Frank N
2nd December 2001, 05:58.53 PM
A simple query for Access users to try on larger databases than Gramps has.
nDIST: 6 or 6.5 or 7
tSURF: "d"
nFAV: 0
tRS: ="*e"
nQPT: 3 or 4
F1: <=4
EP: <=4
ACL: <=5
The results in Gramps's database were
Plays: 347
Win%: 21.9%
ROI: 1.52
That is out of about 7200 total races at those distances, so it finds about one play per 20 such races.
I also found out that this worked well at 5 furlongs and shorter. However, it did very poorly at 5.5 furlongs. My own belief, based on the logic to be presented, is that it won't do as well at distances shorter than 6 furlongs as it does at sprints of 6, 6.5, and 7 furlongs. Maybe somebody else can test this on a larger database.
Here, factor by factor, is the logic behind the query:
The distance and surface are, of course, sprints of conventional distances raced everyday, and this is the basis of the other factors.
nFAV: Horses who were favored their last time tend to draw money today as well, whether they won or not. The public believes that the "smart money" was sent in then, and they react accordingly. Also, horses favored in their last race often get listed at lower morning line odds by the trackman, and this results in lower final odds. If we eliminate these horses, we eliminate, by and large, overbet undervalued types. So we set this to "0", which means we don't want horses who were favored in their last race.
tRS="*e": We're looking for the horse that is believed to be the early leader today. But we're also looking for something else, which will be discussed next. As you'll understand when you've read the next paragraph, we don't want a confirmed front runner (*f), nor do we want pressers, sustained, or rear-running horses.
nQPT=3 or 4: This is the key to the play. What we're looking for are horses who show some early foot, but not so much that everybody else sees it. The fact that the horse is considered to be the likely leader tells us that its early speed is good enough to get to the top of this field. But the fact that it doesn't ALWAYS run early means that the public won't be keying on an obvious front runner. Also, if the horse doesn't always get the lead, it suggests a horse that might be able to run flexibly and sit off of a hot pace today. We want a horse that figures to be ABLE to get to the lead, hence we don't use plain "e" horses (horses who run early but don't figure to get the early lead). But we also want some flexibility, so we don't include "*f" horses. a "*f" horse is both overbet and inflexible. If some jockey/trainer decides to send his/her horse to the lead today and contest a "*f" horse, it will hurt its chances. The fact that the horse figures to get the lead with only 3 or 4 Quirin points suggests that the race is fairly void of early speed, which gives the jockey of our horse more options to choose from as far as setting the pace or laying just off of a suicidal one.
In the database, QPT=2 also worked for a small profit, and perhaps somebody else can test that. QPT=5 did not, mainly because at 5-8 Quirin points, the fact that the horse can run early gets more and more obvious to the public. What's obvious to the public gets bet down.
F1<=4 and EP<=4: Simply, we want to be sure that the horse has enough early foot to the first and second calls to be within striking range, if not on the lead.
ACL<=5: We want to be sure that the horse has raced well against competition in the same ballpark as the other horses in the race.
Perhaps somebody would care to try this in a larger database and report the findings. I'm pretty sure that the ROI is a bit high in the small sample we have, and I'd think something like 1.30 (possibly a little lower) would be more in line with reality.
--------------
As an addendum, requiring minimum off-odds of at least 3-1 produces the following:
Plays: 208
Win%: 18.3%
ROI: 1.92
In this and the original sample were two $60 horses and a $112 horse. This isn't out of line with reality, in my opinion; however, a larger sample might show a lesser ROI.
nDIST: 6 or 6.5 or 7
tSURF: "d"
nFAV: 0
tRS: ="*e"
nQPT: 3 or 4
F1: <=4
EP: <=4
ACL: <=5
The results in Gramps's database were
Plays: 347
Win%: 21.9%
ROI: 1.52
That is out of about 7200 total races at those distances, so it finds about one play per 20 such races.
I also found out that this worked well at 5 furlongs and shorter. However, it did very poorly at 5.5 furlongs. My own belief, based on the logic to be presented, is that it won't do as well at distances shorter than 6 furlongs as it does at sprints of 6, 6.5, and 7 furlongs. Maybe somebody else can test this on a larger database.
Here, factor by factor, is the logic behind the query:
The distance and surface are, of course, sprints of conventional distances raced everyday, and this is the basis of the other factors.
nFAV: Horses who were favored their last time tend to draw money today as well, whether they won or not. The public believes that the "smart money" was sent in then, and they react accordingly. Also, horses favored in their last race often get listed at lower morning line odds by the trackman, and this results in lower final odds. If we eliminate these horses, we eliminate, by and large, overbet undervalued types. So we set this to "0", which means we don't want horses who were favored in their last race.
tRS="*e": We're looking for the horse that is believed to be the early leader today. But we're also looking for something else, which will be discussed next. As you'll understand when you've read the next paragraph, we don't want a confirmed front runner (*f), nor do we want pressers, sustained, or rear-running horses.
nQPT=3 or 4: This is the key to the play. What we're looking for are horses who show some early foot, but not so much that everybody else sees it. The fact that the horse is considered to be the likely leader tells us that its early speed is good enough to get to the top of this field. But the fact that it doesn't ALWAYS run early means that the public won't be keying on an obvious front runner. Also, if the horse doesn't always get the lead, it suggests a horse that might be able to run flexibly and sit off of a hot pace today. We want a horse that figures to be ABLE to get to the lead, hence we don't use plain "e" horses (horses who run early but don't figure to get the early lead). But we also want some flexibility, so we don't include "*f" horses. a "*f" horse is both overbet and inflexible. If some jockey/trainer decides to send his/her horse to the lead today and contest a "*f" horse, it will hurt its chances. The fact that the horse figures to get the lead with only 3 or 4 Quirin points suggests that the race is fairly void of early speed, which gives the jockey of our horse more options to choose from as far as setting the pace or laying just off of a suicidal one.
In the database, QPT=2 also worked for a small profit, and perhaps somebody else can test that. QPT=5 did not, mainly because at 5-8 Quirin points, the fact that the horse can run early gets more and more obvious to the public. What's obvious to the public gets bet down.
F1<=4 and EP<=4: Simply, we want to be sure that the horse has enough early foot to the first and second calls to be within striking range, if not on the lead.
ACL<=5: We want to be sure that the horse has raced well against competition in the same ballpark as the other horses in the race.
Perhaps somebody would care to try this in a larger database and report the findings. I'm pretty sure that the ROI is a bit high in the small sample we have, and I'd think something like 1.30 (possibly a little lower) would be more in line with reality.
--------------
As an addendum, requiring minimum off-odds of at least 3-1 produces the following:
Plays: 208
Win%: 18.3%
ROI: 1.92
In this and the original sample were two $60 horses and a $112 horse. This isn't out of line with reality, in my opinion; however, a larger sample might show a lesser ROI.