The following piece is a personal submission from TIF Executive Director Patrick Cummings
Before tackling the state of timing in North American racing, and Equibase’s increasing role, I believe it is necessary to put some perspective on my background.
After seven years in the investment world, I pursued a full-time MBA, hoping to pivot into racing. Racing has been a life-long passion, and to that point, was a source of fulfilling part-time work for about a decade.
I engaged Trakus as part of my work on a business school project, which led to a consulting relationship and then a full-time role as director of racing information from November 2011 to June 2015. I was given tremendous support by the management team to study the Trakus data and devise ways in which it could be used to a greater degree.
I was recruited by the Hong Kong Jockey Club to lead their racing public affairs team, offered a three-year contract, and fulfilled that before being offered the chance to return home to launch and lead the Thoroughbred Idea Foundation.
Since leaving Trakus in 2015, I have not been compensated by them, directly or indirectly.
This piece is personal, and the tone is as well, in parts. I beg your indulgence.
I also highlight quite a few quotes from Equibase president Jason Wilson. I have interacted with Jason for nearly a decade, and I’ve always enjoyed our conversations and I think we get along quite well. My criticisms of Equibase are not personal, they are professional, but I believe they are necessary to protect the future of the sport, the accuracy of its data and the continued participation and confidence of horseplayers and horse buyers.
The state of race timing in America is not improving as the years pass. It is getting worse.
This degradation, which began as America wandered away from unarguably older, beam-based timing technology, has worsened since the introduction of GPS-based technology to time races at a small but ever-expanding number of North American tracks facilitated by a deal between Equibase and British-based firm Total Performance Data (TPD).
The TPD product, a GPS-based system known as Gmax, has not provided our sport with the accuracy that race timing once had.
For a sport which relies on the confidence of wagering customers to participate via betting, and which has seen betting decline sharply over the last 15 years (nearly 50% adjusted for inflation), a degrading of the quality of information provided to stakeholders is another suboptimal development.
In last week’s Jockey Club Round Table, Equibase president and chief operating officer Jason Wilson doubled down on GPS technology being a key focus going forward.
“…[We] have also recently pursued a new avenue for data collection through global satellite positioning, also known as GPS. GPS, of course, powers multi billion-dollar companies such as Uber and Lyft.
“We use this technology to provide services to tracks ranging from timing to enhanced video graphics. We are also looking into using GPS technology to collect more accurate workout information as well as provide additional data to engage fans. Similar technologies are being used overseas to support in-race wagering, which presents a new potential source of revenue stateside.
“In connection with our new focus on track services, Equibase purchased the timing assets of American Teletimer Corporation earlier this year, becoming the timer at 54 Thoroughbred and harness racing tracks. With this acquisition, we will be able to offer GPS services at these tracks without interruption as well as integrate photo finish and video graphics.
“These service integrations will be our primary focus over the next year as we look for ways to further streamline services for tracks.”
If the early returns on the accuracy of GPS-based timing at tracks in North America, including Del Mar, Golden Gate, Laurel, Mahoning Valley, Penn National, Pimlico and Woodbine, are any indicator, this is a very troubling development for our greater sport.
It makes perfect sense for Equibase to get into this space and increase their business. That isn’t at issue. As what amounts to a data monopoly which sells information to anyone affiliated with the sport, there is a need for that data to be accurate, and it is NOT.
Despite Equibase’s best efforts to lead people into believing so, particularly in remarks made in a recent TDN article, their professional customers are not wrong, nor are their complaining retail customers. Pursuing accuracy is not ignoble, and specifically within the profession of creating value-added figures designed to quantify individual performance.
“We’re proud that our speed figures have earned the confidence of Daily Racing Form readers over the last 28 years, and we can’t risk losing that credibility,” Andy Beyer said. “If we had to work with final times that aren’t precise, everything we do would be undermined.”
Andy Beyer is not wrong, and the Jockey Club, part owners of Equibase, should agree, as the Jockey Club has made it known that accurate figures are important to the valuation of Thoroughbreds.
A Jockey Club executive and Jockey Club member each sit on the North American International Cataloguing Standards Committee (NAICSC) which “establishes the requirement for non-Listed black type races in sales catalogues produce in North America.”
The “quality control requirements” implemented by the group included the creation of a minimum “Race Quality Score,” of which the inputs include “speed figures for the first four finishers in each race over three years. Four speed figures are used to compile the RQS and are provided by Bloodstock Research Information System (BRIS), Daily Racing Form (Beyer), Equibase and Thoro-Graph.”
Even the Jockey Club agrees that accuracy in figures of great importance for identifying the quality of performances in certain races which can significantly increase the value of Thoroughbreds at auction.
So why implement a clearly substandard technology whose accuracy is clearly lacking, producing inaccurate times over many months, and which is expected to expand significantly across dozens of tracks following the acquisition of American Teletimer?
It defies logic, and that is what frustrates long-time, passionate, dedicated, even revolutionary developers like Andy Beyer.
Attempting to qualify the decision to use GPS technology because of the success of GPS-powered entities like Uber and Lyft makes the decision look even more farcical - there is just no legitimate comparison between the two.
While drivers can be dispatched to collect and transport customers using GPS – a function available in almost every modern mobile device – it does not translate to accuracy in timing horse races. If you had to wager on the outcome of an accurate drop-off from an Uber ride, exactly at the spot you requested, my personal experience suggests you would likely be a frequent loser.
But there are no legal betting markets on outcomes related to Uber accuracy because the overall experience is good enough. It’s passable. As it relates to racing, accuracy in time is far more meaningful.
Intricate data used by customers to inform their billions in wagers on events where the difference between significant financial gain and total loss could be a centimeter. Reporting an opening quarter in :23.06 when it was really :21.85 is equal to reporting that distance as six lengths* slower than reality.
Six lengths* might not matter when it is close enough to the restaurant that your Lyft driver’s GPS thinks you are stopped near. In a horse race, six lengths* could be the difference between first and last. Because GPS works for Uber and Lyft should have no consequence for Equibase.
*Thanks to the horseplayers and figuremakers who highlighted my oversight that this time is more like 7 1/4 lengths. See, I told you so.
ACCURACY QUESTIONED BY EXPERTS
Modern “figuremakers” serve as natural checks and balances on the accuracy of race timing and if their many decades of expertise in creating standardized measures of pace and speed in races are any guide, the GPS-based times are routinely incorrect.
Randy Moss, best known in his roles on the NBC coverage of racing and his NFL Network reporting, is a long-time member of the team developing Beyer Speed Figures. Craig Milkowski spent years working on the figures that eventually became the basis for the TimeformUS product. Both have been vocal on social media over years about the degradation of accurate race timing, particularly since GPS timing via Equibase took hold. If two of the sport’s most experienced figuremakers are despondent about the state of race timing, then racing stakeholders are likely being hoodwinked.
If the times are inaccurate, then the public’s ability to compare performances is compromised. That’s bad for everyone in the sport.
The number of issues Moss and Milkowski have publicly noted as inaccurate in recent years are easily in the hundreds. The actual total number of errors is likely much higher.
GRADE 1 RACES IMPACTED BY ERRORS
Most recently, my attention was drawn to a tweet from August 3 that highlighted the final time of the Grade 1 Bing Crosby at Del Mar having been changed from its original publishing.
Del Mar began using the Gmax system this summer. A closer inspection revealed that four races from the day had different times in the Equibase charts. It seems the times have been changed back to their original versions, though experts suggest the original times are inaccurate.
Through a letter in my role with TIF, I alerted the California Horse Racing Board (CHRB) to concerns about the changes, particularly in a Grade 1 race, without any notification being made to the public. Other races on August 2 were changed too, but without public notice.
A CHRB spokesperson responded on behalf of the Board on Tuesday afternoon, stating in part that “the [CHRB] has been monitoring the timing issue at Del Mar and is aware of the problems that have led to the incorrect time…While we appreciate that racing fans want to see the correct splits during their viewing of the races, of primary importance is that the information is accurate for the public’s use for future handicapping purposes, and every effort is being made to ensure that.”
Prompted by the errors witnessed at Del Mar, TIF investigated timing on the same day at some other tracks using Equibase’s GPS technology. Just on August 1, eight races out of 11 at Woodbine had different times from their live feed compared to the published chart that existed on August 4, while six of nine races at Laurel are different too.
Now imagine this when Equibase gets the GPS technology in 50 different tracks!
If this GPS-based technology is working as designed, its accuracy is suspect. If accuracy mattered, a shift to a new method of timing would not have commenced until months of testing ensured confidence in such a switch.
Equibase tweeted an explanation to me, and others, for recent Del Mar timing issues, following my tweets from last weekend where it seems every race on turf there had hand times overriding the original times that were reported.
“Hand times were used over the weekend as Gmax and Del Mar were validating information with respect to start lines, rail settings and lanes.”
Well, hand times may have been used in the official charts, but if all parties knew this was ongoing, then there should have been no posting of inaccurate sectional times on the track’s feed.
Tracks with times that are routinely cited as being incorrect include: Canterbury, Del Mar, Golden Gate, Gulfstream, Indiana, Laurel, Louisiana Downs, Penn National, Santa Anita (select distances) and Woodbine.
Most are now using the GPS technology via Equibase, some use Trakus.
Coverage of the topic has generally evaded mainstream racing media, despite years of stories in the past about this topic. In Wilson’s remarks at the Round Table, he cited the commercial relationship between Equibase and the Daily Racing Form, which may serve as some notice as to why coverage of the timing inaccuracies at Equibase’s GPS installations seem limited in recent months.
Time matters. It has mattered in horse racing for as long as “accurate” times could be recorded. It could easily be argued that race timing in America was more accurate in the 1950s and 1960s than today.
Timing is a tricky business. A 2018 New Yorker story outlined the complexities of keeping time in Olympic events. Omega has been an official timer of the Olympics dating to 1932. Their measurement of time has evolved, improved – not surprising as the New Yorker suggests, given enhancements to technology. It’s notable that GPS-based systems are absent from the technologies used to improve timing.
AN EXAMPLE OF A GPS TIMING ERROR
Here is an example of an error with the GPS system, which the public might not otherwise notice, but which is easily identified by figuremakers, or anyone timing a race using video.
Races 6 and 8 at Penn National on August 12, 2020 were for maidens run over six furlongs on dirt. The sixth was for $16,000 open maiden claimers, while the eighth was a maiden special event for Pennsylvania-bred fillies and mares.
The times published for Race 6 were - :23.05, :47.04 (23.99), :59.80 (12.76), 1:13.17 (13.37).
The times published for Race 8 were - :22.07, :46.15 (24.08), :58.77 (12.68), 1:12.05 (13.28).
The final time suggests Race 6 was 1.12 seconds slower than Race 8, with the majority of the timing difference coming in the opening quarter mile.
But analyzing a replay of the race, from the break of the gate to arrival at the finish line, were almost identical at 1:14.50 (this time eliminates the run-up distance, an untimed portion at the start of the race). Visual observation confirms the gate was placed in essentially the same position.
Again, ignore the fractions – the horses were racing from the break of the gate to the finish in approximately 1:14.50. Yet, one race shows as 1:13.17 and another 1:12.05, a time which suggests Race 8 is about 5.5 lengths* faster than Race 6.
*Another figuremaker adjustment - this is much closer to 7 lengths.
For the figuremakers, without looking any further, Race 8 would likely get a much faster figure than Race 6. On Equibase’s own figures, that’s exactly what happened. The Race 8 winner – Sacred Sho Grl – received a 64 Equibase Speed Figure while Race 6 winner Coast to Coast received a 49.
But the Beyer Speed Figures for the two horses were nearly the same – Race 8 winner Sacred Sho Grl received a 45 while Race 6 winner Coast to Coast received a 46. TimeformUS gave both horses a 70 using their own methods.
While Beyer and Timeform seem to have been attentive to the inaccuracy, Equibase was not, and published a figure as if the two races were really 1.12 seconds different. A difference that wide normally equates to roughly 16 points on the Beyer scale over six furlongs. Based on their assessment, there is almost no difference between the speed of the two races.
Equibase explains their figures as such: “The Equibase Speed Figure tells you how fast a horse has been running in its past races with a single number. Its sophisticated algorithms are based on the horse's actual time in combination with other factors, such as the condition of the track.”
There does not seem to be much sophistication in play at all.
Inaccuracies such as these are replicated across different tracks, distances and surfaces. And with Equibase purchasing the beam-based timing assets of American Teletimer, presumably transitioning more towards GPS times such as those obtained by Gmax at Penn National, the future is anything but bright for those who seek accurate data to inform racing wagering decisions.
ERRORS AT TRAKUS TOO
Race timing has been a function which racetracks sort on their own, often hiring a third party with experience in the space to install equipment and produce the times for races.
Until the early 2010s, timing at tracks in North America was almost universally executed via beam-based technology from companies like American Teletimer, Teleview or a few others. Timing poles would be placed at the relevant points-of-call around the track, an invisible beam activates and when the first horse breaks the beam, the timer is tripped and reported.
While an older technology with some very notable limitations, it is tremendously reliable for timing the lead horse accurately. Occasional errors do occur, particularly if something gets in the way of the beam prematurely – a bird, an outrider, a photographer, etc.
Trakus has provided in-running graphics and post-race data for several North American tracks since 2006. As their business expanded, eventually Trakus times were used as official times. It happened at Gulfstream Park, then Tampa Bay Downs and eventually Santa Anita and Del Mar. Trakus was never devised as a sectional timing system, but it did produce sectional times as part of their total package and after a half-decade at different tracks, was pressed into service, likely because of some cost savings it would bring tracks and potentially offer some flexibility in expanding turf courses.
It is noteworthy that the Trakus technology is not GPS-based, but rather uses “proprietary wireless communications” which it deems “more accurate and immediate than GPS.”
Equibase, and their on-track chart callers, worked with and operated the Trakus system. But in recent years, Equibase made a commercial decision to get into this business for itself.
There have been some rather clear errors with Trakus official times over many years at some specific tracks and distances. Races at one mile on dirt at Gulfstream are particularly troublesome.
While they produce very useful in-race graphics and post-race information, it does not seem as though Trakus is the most accurate solution for in-race timing, either. And that’s the thing – Gmax could produce some other very meaningful benefits too – stride-length data, and individual horse times for those that aren’t leading the race.
In the TDN story from Tuesday, Wilson comments that “timing races is just a very small part of the GPS offering.”
But timing needs to be accurate for wagering customers to make confident, informed decisions and for breeders to identify the best performances.
As for the chart of the race, which is supposed to report the actual times, which helps actual betting customers, actual horse buyers and actual breeders identify the faster horse? If we lose accuracy there, we lose a lot. To hear from the official data provider that timing is just a small part of it all, when accuracy there is probably most meaningful, is monumentally disheartening.
MOVING FORWARD, ACCURATELY
An accurate time is a fundamental element of regulated horse races.
It has become clear that our sport has not evolved with more modern technology, but rather taken a technology, ignored whether it is at least as accurate as the technology it is replacing, and shoved a square peg into a round hole.
Questioning Equibase’s GPS play is not being critical of all innovation and hoping to quash it, it is being critical of technological backpedaling which is being positioned as exactly the opposite.
Bettors invested $11 billion in 2019 to wager on Thoroughbred races in America. Owners spent more than $1 billion at horse sales in 2019 buying Thoroughbreds here. Our sport can do much better, and the status quo is only worsened by Equibase’s role as the sole data-keeper for the sport. Now, they are financially tied to ushering-in a seemingly less accurate method of race timing.
Here are a few recommendations or considerations for the future:
1. Do not publish times on track feeds or in post-race charts if you are not confident in their accuracy. Employ experts to confirm accuracy for you, because it appears those who are in place to confirm accuracy are not enough in this particular capacity. The status quo is misleading viewers, but just how misleading is almost incalculable given the different types of stakeholders consuming bad information.
2. As errors are identified, Equibase should create a publicly available log of all adjustments to times or any other errors instead of just changing charts. Clearly identifying errors, no matter how many, is better than updating them and, effectively, hiding it from the public. Compare it to baseball’s log of scoring decision changes from 2019.
3. No matter how the technology evolves, Equibase should consider working with tracks to time races from the break of the gate, eliminating untimed run-up distances. This will also yield greater accuracy in reporting the exact distances at which races are run. Andy Beyer lobbied for this years ago.
Not to confuse matters, but some graded stakes races (and plenty of non-graded ones too) are being run at distances meaningfully longer than they are reported. If you aren’t familiar with the problems surrounding run-up distances in North American, this piece I penned in 2014 covers it in extreme detail.
4. Since Equibase now owns the timing assets of American Teletimer, at least consider finding a way to integrate the much more accurate beam-based times with the other functions of GPS timing until better alternatives are derived.
5. Carefully consider the long-term applications of technology for tracking morning workouts. At the Round Table, Equibase’s Jason Wilson reported: “We are also looking into using GPS technology to collect more accurate workout information.”
If the Gmax system is struggling to accurately record times in highly controlled environments like a race, its use in tracking workouts, with hundreds of horses starting and stopping all over the track, is highly suspect.
Only one jurisdiction on the planet – Japan – is successfully tracking morning workouts and reporting the information publicly, doing so over a limited number of distances and courses. And, notably, they do not using GPS-based technology.
I am all for innovation in horse racing, and I know that many others would happily embrace accurate technological innovations, properly vetted, to help the sport.
GPS timing, at present, is not one of them, and masquerading it as some antidote for existing timing exacerbates the harm it is already doing.
Transparent, straight-forward approaches should be employed. An honest dialogue with customers is needed. It has become clear Equibase did not get this launch of GPS timing right. Own it and move forward. No one is coming in and displacing Equibase as the sport’s official data provider. We need you, but we need you to be better!
It will make all of racing better too.
Postscript: A new commercial deal
On Tuesday morning, I was alerted to a story from betting industry trade news site SBCAmericas on a deal between GMax operators Total Performance Data and XB Net, a company owned by The Stronach Group, which distributes race signals to international customers for pari-mutuel and fixed-odds betting on American races, including “in-running” wagering.
Remarkably, the deal highlights the data and GPS tracking as fundamental to this offering for horseplayers who can access markets on American racing which are not available to American players. That may actually be a blessing in disguise.
“Thanks to Equibase’s GPS rollout, the pair are launching an “in-running” industry-first for North American horse racing. This product undertakes to revolutionize live betting by delivering reliable pricing on a popular sport which has traditionally struggled to maintain pace with a broader sportsbook marketplace, where over 70% of turnover is generated in-play.
“Now, however, TPD and XB Net have combined to produce what they describe as a dependable in-play offering that crystallizes accurate odds for the most solid in-play, fixed-odds pricing available.
“The feed fuses the PA starting price with the best of automated trading via TPD’s range of consequential in-running analytics, including stride length, stride frequency and sectional timings enabled by saddle-cloth GPS tracking.”
Later in the piece, the following quote appears from TPD boss William Duff Gordon.
“We’re thrilled to partner with XB Net in the States to provide the only fully automated in-play solution on the market. As with any live sport, you must have a service that keeps up in-play, harnessing the most accurate and lowest-latency data feeds to secure genuine market innovation. Our prices process the key variables in real-time, and leverage neural networks to create the best possible in-play price.
“XB Net’s unparalleled racetrack coverage in North America – coupled to their strong heritage in managing content rights, data and odds – meant that they were the logical partners in a shared vision that challenges an outmoded mindset around horse racing trading. Deeper audience engagement is simply our combined aim. So, we’re confident this in-play solution will bring a welcome new dimension to anyone following racing stateside.”
It will certainly be a new dimension.