Understanding ROI using Jockey Stats

[Dan H]

I'm having trouble understanding jockey stats, specifically ROI. I have searched this forum and the web, but most answers provide the "how to" and not the "what it means" to a handicapper. Please help me understand.

Here's two examples from Brisnet PPs:

JKYw/ Turf 369 16% 42% -0.28 (GAFFALIONE)

JKYw/ Turf 117 15% 32% +0.28 (LEZCANO)

The Brisnet explanation: Number of mounts, percentage of winners, percentage of in the money finishes, and return on investment for every $2.00 wagered.

In narrative form, can I say that Gaffalione won 16% (ITM 42%) of his 369 turf mounts for the year - and that if I had bet $2.00 to win on all 369 of his mounts, I would have lost .28 cents per bet?

For Lezcano, I would have won .28 cents per bet?

Here's the dumb-dumb questions?

1. If chasing returns, is Lezcano a better investment on turf?

2. Does the -0.28 ROI suggest Gaffalione wins on lower odds horses on turf?

3. Dows the +0.28 ROI suggest Lezcano wins on higher odds horses on turf?

Thanks in advance.

[cj]

You have it right as far as what it means. Of course, past returns are no guarantee of future performance.

1) Yes, definitely for now, though bettors usually catch on
2) Definitely, since he has a slightly higher win percentage, his winners are paying substantially less.
3) See 2 in reverse.

[Dan H]

Thank you, CJ.

[davew]

lets breakdown Lezcano's stats some

117 races with 15% win rate near 17 wins
.28 per race on 117 races near $32.80

$234 bet on those 117 races returned $266.80 on 17 winners


those winners could be mostly low with a couple very high or many high

a database person could probably list what they were

[Dan H]

Originally Posted by davew

those winners could be mostly low with a couple very high or many high

Good point, thanks. No need to query any database on my behalf.

I have never really used jockey stats as a data point when handicapping, but every time I go to the OTB, I can't help but overhear bettors talking about jockeys and not horses.

And then Saratoga happened.

Television pundits argued about Irad's aggressive riding style and it made me think that maybe there's something there. So I asked the question here.

But, like you (and CJ) inferred, jockey stats listed in Brisnet and DRF don't really tell the whole story.

Thanks for helping me understand.

[classhandicapper]

I recently did some regression analysis using Beyer Speed Figures, my own Class/Form Rating, Jockey Rating, Trainer Rating, Running Style, and Consistency Rating.

The Jockey factor was more important than I thought. It was worth about 17% of the overall rating on dirt.

Of course, these were all my own ratings and the study didn't include a lot of other relevant factors, but 17% is a lot given that list.

As to whether there is any value to be found, I think you are going to have to dig pretty deeply because there's plenty of data out there on jockeys and the public picks up on obvious things quickly.

[MJC922]

Originally Posted by classhandicapper

I recently did some regression analysis using Beyer Speed Figures, my own Class/Form Rating, Jockey Rating, Trainer Rating, Running Style, and Consistency Rating.

The Jockey factor was more important than I thought. It was worth about 17% of the overall rating on dirt.

Of course, these were all my own ratings and the study didn't include a lot of other relevant factors, but 17% is a lot given that list.

As to whether there is any value to be found, I think you are going to have to dig pretty deeply because there's plenty of data out there on jockeys and the public picks up on obvious things quickly.

It's tricky with racing studies because the horses with better figs tend to be the ones with the better trainers and jockeys (can lead to multicollinearity problems). I find the class rating wins 30% all by itself and everything else combined can barely move that needle upward by 1 or 2% more. Maybe I'm doing something wrong. Probably... but I think the weaker the fig the more the rest of it can prop things up. Not saying the Beyer is weak at mid 20s (what's that get these days like 27%?) but say for example the DRF SR+TV would be in the low 20s for win rate so combine that with say avg earnings per start and suddenly avg earnings appears to be more potent than it otherwise would.

[Dan H]

Here's an idea that stems from my inability to find place and show horses with any consistency. I am hoping jockey stats (with other methods) can help me find them. I copied these stats from Brisnet PPs:

Wednesday, September 13, 2023 Race 6

# 1 - JKYw/ Turf 116 13% 33% +0.06 (Corrales)
# 2 - JKYw/ Turf 129 22% 50% +1.39 (Cannon)
# 3 - JKYw/ Turf 146 15% 45% - 0.62 (Geroux)
# 4 - JKYw/ Turf 210 12% 40% - 0.67 (Velazquez)
# 5 - JKYw/ Turf 369 16% 42% - 0.28 (Gaffalione)
# 6 - JKYw/ Turf 460 16% 48% - 0.30 (Ortiz, J)
# 7 - JKYw/ Turf 095 15% 31% +0.39 (Talamo)
# 8 - JKYw/ Turf 495 24% 57% - 0.37 (Ortiz, I)
# 9 - JKYw/ Turf 117 15% 32% +0.28 (Lezcano)
#10 - JKYw/ Turf 343 14% 42% -0.71 (Rosario)

My handicapping process flows from identifying vulnerable/false favorites to using pencil-and-paper to rank winners. I'm no Jimmy the Greek but I'm holding my own on winner selections (mostly chalk).

My betting schema is based on horses ranked A, B, C, D:

$3.00 Win A
$2.00 Exa A with B
$1.00 Exa A with CD
$1.00 Exa B with ACD

To get better at my B, C, and D picks, I'm going to use Jockey stats as a factor (not looking for 3rd place in my schema).

The 6 and 3 are M/L top two favorites. Assume my handicapping process lands me on the top favorite as my A horse, and depending on matching/contrasting running styles and pace shapes, the second favorite is my B horse.

If I use the ITM jockey stats from the list above, and couple them with other ranking factors, my C and D horses lean towards the 2 (50% ITM) and the 8 (57% ITM). I do realize ITM can also mean 3rd place.

So, my hypothetical ABCD is 6-3-2-8.

Any thoughts on using these jockey stats to find place and show horses?

[classhandicapper]

Originally Posted by MJC922

It's tricky with racing studies because the horses with better figs tend to be the ones with the better trainers and jockeys (can lead to multicollinearity problems). I find the class rating wins 30% all by itself and everything else combined can barely move that needle upward by 1 or 2% more. Maybe I'm doing something wrong. Probably... but I think the weaker the fig the more the rest of it can prop things up. Not saying the Beyer is weak at mid 20s (what's that get these days like 27%?) but say for example the DRF SR+TV would be in the low 20s for win rate so combine that with say avg earnings per start and suddenly avg earnings appears to be more potent than it otherwise would.

I just restarted this project yesterday and I'm getting quite a few interesting results.

The first time I ran the regression it was for all dirt stakes.

I tested the results using the weights it suggested and was very satisfied with the win% and roi given this is still pretty basic.

Yesterday I ran it again, but this time it was for just dirt sprint stakes. The weights were very similar, but to my disappointment, the dirt sprint stakes did a hair better with the original weights than the did with those calculated specifically for dirt sprints. I expected improvement. I'm not sure how that's even possible because I have a huge sample of races going all the way back to 2015.

I'm going to do more research tomorrow.

[MJC922]

Originally Posted by classhandicapper

I just restarted this project yesterday and I'm getting quite a few interesting results.

The first time I ran the regression it was for all dirt stakes.

I tested the results using the weights it suggested and was very satisfied with the win% and roi given this is still pretty basic.

Yesterday I ran it again, but this time it was for just dirt sprint stakes. The weights were very similar, but to my disappointment, the dirt sprint stakes did a hair better with the original weights than the did with those calculated specifically for dirt sprints. I expected improvement. I'm not sure how that's even possible because I have a huge sample of races going all the way back to 2015.

I'm going to do more research tomorrow.

Good luck, you might be on to something. I had jocks show up better than expected in some of my myriad studies too. At the time I chalked it up to that factor tying in with 'intent' whereas the trainer factor by itself is kind of detached from intent. Maybe intent in stakes is less of 'thing', unless you're prepping Mage in a grade one. Complex game we play here..

[MJC922]

Originally Posted by Dan H

Here's an idea that stems from my inability to find place and show horses with any consistency. I am hoping jockey stats (with other methods) can help me find them. I copied these stats from Brisnet PPs:

Wednesday, September 13, 2023 Race 6

# 1 - JKYw/ Turf 116 13% 33% +0.06 (Corrales)
# 2 - JKYw/ Turf 129 22% 50% +1.39 (Cannon)
# 3 - JKYw/ Turf 146 15% 45% - 0.62 (Geroux)
# 4 - JKYw/ Turf 210 12% 40% - 0.67 (Velazquez)
# 5 - JKYw/ Turf 369 16% 42% - 0.28 (Gaffalione)
# 6 - JKYw/ Turf 460 16% 48% - 0.30 (Ortiz, J)
# 7 - JKYw/ Turf 095 15% 31% +0.39 (Talamo)
# 8 - JKYw/ Turf 495 24% 57% - 0.37 (Ortiz, I)
# 9 - JKYw/ Turf 117 15% 32% +0.28 (Lezcano)
#10 - JKYw/ Turf 343 14% 42% -0.71 (Rosario)

My handicapping process flows from identifying vulnerable/false favorites to using pencil-and-paper to rank winners. I'm no Jimmy the Greek but I'm holding my own on winner selections (mostly chalk).

My betting schema is based on horses ranked A, B, C, D:

$3.00 Win A
$2.00 Exa A with B
$1.00 Exa A with CD
$1.00 Exa B with ACD

To get better at my B, C, and D picks, I'm going to use Jockey stats as a factor (not looking for 3rd place in my schema).

The 6 and 3 are M/L top two favorites. Assume my handicapping process lands me on the top favorite as my A horse, and depending on matching/contrasting running styles and pace shapes, the second favorite is my B horse.

If I use the ITM jockey stats from the list above, and couple them with other ranking factors, my C and D horses lean towards the 2 (50% ITM) and the 8 (57% ITM). I do realize ITM can also mean 3rd place.

So, my hypothetical ABCD is 6-3-2-8.

Any thoughts on using these jockey stats to find place and show horses?

It might matter somewhat, like if a rider tends to wrap up on a horse as soon as the race is lost then maybe the rider is more of an all or nothing type but then most of them will at least ride for second if they're already that close. Honestly I'm not sure if what you're looking at would matter enough to be worthwhile to your bottom line. My guess is no but don't take my word for it. When I'm looking at placing I'm usually looking more so at the horse and if it has that tendency, however as a horse-factor it's definitely much more overbet than it is underbet because it's so obvious whereas the rider would seem to be a factor which is more subtle when it comes to that. I think there could be riders getting second-rate mounts and they manage to get on the board quite a bit more than expected. Maybe they like to hang back and scrape paint things like that vs sending, you might find some imbalances in WPS there.

[classhandicapper]

Originally Posted by Dan H

Here's an idea that stems from my inability to find place and show horses with any consistency. I am hoping jockey stats (with other methods) can help me find them. I copied these stats from Brisnet PPs:

I never actually tested this with data, but finishing 2nd or 3rd is sometimes the result how the race develops. Sometimes the 2nd best horse gets used hard chasing the ultimate winner and a lesser horse will beat him for second.

So in races where you expect a lot of pace pressure, A over C can be better than A over B if the price is right and you expect B to be part of battle early and C to sit off it.

I'm not sure how to use the jockey factor in that, but maybe the more aggressive the rider is, the more likely he is to be all or nothing???

[MJC922]

Originally Posted by classhandicapper

I never actually tested this with data, but finishing 2nd or 3rd is sometimes the result how the race develops. Sometimes the 2nd best horse gets used hard chasing the ultimate winner and a lesser horse will beat him for second.

So in races where you expect a lot of pace pressure, A over C can be better than A over B if the price is right and you expect B to be part of battle early and C to sit off it.

I'm not sure how to use the jockey factor in that, but maybe the more aggressive the rider is, the more likely he is to be all or nothing???

Good point. Could actually be some validity to that. Back in the day there were never two riders more opposite style-wise at a track than these two (see attached). Dennis would pretty much send everything and Mike would take everything back to last and scrape paint the whole trip. I do see a trend in the WPS ratios between them fwiw.

[classhandicapper]

I did a regression on Stakes, Dirt, Route data this morning and the jockey factor came up mildly more important in routes than sprints.