer. Sure, being able to codify it and to put numbers to it for analysis is a huge achievement but being able to demonstrate a repeatable method not using a computer does not make the results any less legitimate.

[eurocapper]

How do you do it science. You don't blindly test for something empirically (even though some maybe want to believe that). First you have a theoretical framework, then a hypothesis which you test for accuracy. Now this planet for instance, first there was a theory that other solar systems exist, then that other planets exist etc. So IMO one needs to pick trainer mentality and horse psychology (or something similar) first, then test statistically (as a very secondary process) whatever one may come up with. Beyer details in his books the conversations with various insiders that helped his game for instance. Without that insider info, I doubt he would have been able to distinguish horse trouble from race workout.

[traynor]

Originally Posted by DJofSD

Yes, they can differ substantially. That is why these discussions never really amount to anything -- we can even agree on the definitions.

How pattern recognition is performed is open to debate too. Not everything has to be done using a computer. Sure, being able to codify it and to put numbers to it for analysis is a huge achievement but being able to demonstrate a repeatable method not using a computer does not make the results any less legitimate.

You say: "That is why these discussions never really amount to anything." I disagree. Because subjective definitions (an internal, personal set) differ has no bearing on anything external to that set. That is, people believe what they believe about a given topic, but that belief has no effect whatsoever on the content of the topic--it only has effect on their internal, personal set of subjective beliefs and opinions.

Like the man said, "A difference of opinion is what makes a horse race."

[traynor]

Originally Posted by Capper Al

Like I mentioned earlier, patterns are what the old paper and pencil handicappers excelled in. For a starting point, go over the basics as mentioned in Brad Frees book and handicap from there on. We start with the fundamentals and then move on to verify our results with stats. Not stats as a method of finding a way to handicap.

What if Brad Frees was wrong, or was only expounding his personal beliefs, rather than external reality? How did Brad Frees arrive at his beliefs? I am quite sure he didn't just wake up one morning and say to himself, "Eureka! I have The Answer!"

The starting point should be an extensive statistical analysis of "relevant factors" to determine if they are--in fact--relevant. Up to the time Quirin wrote his first book, weight was considered "important" and included in just about every "expert analysis" of probable race outcomes.

You seem to be overlooking the fact that "the basics"--if they are valid--have been arrived at empirically based on results in the real world. That is statistics. Not theory. If those basics are flawed, the weakness of pencil and paper handicapping is that it lacks sufficient data to do more than (relatively blindly) follow the theories of others. To gain the data necessary to do otherwise takes statistics, and statistical analysis.

[traynor]

Originally Posted by eurocapper

How do you do it science. You don't blindly test for something empirically (even though some maybe want to believe that). First you have a theoretical framework, then a hypothesis which you test for accuracy. Now this planet for instance, first there was a theory that other solar systems exist, then that other planets exist etc. So IMO one needs to pick trainer mentality and horse psychology (or something similar) first, then test statistically (as a very secondary process) whatever one may come up with. Beyer details in his books the conversations with various insiders that helped his game for instance. Without that insider info, I doubt he would have been able to distinguish horse trouble from race workout.

In modeling, the process is similar to reverse engineering. That is, rather than a "fishing expedition" hoping to find something useful, the "end result" is known. The process then is to determine how to replicate that desired outcome--to determine the factors and indicators that could be replicated or used to get from Point A to Point B, with Point B being the desired outcome.

I agree completely with your observation above. If the researcher does not have a clear objective, it is a "fishing expedition." Nothing more.