Slides to accompany the Performance Prediction for Thoroughbred Racehorses by Jeff Seder, Founder & CEO of EQB.com
This presentation was delivered as the opening keynote at the inaugural HorseTech Conference on the 18th October 2017 hosted by the Royal Veterinary College London.
2. The Relationship of Selected
Two-dimensional Echocardiographic
Measurements to the Subsequent
Racing Performance of
5,431 Yearlings and 2,003 Two-year-
old Thoroughbred Racehorses
[ excerpts from 72 page refereed published
study – JEVS Vol 23, No 5, May 2003]
3. .
• And musings on the state of
technology in racehorse
performance prediction –
{in a Philadelphia, USA accent--- English and Americans as two peoples
divided by a common language}
4. First: who we are – what we do
35 years experience/research --First
portable ultrasound echocardiography
5. Traditional plus Tech – Predicting performance.
Real results at real racetracks through research married to
multiple tech approaches, traditional expertise, and hard work.
6. www.eqb.com
• 40 years experience as full service bloodstock agents for
Thoroughbred racehorses.
• Millions of dollars in multidisciplinary sports medicine
research with major universities on real racehorses at major
racetracks: biomechanics, biometrics, locomotion (gait) and
exercise physiology. BIG DATA ANALYTICS.
• Spun off medical devices to J&J; holder of patents; inventor of
medical devices for humans and horses.
9. Data available to YOU
• Our studies typically involve thousands of
major racehorses at major racetracks over
multiple years.
• Download papers with lots of data
from our web site; or get a copy
from me here at this conference
www.eqb.com
10. All the many many other avenues we
pursued… ,
.
following to paddock and measuring poop
weights, etc . Led to 1986 new medical of
the year Europe a-mode ultrasound bone
scanner, heart holter monitors, aerodynamic
silks, biomechanics, logarithmic velocity
decay fatigue curves, etc.
11. Examined other techs out there and
their methodologies: e.g.--------
.
• Cow bones
• Morph !
• Stride lengths for each workout distance
• Conformation static vs dynamic, and the problem of parallax
• The “health gene”
• Herd dynamics in races
• The follies of single variables – the magic feed; the magic
horseshoe; etc. (e.g. tobasco sauce)
12. The path to heart scanning --
• The heart paths started with listening to
hearts, and looking at chest sizes.
• Steele’s “Heart Score” from the EKG
• Legend of Secretariat’s heart size (Eclipse?)
• Our Apple llc hybrid monster and new exam
protocols solved the problem of GETTING THE
DATA – BUT then ……
• Years of failure without proper data base.
13. Our researchers and methods
• Harvard (locomotion), MIT (enginjeering),
Univ. of Delaware (biomechanics), Rutgers
(statistics).
• Univ. of Pennsylvania Veterinary School at
New Bolton Center (cardiologists)
• U.S. Olympic Sports Medicine Committee
(Biomechanists).
• Refereed scientific journal publications
• Multivariant discriminant analyses
• 10 years, 7,433 subjects
14. First, before going to our heart scan studies in detail, -- an
unexpected problem: defining “Performance”
How much data did we really need?
Enough to capture enough elites !!
• Averages for the breed/Worldwide performances
• of named Thoroughbred foals born in
• North America between 1985-199410
• Subset of Foals of Foals by top
• population 1985-1994 1% of sires
• % Starters/foals 68.9 84.8
• % Stakes winners/foals 3.2 9.1
• % Graded stakes winners/foals 0.7 3.6
• % Grade 1 stakes winners/foals 0.2 1.2
• Top 1% of sires determined by total progeny earnings for 1985 through 1994.
15.
16. Some fenagle factors in defining
“performance”
Almost half never race,
(but WHICH half?) AND
Only 0.2 % OF RUNNERS
EARN > $200,000 IN A
GIVEN YEAR
17. Even More Of The Unknown and
Unknowable:
• No one variable can tell the whole story.
• And they interact mightily with each other.
• And, there are WILD CARD unpredictable but key
VARIABLES FOR PERFORMANCE: e.g.
INJURY AND ATTITUDE
e.g., Approx 25% injured / yr enough to
compromise performance
18. IS this good enough to define performance ???.
• .
Top performing Total
Bloodstock Agent’s No. Bought # SW %SW***
Name
Mike Ryan 332 32 9.6%
BBA (England) 325 28 8.6%
Shadwell Estate Co. Ltd. 235 25 10.6%
Darley Stud Management 218 33 15.1%
D. Wayne Lukas 216 28 13.0%
Charles P. Gordon Watson‑ 193 26 13.5%
Horse France 188 19 10.1%
John Ferguson Bloodstock 160 15 9.4%
Frank Stronach 145 27 18.6%
Dogwood Stable 140 8 5.7%
Narvick International 137 6 4.4%
James S. Delahooke 136 10 7.4%
Baden P. (Buzz) Chace 130 11 8.5%
Eugene N. Melnyk 107 10 9.4%
Demi O'Byrne 106 26 24.5%
Hartley/De Renzo 105 3 2.9%
*Based on statistics published by July 5, 2003 Thoroughbred Times magazine
**For buyers of 100+ yearlings between 1990 2000 spending an average of $30,000+ per yearling‑
19. .
• WHAT ABOUT OUR TRACK RECORD ??
–Was the research helpful ??
Buyer of multiple world champions, and over
40 grade one stakes winners as unraced young
horses, through multidisciplinary multimodal
approach
20.
21. Oops – needs some updates –
AMERICAN
PHAROAH
• New York Times front page center
June 5, 2015
22. Statistics 101
As my Dad said, “5
flukes are a trend.”
EQB’s KY Derby record- helping
clients “get there”-
[underlined we bought at auction as unraced
youngsters, the
others we consulted on in culling/buying
decisions]
2009 Pioneerof The Nile (2nd,
earned $1,634,200)
2010 Eskendereya (favorite week
before the race, earned $725,200))
2011 Nehro (2nd, earned
$978,381))
2014 Vicar’s In Trouble ( injured,
eased in Derby - earned
$788,454) and
We Miss Artie (mid pack,
earned $783,843)
2016 – Adventist
Oscar Nominated
23. .
• 2nd in KY Derby 2009 & 2011,
[Nehro & Pioneer of the Nile]
and favorite in 2010 {Eskendereya}
until injury 4 days before race]
BACK TO BACK
BREEDERS CUP &
Eclipse Award
WORLD CHAMPS 2008 & 2009
bought as unraced 2 year olds by EQB
[Informed Decision and Forever Together ]
25. .
• BACK TO BACK
LEADING USA OWNER
AWARDS 2008 & 2009
with EQB purchased horses
and
SIX separate horses were
GRADE ONE STAKES WINNERS in 2010
(Eskendereya, who was the heavy favorite for
KY Derby until injury 5 days before race,
Rightly So, General Quarters,
Slip Away, A Z Warrior, Jaycito)
and
Overall 2010 saw us with 20 graded stakes
horses, 29 stakes wins, and 40 stakes horses.
and
26. Back to the heart scan research --
METHODOLOGY
• Multi-disciplinary university based team:
veterinarians; statisticians; MDs;
biomechanists; trainers; jockeys -
in creation, execution and interpretation
of data. Harvard, Univ of Penna,
Univ of Delaware, Rutgers Univ.
• Variety of statistical tests done to show NOT
ONLY what variables “worked”, but also to
eliminate other possible causes
27. This heart scanning study’s
composition versus reality –
(7,433 subjects from major auctions)
• Sale to racetrack performance of 1990-1999 graduates of
major yearling sales –
# % stks % graded
winners Stks winners
• Keeneland Sept 28,176 6.4% 2.0%
• This Study 7,434 6.25% 2.9%
31. Instant violinists?
[ Malcolm Galdwell’s 10,000 hours rule ]
• Not using standard cardiac views
• To get that reproducibility in this cardiac scan
mode, and in height/weight data, took years of
practice for the technicians, and blind test
corroboration studies.
• Small transducer scan angle mistakes mean big
data errors.
• Having the instruction manual and the violin
doesn’t mean you’ll play a symphony.---
35. Normalized Data
[ by percentile within sex/ht/wt group ]
NOTE: effect of growth spurt of puberty
19-22 months of age
36. Normalized data: size matters
[sadly simple truth]
so need to adjust for this to prove anything about the effect of
other variables like the heart.
46. • Do I have time for some cautionary tales
about innovation and technology in horse
racing -- war stories ???
• Use common sense:
– Don’t fall in love with your single variable.
– Don’t believe all of what’s called science.
– Ask for the research, credentials, methodology,
etc.