• Understanding Form Cycles

    POSTED May 31, 2013
    It’s no secret that I’m not enamored with the "bounce theory" — the notion that a tough race will knock a horse off its game for weeks or months at a time. In my opinion, this theory, which was first espoused by followers of “The Sheets” (a term applied to both Len Ragozin’s speed figure service of the same name and Jerry Brown’s Thoro-graph numbers), has been very detrimental to racing.

    It is at least partly due to the bounce theory that we see trainers keep their charges on the sidelines for extended periods of time under the pretext that they are “doing what’s right” by the animals in their care. After all, horses are not machines; they need to recuperate after a strenuous effort, we are told.

    What constitutes a “strenuous” effort is anybody’s guess, however. Sometimes it means a close finish or a prolonged drive (witness Rachel Alexandra in the 2009 Woodward below); other times it is simply a race that produces a swift clocking or extraordinary speed figure.

     2009 Woodward Stakes

    In the latter case, a horse is said to have bounced if its follow-up race features a speed figure lower than the extraordinarily high figure. Of course, on the face of it, this argument is absurd. As I pointed out in “Bounce Baloney,” a piece I did for one of my Kentucky Derby betting guides, in finite data series there is a limit as to how high — or how low — the numbers can go.

    "Hence, as a horse approaches its performance threshold, the possibility of a regression, or bounce, becomes more and more likely, purely from a mathematical standpoint," I argued.

    While not meaning to, Andrew Beyer proved my contention while dissecting his own speed figures in “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 tome. "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.
    Unfortunately, Beyer ignored the obvious mathematical explanation for this data and, instead, offered a mea culpa to Ragozin and “The Sheets” players.

    "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."

    Well, yes and no. Yes, a peak performance is likely to be followed by a decline (again, this is a near-certainty if the word “peak” means anything). But, no, it does not confirm Ragozin’s central tenet — because “The Sheets” guys argue that a bounce is primarily a physical, not a mathematical, manifestation (I contend it is nearly always the latter).

    Still, we can use this concept of bouncing — mathematical bouncing, that is — to help answer that age-old question: How will our horse run today?

    I applied Beyer’s three-and-out rules to my Win Factor Report Form Ratings, which assign a percentage (0 to 100) based on a horse’s finishing position and beaten lengths in a particular race. A 100 percent rating indicates a winning effort; lower ratings indicate lesser performances.

    Here’s what I found:

    (Click on image to enlarge)
    Shockingly (not really), just as improving Beyer speed figures increase the likelihood of a regression, so too do ascending Form Ratings — especially if the most recent Rating exceeds 75 percent. 

    Likewise, deteriorating Form Ratings increase the probability of a form reversal.

    Viewed in this light, maybe the Preakness Stakes wasn’t such a head-scratcher after all. Just two horses — Oxbow and Itsmyluckyday — entered that 1 3/16-mile contest with fading form… they finished first and second respectively.

    (Click on image to enlarge)
    So, with this in mind, let’s take a look at some weekend races:

    May 31

    We’ll start at my home track, Arapahoe Park. In Friday’s opener, one horse shows deteriorating Form Ratings — and he happens to have the best overall speed figures in the field. That, to me, makes Valid Thief very dangerous, especially since the morning line favorite (Clay W. Woodstock) looks pretty weak.

    (Click on image to enlarge)
    Another race that intrigues me is the fourth at Mountaineer Park. Although there are no obvious “mathematical bounce” candidates, the figures suggest that Goose No Fruit and Tempestuous One are the ones to beat. 

    (Click on image to enlarge)
    June 2

    PRM7: In a race that looks wide open, I like the figures on 7-TETON BEAUTY and 6-PRINCESS OF POWER. Both have solid records over the local oval and both come off of deceptively good races.

    (Click on image to enlarge)

    LS6: There’s not a lot of early zip in this race, which makes 1-ENERGY'S PRIDE very interesting—especially since he recorded an insane -15 ESR chasing better over a route of ground last time. This race appears to fit him like a glove… and not the one in the OJ trial.

    (Click on image to enlarge)

    HOL5: Backing strong favorites in maiden races isn’t my cup of tea, but 1-APOSTLE PAUL appears to hold all the aces in this spot… although it is a contentious race.

    (Click on image to enlarge)

    EMD2: If 4-STRONG MOVE gets hammered in the wagering — and I think he might — I’d take a shot with 2-GRAYTFUL HARBOR, whose figures are nearly identical to those of the morning line favorite. Dutching those two (at the right, i.e. profitable, odds) is another way to go.

    (Click on image to enlarge)

    EMD8: 1-BOUND TO WIN pressed a reasonably fast pace while four-wide in his local debut. Since then, he’s had a series of stamina-building workouts and gets a meaningful rider switch. 4-ANKENY HILL set a slow pace in breaking his maiden at first asking, but: a) I’m not sure he needs the lead and b) his late speed ration (LSR) in that maiden-breaker was sensational. Play one or both as the odds dictate.