The Truth About Speed Figures

THE TRUTH ABOUT SPEED FIGURES

by Mark Cramer

Cultures pass down myths from one generation to another. In our horse racing subculture, today’s greatest myth has developed into an icon: the speed figure. We horseplayers have been conditioned to assume that the horse that has run the fastest recently is most likely to run the fastest today. When a horse with a lower speed figure than our most-likely winner comes home first, we are educated to look at the event as an anomaly. The exceptions accumulate in our psychic landscape like chain stores and yet we fail to admit that a pattern, one that contradicts our cherished myth, has developed. My own impressionistic evidence has told me time after time that the fastest horse from the last race of the past performances does not usually win today, and when it does, the payoff is scanty. But when we accumulate anecdotal evidence, we risk deluding ourselves hazy impressionism? When the lines of logic seem blurry, it’s time to make objective measurements.

I decided to do some rigorous research, constructing the parameters in a way that gives the Devil’s Advocate every chance to prove I’m dead wrong.

The theory

Horses that can win races are the ones that can significantly IMPROVE their previous race speed figure. Today’s winner is not the horse with the highest figure from its last race but the horse that is most likely to REACH its highest figure today. Bold-face Beyer figures function essentially as mirages, optical illusions that distort racing reality. Yes, they are more than reasonably accurate most of the time. But they are not worth their face value, for an accurate rendering of the past is not the same thing as an objective prediction of the future. Better stated, the past performances are something that should be seen dynamically, as if they were part of a moving process

Methodology

We’re using the Beyer figures as a research tool, but if the theory is true and winners are horses that run significantly faster today than they did in their previous race, then any other good speed figures would give us the same research outcome. So what we’re really saying is not that Beyer figs are misleading but that accurately adjusted final times from the horses’ most recent race are not good indicators of today’s winners.

(1) I decided to study a group of 150 winners, randomly collected. I would write down the winning Beyer figure and then compare it to the horse’s previous race fig. A plus 18 would mean the horse’s winning race was 18 points faster in its Beyer rating. A minus 7 would mean the horse’s winning race was 7 points slower in its Beyer figure.

When I added all the plus numbers and then the minus numbers, I would be able to come up with a magic number: THE AVERAGE NUMBER OF BEYER POINTS THAT WINNING HORSES INCREASE OR DECREASE FROM THEIR MOST RECENT RACE TO TODAY’S RACE.

My hypothesis expected a significant average increase in Beyer fig because winning horses are the ones that get better and not the ones that were the best. The lower the average increase in Beyer figure for winning horses, the less validity my theory would have.

(2) Beyer advocates still argue that these figures are transferable from different surfaces and distances. But hey, a horse that loves to route and hates to sprint is going to improve its figure when going from sprint to route. Likewise, a horse that loves a fast track and hates the slop, is going to improve from the latter to the former and regress when returning to the slop. In order to eliminate every possible distorting factor in this research, I decided to ONLY CONSIDER LIKE DISTANCES AND SURFACES. This meant:

(a) excluding all dirt-to-turf or turf-to-dirt races, all sprint-to-routes or route-to-sprints, and all fast-to-wet or wet-to-fast changes from my research sample; and

(b) excluding all changes from or to unusual distances, such as 7 1/2 furlongs, 1 3 1/16 or 1 3/8, for which not enough races are run for accurate processing of speed figures; and

(c) allowing for no more than a one-furlong distance change in sprint races and no more than a one-and-a-half furlong change in route races. This would mean that a race involving a switch from 5 1/2 furlongs to 7 furlongs, or one from 1 1/8 to a mile would not be part of the research sample.

This represents a can’t-go-wrong research methodology, since I was precisely eliminating the types of changes that could have made my theory come out better than it deserved to. I would still accumulate the 150-race sample I expected to need.

(3) If after a 150-race sample, the outcome was at all inconclusive, I would do follow-up 150-race samples until a meaningful and conclusive pattern developed.

(4) I would also calculate a ratio between number of winning horses with improved Beyer figures in their winning race compared to the number with 0 improvement or a decrease in Beyer fig.

The research: Sample I

What turned out to be the first sample of 150 races had the following outcomes:

Improvement ratio. For every winning horse whose Beyer fig decreased or stayed the same from its previous race to its winning race, there were four horses whose Beyer figure got better with their winning race. This was a 4:1 ratio in favor of my improvement theory.

Raw average improvement (including winners with higher Beyers and those with lower Beyers): 9.7

In other words, the average thoroughbred winner improves it’s Beyer fig by 9.7!

That seem far too high to me, so I decided to do another sample.

Sample II

With totally different races, the results were as follows:

Improvement ratio. Once again it was 4:1. Four winning horses improved their Beyer figure for every one winning horse that didn’t. It looked as if I’d stumbled on some sort of statistical constant, for the 4:1 ratio now extended over a sample of 300 races.

Raw average improvement. Instead of the 9.7 Beyer points from the previous sample, I now had a 7.93 average improvement. The two averages were close enough to each other to make sense.

Combined 300-race average improvement

The 300-race sample produced an unadjusted average improvement in Beyer figure from most recent race to winning race of 8.8 Beyer points, rounded off.

What does this mean?

With winners’ average Beyer fig jumping an average of 8.8, the horse with the highest Beyer in its previous race has virtually no advantage and can not be called a most-likely winner, unless it figures to improve. In fact, on figures alone, more than half of a typical field has a chance to win. The handicapping question becomes not “Who’s the fastest horse?” but “Which horse will improve the most today?”

Given these research results, for most winners it’s no longer possible to declare that “a horse didn’t figure!”, based on speed figs. In fact, any horse in the field within 8 points of the highest figure has more than a theoretical chance to win, and according to my research results, if that horse is within 15 points of the highest Beyer in the race, it had a reasonable chance to win on speed alone, if we exclude other factors.

Needless to say, other factors like class, form cycle, pace, and excuses in previous races must be considered.

First conclusion

We already knew that the “best-Beyer-previous-race” factor was overbet by the crowd. Now we know that best Beyer last race is not even an adequate predictor of winning horses. In other words, when any horse wins a race, most probably there was one or more other horses in the same field that had higher last-race Beyer figures.

None of this says that speed figures are inaccurate. It only suggests that handicappers must be able to project improvement in order to pick out contenders and winners.

What factors will most likely point to a speed improvement?

Categories that emerged as significant are outlined here.

(1) Three-year-old coming back to races after having been lightly raced as a two year old.

You don’t find a lot of these, but this may be the most consistently accurate projection for a radical increase in Beyer figure. The logic behind the improvement is impeccable. Two-year olds simply run more slowly than three-year olds. But in particular, horses that were lightly-raced as twos, are more likely to quickly reach a projected three-year-old speed rating.

(2) Class droppers. Numerically, the greatest number of winners that showed meaningful improvements in Beyer numbers from their previous race were class droppers.

(This researcher has done extensive research on class drops which has let to a very reasonable theory based on the herding instinct principle. Horses dropping to their proper level can be expected to run faster, for when they are part of a herd of their equals or inferiors, they are more naturally inclined to run in a competitive way. Horses that are outclassed, on the other hand, can be projected to earn a lower Beyer figure than their potential, for once they feel intimidated by more dominant members of the herd, they surrender their fighting spirit.)

In my unpleasant observations of fighting cocks in Panama, I learned this lesson in a visceral way. A rooster entering the ring that “felt” outclassed, began to run away from his rival. Did this mean he couldn’t fight? Not at all. It meant he felt outclassed. The same rooster, when pitted against a member of its own class was capable of fighting to the death.

Horses that suddenly find themselves racing at their proper level, can be projected to “wake up”. Of course, racing is replete with contradictions, and a class dropper that is “damaged goods” may fail to achieve this projection.

(3) Another subset of improvers were the younger class risers, especially those lightly-raced young horses moving up through allowance conditions. These were far fewer in numbers than the class droppers, but at least could be identified, not only by this researcher but by the betting public. Tough to find an overlay in this category.

(4) Horses with no reason to improve. We’d be remiss if we left out this category. Especially in low- and middle-level claiming races, in-and-outers seem to populate most fields. The apparent sudden improvement in these horses may merely be a return to a previous peak that was reached three, four or five races ago. At these class levels, horses are quite susceptible to “the shape of the race”. Their finding a “spot” in which they can race with less stress may suddenly allow them to improve their Beyer figure in a signficant way.

(5) Horses returning after a layoff are prime candidates for radical improvement over their pre-layoff race, provided they have trainers whose horses fire fresh. The final race prior to a layoff is probably an unhappy occasion in which something happens to the horse, thus giving him an excuse for a vacation from the races. Having spent so much time living amongst trainers and talking with them about their horses, I can assure you that two opposite philosophies co-exist on the backside. The type-A trainer is looking for an immediate result and prepares his layoff horse for a peak comeback effort. Type-B trainers prefer to race their horses back up to a peak.

(6) Maiden droppers.

The maiden drop is one of the most potent improvement factors in racing. I clipped 15 maiden dropper winners and searched for something they had in common.

The surprise of this group of 15 was that, with the drop, 14 of them earned a career-high Beyer rating. Another common denominator of these maiden dropper winners is that they’d all shown some competitive running lines in at least some of their previous races. (The type of maiden that follows a field without making a move, as if he were on a merry-go-round, is not likely to satisfy the investor in maiden droppers.)

Final conclusion

Handicappers whose fundamental activity is comparing speed figures and betting the horse with the highest figure will be facing underlay odds and confronting a powerful probability that last race’s speed figures will be altered in today’s race. Handicapping involves the skill of making projections for the future, not calculating an unbending past. There are non-linear curves and bends in the progression of time that make handicapping a geometric rather than arithmetic activity.

If you have no handicapping reason to project an improvement in your horse’s Beyer figure for today’s race, you’re probably looking at a loser.

Handicappers whose fundamental activity is comparing speed figures and betting the horse with the highest figure will be facing underlay odds and confronting a powerful probability that last race’s speed figures will be altered in today’s race. Handicapping involves the skill of making projections for the future, not calculating an unbending past. There are non-linear curves and bends in the progression of time that make handicapping a geometric rather than arithmetic activity.

 

good article, and if you take nothing but this from it, you'll come away with learnign something.

very good read

Good stuff Hound.......

Good read.

Superb article. Absolutely spot on. Think speed figures generally helps to paint a picture of a horse's ability. In a big race like the Derby, surely the goal is to get the horse to peak on the day, not bottom him out in preps running fast times and winning far.

Great food for thought and cheers for that.