Pace Analysis Software

[JJMartin]

Originally Posted by ReplayRandall

Re-read post #142......You'll see the agreement, we're on the same frequency.

Right, so we have races where the running style is an "E" type horse and at first thought, the analysis for those races should be separate from those whose runners do not include an "E". Just some throwing out some thoughts on what would have to be considered. Another area is the rate of pace which after being determined would affect the group's individual energy expenditure. How the energy expenditure is managed by the individual would be important relative to the distance of the race. So I think it is reasonable to suggest that pace analysis is interdependent on many other areas that have to be assessed for the impact of their correlation.

If it is the consensus that my approach is not close enough to what Traynor's intent is, I could possibly change the approach or just stay out of it altogether. And I mean that in a positive tone of course. I think we all have our own perspectives and there is something to be gained from everyone's input but perhaps something of this magnitude will require some leadership and executive decisions or else too much difference of opinion could derail the effort.

[Cratos]

Originally Posted by JJMartin

I take it that the intentional exclusion of any aspect of commonly used systems is in part for the sake of acquiring an edge over the majority?

I am not being pompous, but if it is a law in science and applicable to horserace handicapping I will defer to it, but if it is conjecture used to sell a book or a software program I will avoid it like the plague.

[Cratos]

Originally Posted by JJMartin

Right, so we have races where the running style is an "E" type horse and at first thought, the analysis for those races should be separate from those whose runners do not include an "E". Just some throwing out some thoughts on what would have to be considered. Another area is the rate of pace which after being determined would affect the group's individual energy expenditure. How the energy expenditure is managed by the individual would be important relative to the distance of the race. So I think it is reasonable to suggest that pace analysis is interdependent on many other areas that have to be assessed for the impact of their correlation.

If it is the consensus that my approach is not close enough to what Traynor's intent is, I could possibly change the approach or just stay out of it altogether. And I mean that in a positive tone of course. I think we all have our own perspectives and there is something to be gained from everyone's input but perhaps something of this magnitude will require some leadership and executive decisions or else too much difference of opinion could derail the effort.

This is a very good understanding, but I would look at "work" as opposed to "energy"; they are both measured with the same metic, joules.

[JJMartin]

Originally Posted by Cratos

This is a very good understanding, but I would look at "work" as opposed to "energy"; they are both measured with the same metic, joules.

Ok, I will be away for a while, thanks.

[LottaKash]

Originally Posted by Cratos

This is a very good understanding, but I would look at "work" as opposed to "energy"; they are both measured with the same metic, joules.

I voted for a horse the other day, and he was Jouless...(Is that a word?)..

[Cratos]

It might not look like it, but the group of possible participants (based on their post inquiry to this thread) is growing with the following posters:

• Traynor (the OP; we will urge him to stay)
• BCourtney
• Hypnotist1
• TexasDolly
• Rsetup (maybe)
• Cratos
• JJMartin
• Plainolebill
• Cincyhorseplayer
• DeltaLover
• Southbaygent
• Pondman
• LottaKash
• UltimateBetter
• ebcorde
• traveler
• ultracapper

If I left any poster off who might want to participate or if I included a poster who do not want to participate, I apologize.

[traynor]

Originally Posted by JJMartin

I apologize if I may have contributed to that impression. On further contemplation to your proposal, it seems that what you are describing is a type of system that would mirror or be similar to a neural network. From what you have explained, a user would start off with some perceived scenario and then the algorithm in the software would have to look for patterns that fit the scenario? Then a system of measure that after reaching a certain level of confidence (which would have to be determined somehow) provides validation.

Not so. The initial steps are to find scenarios that can be used in training. That presupposes the scenario is a) repeated in future races, and b) usefully predictive of the outcome of those future races. "Pace analysis" on the order of, "Why, just the other day at Aqueduct, blah blah blah happened, which proves yada yada" although prevalent, pervasive, and (apparently) uncritically accepted as meaningful by many, I have never found to be profitable. YMMV.

There is no algorithm in the software to look for patterns. The intent of the software would be to train the bettor to find his or her own patterns. That presupposes whoever designs the training software has a better grasp of "pace" and its effect (or irrelevance) in specific races than the "common knowledge" regarding pace and its effect (or irrelevance).

[traynor]

Originally Posted by JJMartin

Right, so we have races where the running style is an "E" type horse and at first thought, the analysis for those races should be separate from those whose runners do not include an "E". Just some throwing out some thoughts on what would have to be considered. Another area is the rate of pace which after being determined would affect the group's individual energy expenditure. How the energy expenditure is managed by the individual would be important relative to the distance of the race. So I think it is reasonable to suggest that pace analysis is interdependent on many other areas that have to be assessed for the impact of their correlation.

If it is the consensus that my approach is not close enough to what Traynor's intent is, I could possibly change the approach or just stay out of it altogether. And I mean that in a positive tone of course. I think we all have our own perspectives and there is something to be gained from everyone's input but perhaps something of this magnitude will require some leadership and executive decisions or else too much difference of opinion could derail the effort.

That is simple. My intent is to migrate an almost hopelessly complex and interrelated cluster of software apps that are currently highly "visual" (written, extended, and expanded in Visual Basic over a period of years) into a succinct, elegant, set of scalable Scala apps that accomplish the same thing (or better). Designed from the outset, rather than cobbled together and used mostly because they work. Very well. While so doing, I hope to learn enough about Scala and Spark to embed the latter in the new apps, and essentially automate the machine learning and neural network aspects of Spark (normally referred to as deep learning) as intrinsic components of the apps.

I think the direction this thread has taken may prove useful, and I look forward to seeing the results of the endeavor.

[traynor]

I did not intend that to seem obscure. More directly stated, at the time I started this thread (way before Christmas, almost two months ago) I had some free time. With the racing season rapidly approaching, that free time no longer exists. I have a lot to learn, and a lot to do. And little or none of it is of a nature that I can google it and cut and paste someone else's code as a quickie solution. Nothing personal, but I lack the time to participate in whatever it is that is going to be done. Or not done.

[Cratos]

Regrettably and with sincere apology I cannot continue to be an active participant in the collaborative pace analysis SW effort because I was informed that I am needed to perform work elsewhere.

Therefore, given the input by many bright minds thus far in this thread, this should be a successful project and a rewarding experience; I will try and stay inform by reading various posts within this thread.

[JJMartin]

Originally Posted by Cratos

Regrettably and with sincere apology I cannot continue to be an active participant in the collaborative pace analysis SW effort because I was informed that I am needed to perform work elsewhere.

Therefore, given the input by many bright minds thus far in this thread, this should be a successful project and a rewarding experience; I will try and stay inform by reading various posts within this thread.

You don't have to write it off 100%, you could still have some input here and there.

[JJMartin]

Continuing along the work I've been doing, we see that the "E" horse has a very high percentage of accuracy of being the leader at 1st call (at least in the general sense to begin with). Would anyone be interested in how the "P" horse does when there is no "E" in the race?

[traynor]

Originally Posted by JJMartin

Continuing along the work I've been doing, we see that the "E" horse has a very high percentage of accuracy of being the leader at 1st call (at least in the general sense to begin with). Would anyone be interested in how the "P" horse does when there is no "E" in the race?

Serious questions. What do you think this means? Especially given that the averages/values are generated by mixing (many) different types of races together.

Of what value do you believe this type of research (and the resulting figures/values/percentages/whatever) to be in predicting the outcome of future races?

What correlations are there in the output you post to horses that lead at the first call winning/losing when and when not those horses are the top rank in whatever ranking system you are using?

Do you have clear insight into the parameters used to categorize the horses (by whatever rating system you are using) so that you know a horse designated "E" (or whatever) is actually what most would consider an "E" (or whatever) type horse, and not some spurious, computer-generated nonsense "calculated" by silly "algorithms" designed for convenience in labeling and the appearance of being meaningful, rather than actually defining something useful (and intrinsic) about a horse. And are you confident enough in the accuracy of the designations to believe they truly represent the "preferred running style(s)" (if such a thing actually exists other than in very broad, very general terms) of the horse(s) in question?

[barn32]

Originally Posted by DeltaLover

http://cbsg.sourceforge.net/cgi-bin/live

I'm sure this is something that was written by Professor Irwin Corey.

[JJMartin]

Serious questions. What do you think this means?

I will start by saying I am offering an opinion, an interpretation based on my personal conceptual understanding (or lack of) of the various things involved with analysis. And during the process itself, the understanding of the various factors involved and their interrelations may change with new developments.

Especially given that the averages/values are generated by mixing (many) different types of races together.

Since we are only dealing with 1st call analysis at the moment, I have the following reasoning. The horses per se do not have any concept of what a race is, that they are in one, what the distance is, that there is a finish line etc. They may be aware of the differences in surfaces and other things of course. When the gate opens and the race starts, the herd mentality kicks in and for the most part they will try to run together as close as possible as a group. If there is a strong leader type, that horse will try to get to the lead as early as possible. Judging by the 52% I got by testing the "E", I would say that is a fair assessment. The 1st call is reached roughly 23 seconds from the moment they start running. Most horses are not going to be completely out of energy by that short distance. Based on this we can speculate with reason that a slower top speed horse will exert more effort while he still has energy to keep up with the rest. Therefore distance, race type, jockeys, trainers and other such factors are mostly irrelevant in this sense and will have minimal impact if any on what should be expected at 1st call. Of course as I have stated before, this is subject to change if new evidence suggests otherwise. What I have found is that the average percentage tends to gain as the macro ascends through the year. For example at 1/3 through it may be 51% and at midpoint 52%. The likely reason is that the number of races that the averages are comprised of have increased and contributed to a higher degree of accuracy.

Of what value do you believe this type of research (and the resulting figures/values/percentages/whatever) to be in predicting the outcome of future races?

The idea is to tackle one stage of the race at a time and once that stage is sufficiently predictable, move on to the next segment. You can't predict the end without knowing what happens before. The value of predicting future races is self-evident if they produce positive results when applied.

What correlations are there in the output you post to horses that lead at the first call winning/losing when and when not those horses are the top rank in whatever ranking system you are using?

Actually I am not concerned with that particular thing at the moment. I believe there is too much going on in between to attempt to find a usable correlation from whatever is determined at 1st call to finishing first. As we all know, the leader at 1st call does not always equal winning the race or any expectation of acquiring a positive roi. That being said, when I ran the #1 "E" horse with the condition of being rank 1 for turn time, it would lead at first call with 61% and went on to win for 39% but with an roi five times worse than blindly betting the favorite on every race. So in this example we see how knowing what rate a horse achieves 1st place at 1st call contributes to a higher win percentage and if that was solely the goal, we would have success. But this type of horse is so transparent in it's recorded history that it will be invariably over bet with a terrible roi. However this information may be used to pass a race if we believe we have a contender that shows value but will likely lose to this super "E" horse. So we continue to look deeper into the more elusive correlations within pace analysis until we find a horse that should win at a better price.

Do you have clear insight into the parameters used to categorize the horses (by whatever rating system you are using) so that you know a horse designated "E" (or whatever) is actually what most would consider an "E" (or whatever) type horse, and not some spurious, computer-generated nonsense "calculated" by silly "algorithms" designed for convenience in labeling and the appearance of being meaningful, rather than actually defining something useful (and intrinsic) about a horse. And are you confident enough in the accuracy of the designations to believe they truly represent the "preferred running style(s)" (if such a thing actually exists other than in very broad, very general terms) of the horse(s) in question?

That is the more difficult questions to answer. The whole idea of running styles is a bit vague in my opinion. There are really only 2 running styles. There is the leader, and there is everyone else. When we see an E horse in a race, he will generally be the leader more often than any other designation, I think we can agree on that. But when there is no E horse and just P's and S's. Somebody has to lead. They don't all just decide not to run because an E horse didn't show up for the race. Therefore whoever that leader is, say one of the P horses, what is he? Is he a P or is he an E? When I ran races with the #1 rank that did not contain any E's, I got 42%. That is 10% less than the E had. This indicates that the P's generally prefer to follow than to lead. And that sometimes the same P who led before in a previous race will submit with a higher degree (to not leading) when another non "E" horse leads and when there is no E present. The P horse therefore is more difficult to assess than the E. This is the type of area that if better understood could contribute to a higher roi if a successful method of determination can be developed. I have seen the same horse change its designation 3 times before settling. It is not always an easy thing to determine, therefore the more prior races we have in an individual's history, the better we can make a positive assessment. In other words, more information is better than not that is obvious. Since I am still at the beginning stages of this research I can't really provide any solid answers and my premises could be off, I just don't know yet.