Question for software creators/customizers

[markgoldie]

Any black box creators use a scan of trouble lines so as to compensate for troubled trips, such as dwelling, rearing, stumbling, brushing, checking, colliding, etc? Would seem that incorporation of such trouble lines would skew the power-ratings of the contestants.

[sjk]

I delete lines where the horse never gets out of the gate or where the jock come off before the first call. If I were to use lines like that it would make pace projections more problematic and I don't expect the conditions which caused the problem to repeat in the next start.

[dukeofperl]

I also toss the troubled-lines with the mindset that each horse, given time, will have had an equal amount of good and bad breaks and trips.

[markgoldie]

Thanks for responses. What I guess I was really wondering is if there is an automated way to have a program recognition of trouble-line words in the past performances. On the other hand, not using these types of data files, I don't even know if the trouble line comments are incorporated in them.

But this would seem very important to me. For example, many times I will be looking at a series of pace and final-speed numbers for a field of horses and I will see an abberationally low number for a given horse. If I then go to the standard form, I will see something like "stumbled badly at start" in the trouble line. So I know to throw the line out. On the other hand, I'm wondering how a black-box program that handicaps many tracks at one gulp would handle this problem. Certainly if you average the sub-par performance into the power mix of the horse, you will get a number that is artificially low.

[xfile]

I've got the program looking for every conceivable trouble quote ever mentioned. Since each circuit, for the most part, has a different caller they will word the same thing differently.

[sjk]

I pitch the lines from the database when they occur. There are only a handful of lines each day from around the country where the horse is distanced at the firts call and it is a simple matter to look at the comments to see what was going on.

If you don't have the comments you would be better off to delete all of the lines where the horse was distanced at the first call than to keep them all.

I would be wary of a simple average of all of the legitimate pace figures because ill-advised tactics can distort the true picture.

As an aside I am always amazed at how few people out there have black box software and are willing to answer such queries.

[raybo]

In my personal handicapping spreadsheet, in Excel, prior to AllData, I, also, have the program looking for many trouble comments, of various spellings/abbreviations. When any of these variations are found, that paceline is "highlighted" via "conditional formatting", as well as being, possibly, thrown out from paceline selection consideration.

I always verify what the program is spitting out, prior to finalizing a wager, which includes looking for those "highlighted", troubled pacelines, and manually override those I don't agree with.

[markgoldie]

Thanks for replys. What brought this to mind is two things: (1) Someone doing a completely automated betting program that not only represents hands-off handicapping, but hands-off wagering as well. Here, you have zero input into the process, so if something like a trouble-line scanner is not employed, it would appear that severe errors could be made. (2) I was wondering about a black box program such as Brisnet's Prime Power number. They are cranking out so many ratings over the course of any given day that it would seem implausible to me that a live human is reviewing the trouble lines and adjusting the power numbers accordingly. I could be wrong about that, but I doubt it.

I realize that nothing can substitute for looking at replays when it comes to determining trouble. However, in many cases (like mine), I just don't have the time to afford this luxury. However, through many years of handicapping, I have noticed that a large majority of the time, almost any description of trouble will (meaning it was bad enough to be noted) will affect the animal's final number.

Sjk mentions being wary of averaging of lines and I agree about this. In my own methodology, I try to find a line that is most indicative of how I believe the horse will perform on this day. That's a far cry from averaging.

[dukeofperl]

What I do is set thresholds - I.E. ... a horse is allowed to be 8 lengths off the pace at the half and 6 lengths off the pace at the finish. Not necessarily those precise numbers, but I will use them here as an example.

Then, if the race line violates those thresholds, the line is tossed with little to no thought about why it's being tossed. It could be the horse had trouble or it could be the horse is simply in with a tougher crowd or the horse is having a bad day (not just humans have bad days) or the jockey thinks (s)he had no chance to win so the horse was eased, or even as simple as ... sometimes sh*t happens.

Regardless of the reason, it is tossed so it doesn't taint the remaining race lines that stay within the threshold and will be averaged (so to speak) for comparing horse against horse.

IMO, my philosophy is as simple as it gets and it's what keeps me from going over to the dark side, near insanity, where I would be calculating inches and wind speed and oval-track shapes and the science of dirts.

[raybo]

What I do is set thresholds - I.E. ... a horse is allowed to be 8 lengths off the pace at the half and 6 lengths off the pace at the finish. Not necessarily those precise numbers, but I will use them here as an example.

Then, if the race line violates those thresholds, the line is tossed with little to no thought about why it's being tossed. It could be the horse had trouble or it could be the horse is simply in with a tougher crowd or the horse is having a bad day (not just humans have bad days) or the jockey thinks (s)he had no chance to win so the horse was eased, or even as simple as ... sometimes sh*t happens.

Regardless of the reason, it is tossed so it doesn't taint the remaining race lines that stay within the threshold and will be averaged (so to speak) for comparing horse against horse.

IMO, my philosophy is as simple as it gets and it's what keeps me from going over to the dark side, near insanity, where I would be calculating inches and wind speed and oval-track shapes and the science of dirts.

Your approach sounds logical and probably would work, for general comparisons of horse to horse potentials.

That coupled with a thorough analysis of current condition of, all the horses, would make the process very powerful, indeed.

[raybo]

Thanks for replys. What brought this to mind is two things: (1) Someone doing a completely automated betting program that not only represents hands-off handicapping, but hands-off wagering as well. Here, you have zero input into the process, so if something like a trouble-line scanner is not employed, it would appear that severe errors could be made. (2) I was wondering about a black box program such as Brisnet's Prime Power number. They are cranking out so many ratings over the course of any given day that it would seem implausible to me that a live human is reviewing the trouble lines and adjusting the power numbers accordingly. I could be wrong about that, but I doubt it.

I realize that nothing can substitute for looking at replays when it comes to determining trouble. However, in many cases (like mine), I just don't have the time to afford this luxury. However, through many years of handicapping, I have noticed that a large majority of the time, almost any description of trouble will (meaning it was bad enough to be noted) will affect the animal's final number.

Sjk mentions being wary of averaging of lines and I agree about this. In my own methodology, I try to find a line that is most indicative of how I believe the horse will perform on this day. That's a far cry from averaging.

I think, in a black box configuration, trouble would probably, over time, even itself out. Sometimes your black box is going to profit by trouble and sometimes it's going to lose due to it.

Having said that, if I were going to employ a black box approach, with the goal of making a profit, long term, I would certainly have my software looking for key words/phrases in the trouble lines that would evaluate the viability, of each paceline, for use as a representative indicator of today's performance.

[dukeofperl]

Your approach sounds logical and probably would work, for general comparisons of horse to horse potentials.

That coupled with a thorough analysis of current condition of, all the horses, would make the process very powerful, indeed.

Correct -- and another point ...

The more race lines I have for a horse, the better. Therefore I build a library containing each horse by name and let it build over time with each horses's race line (for each track).

That gives me a pretty good feel for what a horse was able to do then, and should be able to do now, which leaves everything up to lady luck as far as bad trips ... if my numbers say horse-A is the best and she gets a good trip, she should win the race. If she gets a bad trip, that race line will never show up in my program and horse-A will get the nod again when she is favorably placed. She won't have a bad trip everytime.