Jockey Advantage

[Que]

Does anyone handicap strictly by jockeys? I’ve been debating the advantages/disadvantages of this type of handicapping for some time. To test if it’s a viable handicapping alternative I created an approximately 2,600 race database. I then divided the data into a training set of about 2,100 races, and a test set of about 500 races. I then used a genetic algorithm to optimize 83 different jockey stats in the training set. The jockey stats run from very generic, to very race/horse specific. Here some of my observations:

Observation. Overall results look promising, i.e. playing the highest rated jockey in every race, ROI to win was +3% for the training set, and +13% for the test set

Observation. Win percentages (see below) were consistent between both training and test sets. For the most part, the jockeys ranking directly correlate with the actual win percentages. This fact indicates that it may be possible to create a rating system to rank the jockeys (aka horses) in any given race.

Win percentages:
Train Test
Rank 1 14.9 16.9
Rank 2 14.6 14.0
Rank 3 13.7 13.7
Rank 4 14.3 11.5
Rank 5 12.7 11.4
Rank 6 11.8 12.1
Rank 7 9.8 9.7
Rank 8 8.0 9.7
Rank 9 6.8 7.5
Rank 10 6.0 3.2


Observation. Expected win percentages (below) were consistently inconsistent between both training and test sets. For the most part, the public is unable to distinguish between the top five choices based on jockeys alone. This fact, combined with the above observation reinforces my belief that it may be possible to create a profitable jockey rating method.

Expected win percentages (based on final odds):
Train Test
Rank 1 12.8 13.3
Rank 2 13.9 14.6
Rank 3 14.6 14.4
Rank 4 14.9 13.9
Rank 5 13.2 13.8
Rank 6 12.3 11.5
Rank 7 10.3 10.9
Rank 8 9.4 8.1
Rank 9 7.1 7.1
Rank 10 6.7 5.6


Observations. Several factors weighted most heavily in the rating include: the jockey’s record in sprint or routes; the horse’s early speed (indicated by the jockey’s record broken out by the horse’s early speed ranking, and the horse's running position at the first and second call in it’s last race), the horse’s running style, and post position. Several interesting factors that popped up include the horse’s number of workouts since its last race, whether the horse is wearing mud calks, and a combination factor (indicated by the horse's layoff/class change/race type). The factors weighted most heavily, not listed in any particular order, are shown below:

Distance (sprint or route)
Distance (sprint or route) & distance change from horse’s last race
Horse’s Layoff & class change & race type (i.e. clm, alw, mdn, etc.)
Equipment (mud calks)
First call position in horse’s last race
Second call position and finish in horse’s last race
Second call position (sprint/route) and finish in horse’s last race
Odds & finish position in horse’s last race
Horse’s number of workouts since last race
Horse’s running style (sprint or route)
Post position
Post position (sprint or route) & horse’s early speed ranking
Post position (sprint or route), horse’s running style & early speed ranking
Trainer


If anyone has any experience handicapping jockeys, or has any comments on the above, please respond. I'd be interested in reviewing your comments and/or personal observations.

best regards,

Que.

[ridersup]

Que

Just finnished competing in the NTRA head to head contest where you had to pick a winner of two horses in selected races on 4 weekends. Over 5,000 entries and the top 4 got the trip to National Handicapping contest. The winners had 16 correct selections.

Using only Jockey-Trainer and morning line from TSN I got 15 correct. I don't have a data base but from experience I consider the jockey as one of the key factors in any final selection among contenders.

[BIG HIT]

It said some where can't remenber where that the top five jocky win about 80% of race's.Think they were speaking of smaller track's.But have had good luck a better track's two.If you check you will be surpprized at some of the price's.Good luck in your quest.Another strong jocky angle is a jocky that is rideing a horse last three or more times never out of money and rideing him today.

[GR1@HTR]

Q,

Good stats. I keep numbers on Jkys and have used them w/ moderate success but they are very volitle. Jky stats change drastically from month to month and track to track. I assume that what causes these changes range from jockey confidence, jockey agent, and jockey change in tactics.

A few examples, last year Nunez at CRC was 1/91 at one time riding turfers. Now he wins his fair share on the grass.

G Melancon and Theroit were spare to average riders until they started getting mounts from Cole Norman in LAD. Now they are 35% to 40% winners on his horses. Are they really that good? Just step it up when they get his mounts? Who knows…

[karlskorner]

Congratulations. At last, after all this time, someone is paying attention to the human being on board. Who and Why are pieces to fit in the puzzle.

Karl

[Jaguar]

Que,

You might find Mike Nunamaker's great study on impact values very useful.

Also, if you know anyone with an old Multi-Strats(Titan Series) handicapping disc in their diskette box, take a look at the jockey/trainer algorithm the program uses, it's excellent.

Today connections are so important in the racing game, especially in races where the horses are sound. The jockey/trainer patterns in Maiden Special Weight, Allowance, and Stakes races reveal when a conditioner is dropping an animal in for the tag.

After all is said and done, even though we love the challenge and romance of the racing game, it's all about money. Trainers often tip their hand and my computer sometimes catches them at it.

It's quite satisfying when the machine assigns the jockey/trainer combo a big number, and time to take the rubber band off the bankroll.

All the best,

Jaguar

[Rick]

Que,

Have you checked jockey's ROI? Another one that may be helpful is number of starts by the jockey for the trainer, especially for first-time starters.

It's interesting that a lot of the significant categories are things that are commonly available for trainers but not for jockeys. I suppose it's not surprising that information that most others don't have would be more valuable.

Jockey stats are one of very few things that can improve the ROI on favorites and anything that improves favorites will probably improve your top selections.

[Que]

Thanks for the replys. I'll keep researching it... but, the initial stats look promising. My only concern is the low win percentage... I've typically found methods with positive ROIs, but have win percentages below 15% are above my pain threshold... i.e. too many long losing streaks with a few bombs in between.

Rick, yes I do check Jockey ROIs. In fact, all my Jockey angles and Owner angles are mirror images of my trainer angles. I found it was easier to change a few lines of code and track jockeys and owners using the same program and database that I use for tracking trainers, rather than develop separate databases/programs. Also, some of the results you get on going far beyond simple jockey/owner stats is very interesting.. albeit you need to pass it through the common sense test. But I think it brings in the jockey agent angles... assuming good jockey agents are knowledgable enough to find their jockey good mounts.

Que.

[hurrikane]

Great stuff Que, as usual. A few people over at HTR have been using jockey data with some success. Most of the ties in the HTR data are decided by jockey ranking. Massa also developed a Jky ranking number similar to what you are doing that takes several things into account and give you a kind of batting average/rating number.
It's not my intention to sell you on HTR but just to say I find the jockey perspective is as important as the trn/own.
Only thing you might add if you have'nt already.
Impact values. especially when you are talking run styles or early speed. Is it a function of the track or the jockey?

[karlskorner]

If "Pace makes the race", who sets the pace
If the rail is the shortest distance home, who goes to the rail
If a small opening appears, who guides the horse through
If the horse loses interest at the top of the stretch, who gets his attention with a whip

The Jockey. Don't ignore him/her

Karl

[Larry Hamilton]

I like your technique of making a point...having said that, you left out a question

If he follows instructions exactlly, who gives him another ride

[karlskorner]

Becoming a good Jockey is part of an all important area known as "judgment", a skill that comes through experience and exhausitive study of racing films. Inexperience and incompetent riders are forever getting trapped by a wall of horses, a situation known as a "switch". The best riders are the ones who keep their horses out of trouble by anticipating the moves and strategies of other riders. Part of the "judgement" necessary requires a sense of PACE. "A clock in his head" Devloping the "clock" takes untold hours of hard work, the jockey first counts mentally while someone holds a stopwatch. Then spends countless hours on the track translating that sense to the race track. The lack of a sense of PACE is why the only strategy many jockeys are capable of is pushing a horse as hard as they can until it stops. While this 'strategy" wins some sprints, it seldom wins races over a mile.

Implementing propery strategy requires mastering proper "technique" This means what horsemen call a good "seat" on a horse, sitting low, to cut the wind risistance and as still as possible to avoid disrupting the horse's running rythm and balance. Technique also means learning to control a horse in the starting gate and the ability to get the horse out of the gate in good order. It means handling the reins properly to get the horse to relax, to accelerate or to change "leads" around a turn. Finally it means handling a whip properly and being able to change over instantly from whipping with one hand to using the other when the need arises. Many jockeys are able to whip only with one hand, a severe disadvantage. Great jockey add a final quality known as "great hands" or "feel". A communication between horse and rider, that coaxes the maximum performance from an animal, something that many riders sorely lack.

Don't ignore the Jockey

Karl

[Larry Hamilton]

What you are describing are the qualities and abilities of the athlete (in almost any sport).
So the real underlying argument is who is important--coach or player. The answer is both, the rider and horse are the athlete, the trainer is the brains.

The problem here is circular. How do you guage the ability of the jock when he is restrained by the trainer and how do you guage the intent of the trainer when he is limited by the abilities of the rider and horse.

A side bar: Andy Beyer called the jocks "pinheads" in one of his speed books

[Rick]

Que,

Does previous jockey ROI predict future ROI to any significant degree?

[karlskorner]

The closest Beyer has ever been to a horse was his failure to jump over the "droppings" in the paddock. I don't think his mind is of the same set as it was 20 years ago, especially when a good number of the jocks earn much more than he as a word jockey. "Pinhead" is a good word when somebody does something that you don't understand.

Karl

[JimG]

When I was doing mostly pen and paper handicapping, jockey/trainer combos were a big part of the handicapping equation for me. Since I have drifted predominantly to computer handicapping and trying to find value from the contenders, I have gotten away from that.

Jockey/trainers are more of a tiebreaker for me now between horses I deem as close in ability.

Always give a strong look to a trainer that switches to his "go to" jockey for the upcoming race.

Jim

[Que]

Rick,

You asked, "Does previous jockey ROI predict future ROI to any significant degree?" That really is the one million dollar question. It’s also a question that I don’t have an easy answer to. As an example, let’s examine the winning jockey’s of Friday’s 1st race at Belmont Park (which I selected randomly—trust me).

Castellano Javier -- Positive ROI angle’s before that race (with < 30 mounts)


Record for 2002 (prior to race)
Belmont Park (no track change) 207 29-36-33 1.09
Odds: 2/1 to 4/1 202 54-51-33 1.10
Odds: 2/1 to 4/1 & route 103 30-23-20 1.24
Odds 2/1 to 4/1 & horse’s last race finish 4th-5th 39 13-7-6 1.39
October 44 8-8-7 1.32
Post 1 & route 42 12-6-4 1.34



Record for 2001
Belmont Park (no track change) 190 19-24-29 0.74
Odds: 2/1 to 4/1 239 42-60-44 0.76
Odds: 2/1 to 4/1 & route 109 15-30-18 0.61
Odds 2/1 to 4/1 & horse’s last race finish 4th-5th 54 11-18-6 0.91
October 84 4-11-16 0.20
Post 1 & route 90 14-15-11 0.75



Record for 2000
Belmont Park (no track change) n/a
Odds: 2/1 to 4/1 529 93-93-111 0.69
Odds: 2/1 to 4/1 & route 191 35-30-42 0.73
Odds 2/1 to 4/1 & horse’s last race finish 4th-5th 101 16-17-21 0.65
October 136 19-23-21 0.37
Post 1 & route 86 13-10-17 0.69



Record for 1999
Belmont Park (no track change) n/a
Odds: 2/1 to 4/1 125 29-20-23 0.89
Odds: 2/1 to 4/1 & route 50 15-8-3 1.23
Odds 2/1 to 4/1 & horse’s last race finish 4th-5th 16 7-3-0 1.88
October 38 7-10-6 0.45
Post 1 & route 26 8-5-3 1.75


Since this is only one jockey, in one race, it’s impossible to make any conclusions from the above data, but it does illustrate the complexity of answering this question. For example, how far do you go back to calculate the ROI, e.g. three years, two years, one year, 90 days, 30 days. Etc. Does Castellano’s stats in 2002 mostly at the NYRA have any correlation to his earlier stats at Crc and GP in 2000? Plus, if you use a very small time window are you playing what is really hot or are you fooling yourself and just playing random noise? The answers to these questions are just as important as the question of what are the most significant jockey angles to follow. And even more bothersome is that a significant angle for one jockey, maybe an insignificant angle to another jockey. For example, some jockeys may favor horse’s with certain running styles, while for other jockeys it really doesn’t matter.

I’m sure all of you already know this, therefore I’m probably just preaching to the choir, but unfortunately this game has no easy answers.

Que.

[ridersup]

In one of his books Beyer severely criticized Mario Pino and I believe he recently got his 5,000th win.

[Rick]

Que,

Yeah, I realize that there are countless ways you could break it down, but I was interested in whether there was any value in the most readily available stats. For me that would be for the previous year since I use BRIS past performances. I have found some value in setting minimum requirements for jockeys in order to qualify selections but haven't incorporated any sort of rating so far. My impression from past studies has been that those with the best stats get the best horses anyway, so it doesn't usually improve ROI to give extra credit for impressive stats. But, those with really bad stats do negatively affect the ROI.

[Boris]

Que,

Perhaps an easier question to answer, perhaps not. Look at the jockeys first ride of the day. I view it as the "why did the jockey come in early for this race?" angle. Cheaper the race the better relative to other track purses. Focus on top 10 jocks at the track where they have a mount in race 1 or 2. Minimum odds of 5-1. Prefer a gap of two races before their second mount of the day, but not a hard rule. In other words, if they skipped this mount they could have taken a nap.

I'm not playing Belmont right now, but noticed Luzzi brought home a couple of bombs last week in his first ride of the day. Anyway, another way to look at it.

[Larry Hamilton]

I would suggest to youze guys tring to ferret out a sweet angle about jocks dont consider the following:

Trainer, Track, Distance, Class, Type, Track Condition, EPS

Why? These simplistic connections of how well a jock did given these "filters" has been beaten to death and has yeilded nothing yet.

Ok, what then? How about something like this. How well does this jock do when he leads at the half on an E horse. How well does this jock do when trailing by 5 at the half on an S horse. What I suggesting is that you try to find value in how well a jock does on a specific running style at a specified distance ahead or behind at the half, 1st call or top of the stretch...THis addresses the jocks skill. Then wrap each of your results with a standard deviation (of beaten lengths) and you have some idea if the jock contributes or is along for the ride.

When you have that roughed out THEN use the old standbys mentioned above.