Frank N
19th October 2001, 05:02.38 PM
One of the most overbet pieces of information in the past performances is the finish position and beaten lengths. Those and the speed rating of choice are probably the most overused pieces of information. They are the first thing that most handicappers look at, and they're the basis of elimination for many people.
It's also well known that horses who can get to a clear lead early in the race generally have an excellent chance of winning at sprints. But most handicappers look at the second call, and indeed, most pace programs are constructed around the second call. And as Ken and others have pointed out here, an underused factor is a properly constructed first fraction (Ev or F1).
To begin with, those of you who have Access should try this query. I'll do them right along with you on my own database, which contains the same basic factors. Try this:
nDIST = 6
tSURF = "d"
nBLF >=0
rF1 = 1
For those of you who don't use Access, we're looking at all horses in 6 furlong dirt sprints with the top ranked Early Velocity (Fraction 1, or F1), regardless of their finish position or beaten lengths.
I know that using nBLF >= 0 will not filter any horses. I'm putting it there so that it's consistent for later, when we'll use it.
Depending on the tracks you include in your database, you'll come up with something in the ballpark of what I do:
Win%: 21.3%
ROI: 0.95
This is about right in terms of win% and ROI for Ev/F1, and this factor is the best single factor in terms of ROI for use at most sprint distances at most tracks.
Now, let's look at all horses who won their last race. We modify nBLF so that it's nBLF = 0 :
nDIST = 6
tSURF = "d"
nBLF = 0
rF1 = 1
The new report should look like this:
Win%: 24.5 %
ROI: 0.83
Now modify nBLF to include horses who did not win but who finished within 2 lengths of the winner as below:
nDIST = 6
tSURF = d
nBLF > 0 and <= 2
rF1 = 1
You should come up with something similar to this:
Win%: 25.5%
ROI: 0.81
It's not a fluke that the win percentage is about the same or a little better, since horses who won their last race often either have to take a rise in class or carry more weight. Horses who did not win generally get the same (or less) weight assignment and are eligible for the same (or less) race conditions (such as allowance races or claimers for non winniers of etc.). And while I'm not a big fan of weight, horsemen are, and it affects the way they enter horses.
The problem, of course, is that the public sees these wins or close-up finishes and overbets the horses that are involved. And for reason to be explained later, two is the magic number for beaten lengths.
Now let's look at top Ev horses who finished over 2 and up to 5 lengths from the winner in their last
nDIST = 6
tSURF = d
nBLF > 2 and <= 5
rF1 = 1
This is what I came up with:
Win%: 25.5%
ROI: 1.07
The same win percentage, but the ROI jumps like a rocket.
Now let's move a little further out to beaten lengths greater than 5 and up to, but not including 10.
nDIST = 6
tSURF = d
nBLF > 5 and < 10
rF1 = 1
This is the result:
Win%: 18.8%
ROI: 0.85
I'll save a comment on this result until after the next query.
And finally, let's use beaten lengths of 10 or greater:
nDIST = 6
tSURF = d
nBLF >=10
rF1 = 1
This is what my database returns:
nDIST = 6
tSURF = d
nBLF >=10
rF1 = 1
Win%: 16.6%
ROI: 1.10
What happened in these examples? Basically, the ROI of the horses in each example is fitting into the public's preconception of what beaten lengths should be, relative to odds. The "close finish" of 2 or less lengths that authors have pounded into the heads of racegoers for years has shaped the betting public's betting habits to the point where most such horses go off as underlays. From there up to 10 lengths, things seem to be treated pretty much the same in the minds of the public. But they're far from the same in reality.
So winners and horses who finish up close (2 or less lengths) ran decent races, but it's obvious to the public, and they end up getting overbet.
Horses who run a bit further out (in this example, up to 5 lengths) also ran decent races in their last, but they're often underbet by the public because of the two-length maxim, and they often end up at decent prices. Many of these horses may have finished well, and been beaten by some horse who was at the top of its form and won easilly.
As the horses get further away from the winner, the quality of race that they ran in their last, tends to deteriorate a little. But from 2 up to 10 lengths, it doesn't seem to make a lot of difference to the public. In reality, the performance sems to decline at around five lengths beaten at the wire. But the public continues to treat the horses as they did the ones from 2 to 5 lengths until they reach 10 beaten lengths at the finish.
10, being a double-digit number, is the area in which the public starts to disregard the horse's chances. But in reality, I've found that once the beaten lengths exceeds about five, the performance of an early velocity horse doesn't deteriorate that much. There is some decline, but it isn't as marked as one would think. With other running styles, this is not as true, but once a horse who runs early gets to be beaten by about five lengths, it's generally on its way to backing up and finishing out of the money, and if it's clear that the horse isn't going to get a paycheck, often the jockey (much of the time on the order of the trainer) will just ease up on it to not use it any harder and avoid the risk of blowing it out or injuring it. So a horse that could finish sixth, 8 lengths from the winner at the end if kept to task, will finish tenth, 15 lengths from the winner....because it wasn't urged anymore or possibly was even put back under a hold. While it does appear to darken the horse's form a little (and no doubt that is on the mind of many trainers for betting purposes in later races), it isn't really cheating, because the horse didn't have a chance anyhow and is simply being eased up to avoid hurting it. Also, many horses who are not quite ready for a full effort may be freshened off of a layoff or some other change in routine, or given a workout within the race, then eased back to wait for another day.
Therefore, once the beaten lengths of an Ev horse get to 10 and higher, it's given a new classification in the minds of the public and the value returns.
To show you that this wasn't a fluke, let's go a couple of steps further and use horses who were beaten at least 20 and up to 30 lengths.
nDIST = 6
tSURF = "d"
nBLF >=20 and <30
rF1=1
My results are:
Win% 19.4%
ROI: 1.08
While horses who were beaten from at least 10 up to 20 lengths returned:
nDIST = 6
tSURF = "d"
nBLF >=10 and <20
rF1=1
And we come up with:
Win%: 16.2%
ROI: 1.18
This could well be a fluke in the data, and somebody with a large database may wish to test these numbers. But it might make sense in the light that the horses who finished from 10 to 20 lengths may include more of those who had been used a little more or may be habitual quitters , whereas the horse who finished from 20 to 30 lengths may have included more that have been freshened or simply were in over their heads and weren't used after the first call.
I did test the 10-20 and 20-30 with nODDS (final betting odds) of <15 (less than 15-1 final odds) , and surprisingly, the 10-20 group ROI dropped to 0.98....but the ROI of the 20-30 group ROSE to 1.47.
Of course, there's a line to be drawn somewhere, and it seems to be conveniently located at 30 legnths behind the winner. Horses who finish that far back include a lot of horses something is wrong with, and the ROI drops off the board. My figures for:
nDIST = 6
tSURF = "d"
nBLF >=30
rF1=1
returns this:
Win%: 13.1%
ROI: 0.44
However, I stress that the sample size of this last group is relatively small, compared to the rest of the samples, and it only takes one bomb to turn that into a positive expectation.
These were the results of my database, which uses different data than HTR users have, and the results with HTR may be different. Perhaps somebody can test this and let us know.
The point, however, is that if you are more likely to succeeed if you look for value within parameters that are sound in logic, but fall outside of the generally accepted beliefs of the public.
It's also well known that horses who can get to a clear lead early in the race generally have an excellent chance of winning at sprints. But most handicappers look at the second call, and indeed, most pace programs are constructed around the second call. And as Ken and others have pointed out here, an underused factor is a properly constructed first fraction (Ev or F1).
To begin with, those of you who have Access should try this query. I'll do them right along with you on my own database, which contains the same basic factors. Try this:
nDIST = 6
tSURF = "d"
nBLF >=0
rF1 = 1
For those of you who don't use Access, we're looking at all horses in 6 furlong dirt sprints with the top ranked Early Velocity (Fraction 1, or F1), regardless of their finish position or beaten lengths.
I know that using nBLF >= 0 will not filter any horses. I'm putting it there so that it's consistent for later, when we'll use it.
Depending on the tracks you include in your database, you'll come up with something in the ballpark of what I do:
Win%: 21.3%
ROI: 0.95
This is about right in terms of win% and ROI for Ev/F1, and this factor is the best single factor in terms of ROI for use at most sprint distances at most tracks.
Now, let's look at all horses who won their last race. We modify nBLF so that it's nBLF = 0 :
nDIST = 6
tSURF = "d"
nBLF = 0
rF1 = 1
The new report should look like this:
Win%: 24.5 %
ROI: 0.83
Now modify nBLF to include horses who did not win but who finished within 2 lengths of the winner as below:
nDIST = 6
tSURF = d
nBLF > 0 and <= 2
rF1 = 1
You should come up with something similar to this:
Win%: 25.5%
ROI: 0.81
It's not a fluke that the win percentage is about the same or a little better, since horses who won their last race often either have to take a rise in class or carry more weight. Horses who did not win generally get the same (or less) weight assignment and are eligible for the same (or less) race conditions (such as allowance races or claimers for non winniers of etc.). And while I'm not a big fan of weight, horsemen are, and it affects the way they enter horses.
The problem, of course, is that the public sees these wins or close-up finishes and overbets the horses that are involved. And for reason to be explained later, two is the magic number for beaten lengths.
Now let's look at top Ev horses who finished over 2 and up to 5 lengths from the winner in their last
nDIST = 6
tSURF = d
nBLF > 2 and <= 5
rF1 = 1
This is what I came up with:
Win%: 25.5%
ROI: 1.07
The same win percentage, but the ROI jumps like a rocket.
Now let's move a little further out to beaten lengths greater than 5 and up to, but not including 10.
nDIST = 6
tSURF = d
nBLF > 5 and < 10
rF1 = 1
This is the result:
Win%: 18.8%
ROI: 0.85
I'll save a comment on this result until after the next query.
And finally, let's use beaten lengths of 10 or greater:
nDIST = 6
tSURF = d
nBLF >=10
rF1 = 1
This is what my database returns:
nDIST = 6
tSURF = d
nBLF >=10
rF1 = 1
Win%: 16.6%
ROI: 1.10
What happened in these examples? Basically, the ROI of the horses in each example is fitting into the public's preconception of what beaten lengths should be, relative to odds. The "close finish" of 2 or less lengths that authors have pounded into the heads of racegoers for years has shaped the betting public's betting habits to the point where most such horses go off as underlays. From there up to 10 lengths, things seem to be treated pretty much the same in the minds of the public. But they're far from the same in reality.
So winners and horses who finish up close (2 or less lengths) ran decent races, but it's obvious to the public, and they end up getting overbet.
Horses who run a bit further out (in this example, up to 5 lengths) also ran decent races in their last, but they're often underbet by the public because of the two-length maxim, and they often end up at decent prices. Many of these horses may have finished well, and been beaten by some horse who was at the top of its form and won easilly.
As the horses get further away from the winner, the quality of race that they ran in their last, tends to deteriorate a little. But from 2 up to 10 lengths, it doesn't seem to make a lot of difference to the public. In reality, the performance sems to decline at around five lengths beaten at the wire. But the public continues to treat the horses as they did the ones from 2 to 5 lengths until they reach 10 beaten lengths at the finish.
10, being a double-digit number, is the area in which the public starts to disregard the horse's chances. But in reality, I've found that once the beaten lengths exceeds about five, the performance of an early velocity horse doesn't deteriorate that much. There is some decline, but it isn't as marked as one would think. With other running styles, this is not as true, but once a horse who runs early gets to be beaten by about five lengths, it's generally on its way to backing up and finishing out of the money, and if it's clear that the horse isn't going to get a paycheck, often the jockey (much of the time on the order of the trainer) will just ease up on it to not use it any harder and avoid the risk of blowing it out or injuring it. So a horse that could finish sixth, 8 lengths from the winner at the end if kept to task, will finish tenth, 15 lengths from the winner....because it wasn't urged anymore or possibly was even put back under a hold. While it does appear to darken the horse's form a little (and no doubt that is on the mind of many trainers for betting purposes in later races), it isn't really cheating, because the horse didn't have a chance anyhow and is simply being eased up to avoid hurting it. Also, many horses who are not quite ready for a full effort may be freshened off of a layoff or some other change in routine, or given a workout within the race, then eased back to wait for another day.
Therefore, once the beaten lengths of an Ev horse get to 10 and higher, it's given a new classification in the minds of the public and the value returns.
To show you that this wasn't a fluke, let's go a couple of steps further and use horses who were beaten at least 20 and up to 30 lengths.
nDIST = 6
tSURF = "d"
nBLF >=20 and <30
rF1=1
My results are:
Win% 19.4%
ROI: 1.08
While horses who were beaten from at least 10 up to 20 lengths returned:
nDIST = 6
tSURF = "d"
nBLF >=10 and <20
rF1=1
And we come up with:
Win%: 16.2%
ROI: 1.18
This could well be a fluke in the data, and somebody with a large database may wish to test these numbers. But it might make sense in the light that the horses who finished from 10 to 20 lengths may include more of those who had been used a little more or may be habitual quitters , whereas the horse who finished from 20 to 30 lengths may have included more that have been freshened or simply were in over their heads and weren't used after the first call.
I did test the 10-20 and 20-30 with nODDS (final betting odds) of <15 (less than 15-1 final odds) , and surprisingly, the 10-20 group ROI dropped to 0.98....but the ROI of the 20-30 group ROSE to 1.47.
Of course, there's a line to be drawn somewhere, and it seems to be conveniently located at 30 legnths behind the winner. Horses who finish that far back include a lot of horses something is wrong with, and the ROI drops off the board. My figures for:
nDIST = 6
tSURF = "d"
nBLF >=30
rF1=1
returns this:
Win%: 13.1%
ROI: 0.44
However, I stress that the sample size of this last group is relatively small, compared to the rest of the samples, and it only takes one bomb to turn that into a positive expectation.
These were the results of my database, which uses different data than HTR users have, and the results with HTR may be different. Perhaps somebody can test this and let us know.
The point, however, is that if you are more likely to succeeed if you look for value within parameters that are sound in logic, but fall outside of the generally accepted beliefs of the public.