In a battle between Brisnet Prime Power Rating’s highest ranked horse, Fly (133.25 BPPR) at 3-5 in the 8th race at Gulfstream today, and the HW’s top selection, #6 Ms. Meshak (122.31) at 11-1, the Handicap Wizard proved to be the more accurate model (exponential model), as Meshak won in a romp. Ms. Meshak had a 45% combined winning probability and 68% winning estimate by speed, making her an easy choice to win. She did, paying $25.80.
The Handicap Wizard had winners in five out of nine non-maiden races at Keeneland on Saturday, including the G1 Blue Grass Stakes with Vekoma. This was with the soon to be released exponential model in Version 4.0 with weightings of 0.5% for SPD (final speed winning est.) and WP (winning percentage, adjusted for stakes). The weightings reflect the configurations that were used in the algorithm for the original version of the Handicap Wizard. Version 4.0 will have dual functionality between the model with exponential smoothing and the standard model, which uses equal weighting for all past speed figures in calculating means and standard deviations for simulation. The benefit of exponential smoothing is that, theoretically, it captures the form cycles of the horses by weighing the most recent speed figures higher than latter ones in calculating means and standard deviations. The following results underscore the success of that approach.
However, this does not mean that considering the component speed figures, too, in a race isn’t worthwhile. On the contrary, had one keyed on the LP estimate of #8 Lantiz in the 5th race with a monster 45.8% value in that 1 1/8 mi. race on the turf, one would have found a winner that paid $94.40. Every race is a puzzle with its own set of dynamics, where the results of the main algorithm as well as the component estimates go into solving that puzzle.
Another example was in the 6th race, the G3 Commonwealth run at 7f on dirt for four-year-olds, with #11 Bobby’s Wicked One. At first glance, the horse was rated mid-pack by the Handicap Wizard with a combined winning probability of 7.53%, which was close to how the market was assessing him with odds of 13-1. Yet, despite having a low winning percentage of just 4.7%, the horse had final speed winning estimate of 10.4% and a TRIO winning estimate of 20.0%. The TRIO estimate (found in HW Version 4.0) is a simulation of E1, E2 and LP figures of a horse added together. In this particular case, Bobby’s Wicked One’s components together in simulation stacked up well against those of the other horses, where he rated third below the two favorites, #2 Recruiting Ready and #4 Uncontested. The 20.0% TRIO estimate would equate to fair value odds of 4-1, much less than the 13-1 price the horse was getting in the live odds. You would never see the value of a horse like Bobby’s Wicked One without the simulation engine of the Handicap Wizard revealing such data.
Racing fans will be watching three prep races today in the Wood Memorial, the Blue Grass Stakes and the Santa Anita Derby for clues about who will be running at Churchill on the first Saturday in May. The way the Handicap Wizard’s projections will be calculated today for these races is using exponential smoothing, a feature which will be incorporated into the program’s next version, version 4.0, to be released in the near future. Exponential smoothing puts greater weight on the most recent speed figures in calculating means and standard deviations for the simulation. The value of the approach can be seen in its winning speed estimates from last week’s Florida Derby, which saw Maximum Security go-to-wire and maiden, Bodexpress, finish second. The improbable outcome resulted in the $1 exacta paying $257. The winning speed estimates using exponential smoothing had both horses in its top four as the third and fourth selections. Why the horses did not appear higher in the combined winning probabilities is because neither horse had won or even competed in a stakes race before the Florida Derby, causing their winning percentages, adjusted for stakes, to be lower.
Florida Derby
Wood Memorial
Blue Grass Stakes
Santa Anita Derby
The next race on the Derby trail is this Saturday’s $1 million Florida Derby at Gulfstream Park, run at 1 1/8 miles on dirt for three-year olds. The contest features three stakes winners in Code of Honor, Harvey Wallbanger and Garter and Tie. However, that is not fully indicative of the talent in this race, which includes the highly touted, Hidden Scroll, from the Bill Mott barn, who wowed in winning his maiden race at Gulfstream at $60K stakes by 14 lengths at the end of January, when he earned a Brisnet final speed figure of 102. Code of Honor won the G2 Fountain of Youth Stakes at 1 1/16mi. at Gulfstream in early March, holding off a surging Bourbon War at the wire. Harvey Wallbanger defied long odds of 25-1 to win the G2 Holy Bull at 1 1/16 mi. at Gulfstream in February, benefiting from an excellent rail trip that allowed him to overtake Everfast down the stretch, who finished second. Garter and Tie boasts the highest final speed figure in the field at 103, which he received in winning the $75K Smooth Air Stakes at 1 mi. at Gulfstream in December. However, he has had less success in his last two starts, with a 3rd place finish in the $100K Mucho Macho Man Stakes at Gulfstream in January and a 6th place finish in the Holy Bull.
After changing the weights to 0.15% for E2, to 0.15% for LP and to 0% for E1 to account for the longer distance in this race, the Handicap Wizard is calling the race close between Hidden Scroll and Garter and Tie as favorites. However, because the program calculates its winning speed estimates from simulation that does not factor in the regression both horses have had in their last starts, the combined probabilities of 25% of these horses should be viewed with caution. In the case of Hidden Scroll, the horse is seeing a jockey change to Javier Castellano from Joel Rosario, who arguably ran Hidden Scroll too hard in the opening fractions of the Fountain of Youth, causing the horse to come up short in the end. Castellano won the Florida Derby in 2014 with Constitution. The Handicap Wizard’s third selection, Code of Honor, looks to be a solid choice following his win in the Fountain of Youth Stakes. He is the class in the race in terms of money finishes and has the lowest risk in the race. Still, he will need to be in top form, if he is going to fend off Bourbon War in the final eighth, who has the strongest LP estimate in the field. The Handicap Wizard rates Bourbon War fourth, but he appears to be the best suited for the 1 1/8 mi. distance of this race.
Another horse that will be charging late is Harvey Wallbanger, who has shown good progression in his five races, posting his best final (95), E2 (100)and LP speed figures (100) in his last start, according to Brisnet. Therefore, the model might be underestimating his potential in this race by rating him as a 6th selection. His winning percentage, adjusted for stakes, of 15.7% as second highest is likely more indicative of the relative strength of this horse. In an interview with America’s Best Racing, co-owner, Wayne Cutler, praised his horse as one having heart and knowing where the finish line is, while running to the competition. It’s also worth noting that the horse worked 4 furlongs on March 23. In 47.1, finishing 1st of 99 horses.
The last horse worth mentioning is lightly run, Maximum Security, who comes into the race undefeated in three starts at Gulfstream, albeit against allowance competition. It will be the first time he stretches out to two turns under rider, Luis Saez, who is second only to Irad Ortiz Jr. in the Gulfstream jockey standings this year.
The Handicap Wizard had three of the top four finishers correct in the Louisiana Derby with its top four selections. Its first selection, War of Will, suffered a hing leg injury coming out of the gate and was not a factor in the race. The horse that won, By My Standards, defied long odds to beat the Pletcher trained, Spinoff, at the wire.
To put By My Standards performance in perspective mathematically, his mean final speed figure before the Louisiana Derby was 84.75 in four starts with a standard deviation of 4.27. In the Louisiana Derby he put up a 102 figure, which was 4.04 standard deviations above the mean. The equivalent probability is 0.0027% or 1 in 37,417.
If we go off his last fig of 90, the jump to a 102 final speed figure is still 2.81 standard deviations higher. The probability of that happening from 90 as a mean using the same standard deviation is 0.2477% or 1 in 404.
Pro-Handicap Analytics 