One thing that will be clear at the end of our series this week on speed figures is that no two numbers are the same. Each of the figure makers we're exploring takes a different approach to rating race performance, and understanding the differences between them is critical to getting the best use out of whichever figure one uses.
In the first installment, we learned the Beyer Speed Figure boils down to three major components — final time, distance of the race and the inherent speed of the racetrack.
The TimeformUS formula provides handicappers with a different lens, based primarily on final time, pace and weight. Of those, pace is a key differentiating factor.
“We make pace figures thinking pace has influence on the final time, as much as the horses do,” said TimeformUS chief figure maker Craig Milkowski.
While pace plays a vital role in the TimeformUS model, one thing that doesn't is “par” numbers, the averages used by some figure makers to assess how fast or slow a specific race is normally run, based on class and distance.
“No pars is a big one for us,” said Milkowski. “We don't pigeonhole that a $10,000 maiden claimer should run this. We actually look at the horses and what we think they should run.”
Essentially, TimeformUS provides more of a projection figure than a past performance number. If a 3-year-old runs a figure of 80 in January and doesn't run back until July, Milkowski said, TimeformUS won't pin him with a figure of 80 in July. It'll be higher due to likely development.
Like all figure makers, TimeformUS uses a track variant to account for the many influences on race conditions at a given track on a given day. Milkowski said at one point in his career making figures, he tried to parse all the specific variables, like wind speed, track maintenance, time of day, etc; it proved impossible. Still, while some figure makers are satisfied just knowing track conditions changed, TimeformUS does try to understand what happened.
“We don't try to figure it out individually but we're aware of those things, you know weather changes, going from day to night, wind. If we have to look up wind, we know where to do that,” Milkowski said. “We don't take it lightly where we say the track was, for example, 20 points fast and then the last three races suddenly we're saying it's 10. We try to figure out why.”
In the scenario Milkowski describes, and other figure makers do this as well, TimeformUS might create a separate variant for the last three races, meaning horses that got a 100 rating in one of the first seven races would receive a final figure of 80, while a horse that ran a 100 in one of the last three races would get a 90.
Unlike some figure makers, who might only look at the winner or on-the-board runners in a race to determine if a change in variant is warranted, TimeformUS examines the top eight finishers in every race. For Milkowski, the bottom line is: never assume the track stayed the same all day.
“Having done this for a long time, you make a lot of bad figures if you make that assumption.”
What's the most common trap figure makers fall into, according to Milkowski?
“I think the most common is we underestimate how much winners improve when they win races,” he said. “Most people are using a projection method or a par or something like that. But so many horses — I think this is quoting the Beyer study from a long time ago — I think they found the average winner improved more than 10 points over their previous figures. When horses win, they're usually improving and often by a lot and the younger the horse the more that's true.
“So if you don't account for that, your figures are going to shrink over time.”
Milkowski is especially cognizant of the potential for huge figure jumps when dealing with 2- and 3-year-olds. In the first part of this series, we heard from Andy Beyer about adjusting “suspicious” figures after horses run back in subsequent starts to validate or disprove a previous number. We mentioned a specific example — juvenile Bolt d'Oro's win in last year's FrontRunner Stakes at Santa Anita. While Beyer downgraded his speed figure on Bolt because it reflected the best 2-year-old number in 25 years (an unlikely scenario from his perspective), TimeformUS actually left its figure alone.
“Just my personal experience is, most of the time, I trust the clock,” said Milkowski. “Now if it's an entire field and all of a sudden they get every horse in the field running new lifetime bests even when they finish seventh or eighth that's going to raise flags for me to say hey you know maybe something happened here.
“When (Bolt) ran as fast as the older males and faster than the fillies did that day, that was one where I was like here's a really good 2-year-old that just demolished the field, why am I going to knock that figure down?
“He didn't necessarily back that one up going forward, but that happens with 2-year-olds. They don't always run back; the rest of the field actually did.”
While Milkowski usually chooses to trust the clock, often it's his own timing of a race. Anyone who follows him on social media knows he has a major pet peeve with official times in many cases.
Sometimes, Mikowski said, the data is entered incorrectly into Equibase and if something seems fishy, he'll retime the race from video. In other situations, the time gets recorded properly in Equibase but there's still something wrong with it. Video usually provides the answer — for example, why the beam at the starting gate triggered the clock prematurely.
“With the beam system, it could be an outrider being in the wrong place tripping the beam,” Milkowski said. “I saw a race at Los Alamitos where the three-quarter mile time just looked way too fast. I always go and watch the replay and when I did, as the horses are getting close to (the beam) you see a huge flock of birds sitting on the rail just take off flying. I'm sure they tripped the beam before the horses got there.”
While birds can be a problem for the beam trigger, Trakus and GPS-based timing systems have their own issues.
“I'm not going to sit here and say they're a second off or anything like that but they're just not as accurate as what we're used to with the beam system,” said Milkowski. “And from what I understand they actually still rely on somebody pressing a button when the gate opens. You know he could easily be a half second off. And I have seen even worse than that and half a second is an eternity in horse racing.
“It'll throw figures off five or ten points, which sticks out like a sore thumb when you're trying to make a track variant.”
In the last week, it should be noted, three North American tracks officially installed a British company's GPS-based timing system. Woodbine, Laurel Park and Pimlico have been testing Total Performance Data for several months and the feedback Milkowski has gotten about it has been positive. The tracks seem to like it.
“We have been using GPS timing since the beginning of July and have been pleased with not only its performance, but the flexibility it offers,” Jonathan Zammit, vice president of Thoroughbred operations at Woodbine, told the Daily Racing Form.
Returning to the idea that pace is an important component in making TimeformUs figures, let's revisit a case we brought up in the first installment about Beyer Speed Figures. On Aug. 11 at Arlington Park, the 3-year-old Carrick won the Secretariat Stakes in a final time of 2:01.04. Two races later, at the same distance, getting a similar trip, 4-year-old Robert Bruce won the Arlington Million in a time more than a second slower than Carrick. The figures from some sources rated Carrick's performance higher. Like Beyer, TimeformUS gave Robert Bruce the better number (126 vs 118).
Milkowski concurs with Beyer that slow paces often muck up the final times of turf races.
“At TimeformU.S., I have the advantage that we incorporate pace,” Milkowski said. “On turf, late speed is a lot more important than final times. I've learned that final time is almost irrelevant on turf.
“Me and my partner, we're actually working on a new way to do it where we could show that the race in which Robert Bruce ran slower than Carrick on final time would still get a higher rating overall because Robert Bruce is the better horse, the faster horse at that time.
“He's an older horse so he should be, but we don't have it perfected yet because I got to tell you, it's a tough nut to crack.”
TimeformUS pace figures are now available with DRF Formulator. While early/late pace ratings show where a horse tends to be over multiple races, TimeformUS figures are provided for every race, revealing how fast each horse ran at different points of call.
New to the Paulick Report? Click here to sign up for our daily email newsletter to keep up on this and other stories happening in the Thoroughbred industry.
Copyright © 2019 Paulick Report.