Secretariat Didn’t Bounce — and Your Horse Probably Won’t Either

Myths are part of what makes sports fun.

The “Curse of the Bambino” supposedly doomed the Red Sox for 86 years after they traded Babe Ruth. Michael Jordan allegedly never lost a game when wearing UNC shorts under his Bulls uniform. And don’t even get me started on the “Curse of Apollo” or the “Madden curse.”

But while most sports myths make me smile, one makes me want to weep, gnash my teeth, and kick anybody who uses cliched phrases like “at the end of the day” in their “low-hanging fruit.”

Of course, I’m talking about the Bounce Theory.

For the uninitiated, the Bounce Theory suggests that when a horse records a superior speed figure, it's likely to regress sharply next time out. Why? According to the theory’s disciples, because the horse is too "spent" from its last effort — not tactically overmatched, not off-form, just exhausted.

This sounds reasonable (if we ignore the fact that most races are under two minutes long), even humane. But, generally, it’s just gobbledygook wrapped in balderdash with poppycock and a warm glass of hogwash on the side. Yet, sadly, this inane theory has influenced how many trainers campaign their horses — keeping them in bubble wrap for fear of a mythical letdown, all while bettors dump perfectly logical contenders for fear they “left it all on the track last time.”

The Bounce Theory didn’t start in a shady backroom or Reddit thread. It was born from numbers in The Sheets and Thoro-Graph — sophisticated speed figures produced by Len Ragozin and Jerry Brown, respectively.

Once a skeptic, Andy Beyer, creator of the famous Beyer Speed Figures that appear in the Daily Racing Form, became a staunch advocate in 1993 while writing Beyer on Speed.

“When the Racing Form undertook its study of the Beyer Speed Figures, I had a chance to examine some patterns and determine if they had any bearing on a horse’s future performance,” Beyer wrote in that 1993 classic. “I asked the computer what happens after a horse has run a sequence of three improving figures or three declining figures on the dirt during a four-month period. If, for example, he earns figures of 70, 65, and 60, is he likely to continue this decline? I expected that such trends would have little or no value as predictors of the future, but I was dead wrong. Of 3,592 horses who had run three declining figures,

* 1,173 (32 percent) continued their decline,
* 134 (4 percent) ran the same figure as in the previous start.
* 2,285 (64 percent) ran an improved figure.

Beyer then went on to document the results of 4,518 horses that showed the opposite pattern — a series of three improving figures, which he later termed the “three-and-out” pattern due to the abysmal results:

* 1,125 (25 percent) improved again in their next race.
* 192 (4 percent) ran the same figure as in the previous start.
* 3,201 (71 percent) declined in the next race.
* 2,368 (52 percent) declined by 6 points or more.

“For me, the most important lessons to be drawn from the existence of the three-and-out pattern are conceptual ones,” the speed figure guru wrote. “It verifies Ragozin’s premise — which I had always doubted and often derided — that patterns of past figures may indeed be meaningful. It suggests that many other such patterns may exist. And it confirms Ragozin’s central tenet that a peak performance is apt to be followed by a decline.”

No, it doesn’t. It proves that people often accept baseless explanations if the premise is true. What Beyer observed is simple math, not some mystical event. A “peak performance” means an athlete has hit their upper limit; it doesn’t mean they’re now fatigued and need to recuperate.

When Wilt Chamberlain scored 100 points on March 2, 1962 — the statistical equivalent of seeing Bigfoot and the Loch Ness Monster on a play date with D.B. Cooper — he followed it up with a 58-point game two days later. Only 46 players in history have scored 58 points or more in an NBA game. Wilt didn’t bounce. He just came back to Earth (well, his Earth).

What about Secretariat? He set a track record in each "jewel" of the Triple Crown — Kentucky Derby, Preakness (after the fact), and Belmont — in 1973. Even though his numbers weren’t commercially available until years later, Beyer said in an interview with PBS News in 2011 that Secretariat would have earned a figure of 139 for his Belmont romp, “which is by far the best of any horse I’ve seen.”

Surely, Big Red bounced to the moon in his next start four weeks later? Well, he won by nine lengths in a time that was a fifth of a second off the track record. I'll go out on a limb and say, "no."

And therein lies the problem with the Bounce Theory: it treats normal regression like a supernatural occurrence.

Worse, when bounce proponents try to prove their point, they usually do it with their own numbers — like an Italian pizza chef rating his own pie. But bettors don’t get paid for style points or speed figures. You don’t cash a ticket that says “ran a 103.” You get paid when your horse wins.

So, I decided to look at wins and returns instead of speed figures in my database of random races from 2011 to 2023. Spoiler alert: It didn't go well for Bounce Theory advocates..

The Bounce Theory vs. Reality

To keep things fair, I only considered horses that raced over the same surface as their last start. This produced 13.4 percent winners and a $1.47 average $2 win return for all 516,378 qualifiers.

Category	Starters	Win Rate	$2 Avg. Win Return	Δ Win Rate	Δ ROI
Baseline	516,378	13.4%	$1.47
Improving Speed Figures	94,875	14.7%	$1.50	🟢 +1.3 percentage points.	🟢 +3 cents
Declining Speed Figures	64,517	11.3%	$1.40	🔴 –2.1 percentage points	🔴 –7 cents

Horse with three improving speed figures — the dreaded three-and-out pattern — increased the baseline win rate by 1.3 percentage points to 14.7 percent, and the average return by approximately three cents per $2 wager.

Conversely, three declining speed figures showed a win rate of 11.3% (down 2.1 percentage points) and an average return of $1.40 (about 7 cents less than the baseline figure).

In other words, the results were the exact opposite of what the Bounce Theory predicts — and in perfect alignment with the laws of math.

I haven’t been this shocked since Kim and Kanye said, “I don’t.”

So, the next time someone tells you a horse is “due to bounce,” smile politely, check the PPs, and maybe give that horse another look — especially if the odds are right.

Math doesn’t lie, and horses don’t bounce.

The Bounce Theory vs. Reality

More Horse Racing Stories: