Longshots: Speed Improvement

[FakeNameChanged]

For anyone interested, I'd like to share a home grown system I've developed over the last decade. Background: While I'm fairly new to this board, my handicapping days go back about 35 years, more or less. Being a former engineer, numbers and statistics are what attracts me to the races, plus cashing a ticket. I apologize to any of the good handicappers on this board, if you feel my system is too rudimentary in this presentation.

So I made a lengthy study on winners, 4-1 odds or higher to see what factors showed up in their PP's. I included about 15 different factors, and two things stood out. My original study was over 500 winners at tracks all across the country. I've since done about three more follow up studies of 150, 200, and 100 winners and the numbers repeated almost to the exact percentages.

The number one factor hands down was speed improvement in it's last race. Speed improvement(SI) was evident in 65.6% of all winners studied. The number two factor was the horse made a middle move in its last race. MM being defined as a gain in either positions or lengths from any p.o.c to the next p.o.c., not including the stretch. Gains in the stretch were a separate factor. MM's were evident in about 60-61% of all longshot winners. Since SI was easier to quantify, I settled on using this one factor to find a probable longshot play.

My selection method was to quantify the speed number improvement in its last race from its previous race of similar distance. I only used two distance descriptions, either it was a sprint or a route to be similar. So 7f and below were sprints, and 1M and above was a route. If last race was 6f and no other races in the pp's was a sprint, then I used the 2nd last race as the base point, even if it was a route. My preference was to have the last two or three races at the same distance or similar distance. Today's race distance did not matter in identifying where the speed improvement was important. I also prefer the surface to be the same, but speed improvement going from dirt to turf has shown me to be a valid improvement. For speed improvement to be valid, the last race had to be in the last 35 days. I didn't arbitrarily pick that number but let the data show me what was a break before the data was invalid. My original training started with Tom Ainslie handicapping book, where he used 28 or 30 days to decide if a horse was a good play. I think he assigned extra credit for 7 days or less and deducted points if it was over 21 days, but my memory may be wrong.

I also found that you can use Beyers, the old track speed rating and variant, or the Equibase speed numbers to make your speed comparison. They all work equally well. My preference is the Equibase speed numbers but they all work, as long as you're consistent in your comparison. I don't look at times at all. I trust the speed numbers to represent the time and track bias, or variant for the day.

My preferred tracks to use Speed Improvement for longshots are Belmont, Aqueduct, PARX, Penn National, LRL, GP, DEL, and MNR. I've also used and had winners at Tampa, Arlington, GG, and SA. I generally don't use Sp Impr for G1 Stakes, but have had the winner in the Belmont stakes a few years back, and the exacta in the KY Dby or Preakness a couple years back when First Dude ran second. I gave that to a bunch of guys at OTB, and they thought I was gold. To be honest, my record in the Triple crown races sucks. I told them to box First Dude with the top three favorites, whom I couldn't separate.

I show a couple of recent examples so you can look at the PP's for yourself. I will only show horses with any Speed improvement in their last race. This method also id's the favorite. More often than not, the top SI horse is overlooked by the public. My apology for not using screen shots, since I don't subscribe to online PP's.

1st race at MNR(10-9-16)-7f-Turf-Clm 15000-14000:
#1 3-1 ML Red Dart: +21 SI(70 fig) lr.-both sprints
#5 9-5 ML Ship Ahoy +31 SI(62 fig) lr.-both routes
#9 10-1ML Santos Symphony +39 SI(56)-both sprints
This is not my ideal Speed improvement play since all three have very high SI's. I prefer only one or two with min. +5 SI or double digits, but it was still a play. Notice this horse had the lowest last race speed rating, which always gives me pause, but it was still a play.
Results
#9-$74.20-21.20-6.80
#5 - 3.80-2.40
#1 - -2.40
Exacta 307.40, Triple: 756.60

While these results are not normal, I just wanted to show that this method still picks longshots. And to be clear, I often box the SI play with top one or two favorites which both of the place and show horses were. I almost never play the triple, but regularly cash exacta tickets. Mountaineer has produced my biggest winners using this method, but it works quite well at big tracks like Belmont as well.

7th PARX(10-3-16)-10000 CLM-1M70Yd
#7-15-1ML-Big Gally's Answer +20 SI-(76 spd rat)-l.r.route, 2lr-7f.
Results #7 won $27.40-10.40-6.40
After I made this bet at the OTB, I went back to my seat and saw to my dismay that this trainer was 0/40 this year! I swore to myself, but said WTH, let it ride. Big Gally's Answer wired the field and hung onto win. I keep telling myself, that my old handicapping methods would have probably thrown this horse out. The 2nd SI horse at +6, ran 4th. This was a 14 point difference. It's not unusual to have only one SI in a field.

from the same day at PARX:
10th-PARX-10000 Clm-7f.
#3 10-1 ML-Fat Albert-+17 SI (76) p.t. odds 15-1-both races sprints
#6 7-2 ML-Only The Truth +16 SI (75) p.t. odds 7-2-both routes
#9 20-1ML-Thomas Knight +16 SI(88) p.t. odds 21-1-both routes
I dutched the top two odds' SI's to win and place. #9 also had a huge speed rating difference in its last race, which I occasionally use to break ties, since no horse had a +5 differential in SI. I often skip races with several SI with high ratings. (side note: #9's speed number l.r. was highest in race for all).
Results:
#9- $44.00-12.80-8.60; boxing these three didn't give me the exacta. Two nice longshots from one track in a day is very satisfying. This method also picks many horses paying 5.00 to 8.00, so it looks for the improving race horse.

Finally at Belmont:
2nd BEL(10-9-16)-Clm 16000-6-1/2f-
#2 6-1 ML-Sol The Freud +16 SI (93)-both races sprints-pp 6-1
#4 5-2 ML-You Know I Know +11 SI(100)-both races sprints-pp 2-1
#5 15-1ML-Because I'm Happy +19 SI (92)- " " " -pp 22-1
the #5 has +3 differential and +6 with the 2-1 odds #4.
Results
#5 won $46.80-14.80-5.70,
the 7-2 third fav. ran 2nd, and #2 showed at 5.10. I showed these races from yesterday and 10/3, to show it's still current. I have a database where I've recorded upwards of 300 actual winners I've hit over the years. True that many of them paid 7.00, but if they were the only play in the race with Sp Improv. then I played it if odds were 5-2 or higher.

The 64 dollar question will most likely be, what is the hit rate or ROI on this method? I don't have that answer, but the average mutual for all top SI's including low odds was $12.60. When I set the min. at 5-2, it rises considerably. I can't provide that number at the moment, as my old laptop crashed, and my new Chromebook does not have Excel on it, so my database is currently unavailable. I have a cloud on Microsoft, but it's not allowing me to access my Excel files, and I hate paying for something I've already bought. When it comes available, I will share that data.

I realize, many if not all of you probably program your handicapping and running a test is probably no big deal. My programming skills stop at Excel, and all my data crunching was done laboriously with hundreds of Racing Forms done on the kitchen table sometimes into the wee hours of morning. Any questions or criticism's, let me have them.

Let the beatings begin. Thanks for reading. Whosonfirst.

p.s. Sp Improv. is only valid if horse finished within 10 lengths in l.r. If horse was eased, impeded, or had any incident in 2nd l.r. I used the 3rd race back to make speed comparison.

[FakeNameChanged]

More Speed Improv plays at Belmont on 10/9/16:
Race 5-#6 +18/+13 placed to odds-on fav. #5, but the exacta paid $11.20 only six horses.
Race 6-#6 +16 won 14.00-7.00; #4 +13 place 5.60; exacta paid 70.00
3 winners including the two mentioned on previous post.

Belmont 10/10/16
Race 6-#2 +5 Won $4.10(only 4 horses)
Race 7-#11 +11 Won $5.20(again only 4 horses)
day not looking too good, several races with short fields sparse data in Mdns.
Race 10-Turf Mdn Clm 40K-#4 Road to Perfection +19 Won $59.00-18.60-5.70
#4 had a 59 speed rating(equibase), which was 2nd lowest out of 6 total horses. The #8 was also a +10 and finished last. Boxing #4 with five other horses in the exacta paid $290.00.

Here is how the LR speed ratings look for Race 10: #3-74, #4-59, #8-89, #11-84, #12-80, #13-27(comment dull fits), However when I look at the old DRF speed ratings, #4 was third highest at 82. I'm not sure why the equibase ratings don't track better with the old DRF ratings. Comparing speed Improv with the old DRF sp ratings show +16 for #4, so my original study still applies.
(edit): I forgot to mention that Maiden claiming races are some of my favorites to play. Last race top speed ratings win at decent prices in Mdn. claimers. MSW's not so much.

[maddog42]

Thanks for sharing whosonfirst.

[FakeNameChanged]

Originally Posted by maddog42

Thanks for sharing whosonfirst.

You're welcome maddog42. Hopefully it's something you can use with your regular h-capping.

One revision to my earlier edit. I looked back at my records and I still use the SI number even if the horse is 12 lengths from the winner in the last race. I know that seems like a lot, and I prefer it to be 8 lengths or less, but if a big Sp Impr shows and horse is 11-1/2 or even 12 l's from the winner, I use it. Will probably only be a minimal saver bet, "just in case". To be fair, I don't get a lot of +20's SI and 12 lengths behind, but it does happen. I've seen an SI at 17-1/2 lengths win, but that was probably just a random number occurrence than anything meaningful.
I also mostly avoid Stakes, where I said previously only G1. Graded Stakes are fun to watch, but my money usually stays on the sidelines.

[kevb]

How do you define speed improvement?

[maddog42]

Originally Posted by Whosonfirst

You're welcome maddog42. Hopefully it's something you can use with your regular h-capping.

One revision to my earlier edit. I looked back at my records and I still use the SI number even if the horse is 12 lengths from the winner in the last race. I know that seems like a lot, and I prefer it to be 8 lengths or less, but if a big Sp Impr shows and horse is 11-1/2 or even 12 l's from the winner, I use it. Will probably only be a minimal saver bet, "just in case". To be fair, I don't get a lot of +20's SI and 12 lengths behind, but it does happen. I've seen an SI at 17-1/2 lengths win, but that was probably just a random number occurrence than anything meaningful.
I also mostly avoid Stakes, where I said previously only G1. Graded Stakes are fun to watch, but my money usually stays on the sidelines.

With the decline of racing fields, I have been looking for ways to eliminate the favorite. One thing I tested was favorites that had been beaten 6 lengths or more in the previous race regardless of surface or distance. The favorites that had beaten by 6 lengths or more in the previous race fared very well, and in fact had a 2 percent higher win rate. Don't take this as gospel, since it was only a 300 race study, and 2 percent is not significant at all. There were favorites that won despite losing there previous race by 12,17 or even 20 lengths. Usually there was a mitigating factor such as trouble, wrong surface or distance.

[FakeNameChanged]

Originally Posted by kevb

How do you define speed improvement?

Maybe my explanation above was too obtuse. Let's try it like this.
2nd Race-Belmont(10/9/16)-6-1/2 furlong-Clm 16000

#5 Because I'm Happy
15-1 ML
25Sep16- 5Bel -6-1/2f-Clm12500 - 92 - fin 3 3-3/4 l's
31Jly16 - 5Sar-6-1/2f-Clm20000 - 73 - fin 5 12 l's

Subtract the 73 speed fig from 92 fig in last race and you get +19 Speed Improvement number. Since it was less that 35 days from today's race and it's our highest number within 12 lengths, it's our play. The fact that last race was a class drop from 20 down to 12.5K does not matter to me. And the rise in class this race is actually a plus in my mind. It helps to keep the public off my play, and it's what sharp trainers often do.

[jpjpicks]

It's nice to see some handicappers thinking outside the box !!! whosonfirst like you said reg handicapping would have not got some of the horses. like I have found out horses that run out last race but have good turn time plus late pace tend to run good next time out when you play system you got to take the good an bad
Ps: whosonfirst keep up the thinking an work very nice of you to share your system

[thaskalos]

I don't know if I understand you completely here. What you call "speed improvement" is just a speed rating comparison, right? You just take the last race speed rating, and you compare it to the SECOND-last race speed rating...while looking for a numerical "improvement" in the most recent start.

What if the second-last race is atypically slow for the horse...while his last race was a normal one? Wouldn't that lead to a misleadingly-high "improvement" indicator? What I mean is...what if the last 4 speed ratings for the horse are 78, 61, 74, 80. You would compare the most recent one (78) to the one right before that (61), and you would conclude that there was an improvement of 17 points...while the truth is that the second-last figure is suspiciously low, and the horse ran an even bigger rating in the fourth race back. Is this really an "improvement"? And...what if there are horses in the same race today with much higher speed ratings than the 78 or the 80 that our horse here shows?

IMO...you are placing all your eggs in the speed-rating basket...and I can't call that a good maneuver. A lot of different factors go into the creation of the speed figure...and the speed figure alone is not accurate enough to determine a horse's improvement, or its decline. Speed figures, although important, cannot stand alone in the handicapping process.

[pandy]

Whosonfirst, congratulations on a very good post. You did your homework.

[pandy]

Originally Posted by thaskalos

I don't know if I understand you completely here. What you call "speed improvement" is just a speed rating comparison, right? You just take the last race speed rating, and you compare it to the SECOND-last race speed rating...while looking for a numerical "improvement" in the most recent start.

What if the second-last race is atypically slow for the horse...while his last race was a normal one? Wouldn't that lead to a misleadingly-high "improvement" indicator? What I mean is...what if the last 4 speed ratings for the horse are 78, 61, 74, 80. You would compare the most recent one (78) to the one right before that (61), and you would conclude that there was an improvement of 17 points...while the truth is that the second-last figure is suspiciously low, and the horse ran an even bigger rating in the fourth race back. Is this really an "improvement"? And...what if there are horses in the same race today with much higher speed ratings than the 78 or the 80 that our horse here shows?

IMO...you are placing all your eggs in the speed-rating basket...and I can't call that a good maneuver. A lot of different factors go into the creation of the speed figure...and the speed figure alone is not accurate enough to determine a horse's improvement, or its decline. Speed figures, although important, cannot stand alone in the handicapping process.

He's not always using the last two races, because he's only using speed figures from comparative distances. So if a horse raced in a sprint last race, but in a route the race before that, then in a sprint three starts back, he'll use the sprint from three races back...he's looking for improvement from the prior sprint race, or the prior race at a similar distance.

[thaskalos]

Originally Posted by pandy

He's not always using the last two races, because he's only using speed figures from comparative distances. So if a horse raced in a sprint last race, but in a route the race before that, then in a sprint three starts back, he'll use the sprint from three races back...he's looking for improvement from the prior sprint race, or the prior race at a similar distance.

I understand that...I read his initial post. He is making distinctions between sprints and routes...but he is still depending on the speed figures for the entirety of his handicapping. You can add, subtract, divide, average, or do whatever you want to those speed figures...and they STILL won't lead you to the promised land. The speed figures are only a PART of the handicapping puzzle...they can't be the only thing that the horseplayer is concerned with.

[pandy]

Originally Posted by thaskalos

I understand that...I read his initial post. He is making distinctions between sprints and routes...but he is still depending on the speed figures for the entirety of his handicapping. You can add, subtract, divide, average, or do whatever you want to those speed figures...and they STILL won't lead you to the promised land. The speed figures are only a PART of the handicapping puzzle...they can't be the only thing that the horseplayer is concerned with.

He's using this as a spot play longshot method. If I were going to use just one factor, it would be pace figures, but as I've stated before, I think you could just use speed figures if you have a good method of interpreting and analyzing the speed figures. Again, just for spot plays. Obviously, if you just bet all of the top figure horses you will lose. But there are other ways of using the figures.

[ReplayRandall]

Originally Posted by thaskalos

The speed figures are only a PART of the handicapping puzzle...they can't be the only thing that the horseplayer is concerned with.

Of course, speed figures are just a part of the handicapping puzzle, Gus. But when a player starts the process, Whosonfirst wrote a well thought plan to at least give him the "improving" contenders, with his methodology. From there, anyone can go in any direction they choose to arrive at possible plays which have long-term positive value......I give whosonfirst credit for a nice starting approach...

[thaskalos]

Originally Posted by pandy

He's using this as a spot play longshot method. If I were going to use just one factor, it would be pace figures, but as I've stated before, I think you could just use speed figures if you have a good method of interpreting and analyzing the speed figures. Again, just for spot plays. Obviously, if you just bet all of the top figure horses you will lose. But there are other ways of using the figures.

I read his post very carefully...and he never used the term "spot play" even once. Nor does he say that he has "other methods" in his handicapping arsenal. He said that this "home grown system" was his "selection method". As far as I -- or anyone else -- could tell...this might be the OP's only handicapping method. And that's why I responded in the way that I did.

In Andy Beyer's book Beyer on Speed, he talks about the myriad ways that horseplayers have invented in order to properly analyze these numbers. Beyer has tried all these ways too...and so have I. When the Beyer figures first got included in the Racing Times...I thought that I finally had the "key to the mint". For YEARS I sliced and diced these speed figures, looking for that magic combination...and I even considered the "improvement" angle that the OP describes here.

Alas...this game is a little more complicated than that.