• The Favorite-Longshot Bias

    POSTED Feb 23, 2012
    Ask a roomful of value bettors the key to success at the racetrack and you’re likely to hear the term “contrarian” bandied about like the word “sucks” at the screening of a Michael Bay film. In other words, to make money at the track, one must be willing to go against the crowd from time to time — zig when others zag and ying when others yang.

    Sometimes this entails doing things that seem counterintuitive.

    Take, for example, betting to place. While most “pros” scoff at the mere notion of it, those that do espouse place wagering generally only do so if the horse in question is going to post at a decent price. No less an authority than Tom Ainslie advocated just such an approach in his “Encyclopedia of Thoroughbred Handicapping.”

    “In fields of seven or more horses, betting to place is acceptable when the selection is 7-2 or greater and the favorite figures to be out of the money,” Ainslie wrote.

    “In a field of six, betting to place is acceptable when the selection is 3-1 or greater and the favorite figures to be out of the money; in a field of five, betting to place is acceptable when the selection is 7-2 or greater and the favorite figures to be out of the money,” the father of modern-day handicapping opined.

    Of course, Ainslie’s rules make sense if one is concentrating solely on returns, rather than ROI. In a great chapter of “Betting Thoroughbreds” entitled LaPrevoyante to Win, My Wife to Place, Steve Davidowitz reached a similar conclusion after a modest win bet on Princess Doubleday turned sour in the 1972 Spinaway Stakes at Saratoga.

    Despite his wife’s suggestion that he bet Princess Doubleday to place in the event, Davidowitz and buddy Andy Beyer (yes, that Andy Beyer) stuck to their guns and backed the 40-1 Princess Doubleday primarily to win.

    Naturally, the heavily-favored LaPrevoyante won the Spinaway, while Princess Doubleday ran second, paying a juicy $15.20.

    “We’re both idiots,” Davidowitz screamed at Beyer when the reality of their “blunder” hit home 15 minutes later. “Suppose we had been told LaPrevoyante was a late scratch and wouldn’t be in the Spinaway. What price would we then have made Princess Doubleday to beat the rest of that weak field? Even money? Seven to five? Well, by ignoring the place pool, by being so longshot conscious, we just passed up the sweetest 6-1 prime bet of the year.”

    Well, yes and no.

    Davidowitz’s logic is undeniably sound — as is Ainslie’s in seeking races with a vulnerable favorite. Given that both the first- and second-place finisher affect the payoffs, it makes sense to eliminate what is (theoretically at least) the lowest price on the board when betting to place.

    But both of these handicapping luminaries ignored the most basic element of the game when they took the positions they did — mainly, that one must always strive to be on the right side of the percentages if one is to achieve long-term success at the racetrack.

    And the right side of the place-betting equation is betting favorites and other lower-priced animals in the middle, not longshots.

    This is due to what many scholars have called the “longshot-favorite bias.” Essentially, this theory states that the American (and studies have shown that the phenomenon is, in fact, most prevalent in the States) wagering public, as a whole, tends to overbet longshots and underbet favorites, especially in the place and show pools.

    In a 2006 piece that appeared in the Southern Economic Journal, authors Dr. Marshall Gramm and Douglas H. Owens documented this phenomenon.

    Gramm and Owens found that “34.7 percent of all win bets are on the race favorites, while only 30.7 percent of the place bets and 29.6 percent of the show bets are on race favorites.” By way of contrast, “the least favorite horses receive 1.9 percent of all money bet in the win pool, but 2.3 percent of the place pool and 3.1 percent of the show pool,” Gramm and Douglas pointed out.

    Obviously, if the pools were efficient, a horse’s place and show pool percentages would mirror its win pool percentage, the “Silky Sullivan” problem notwithstanding. (The Silky Sullivan problem, named for the come-from-the-clouds closer who won the 1958 Santa Anita Derby, relates to horses that typically either win or finish out of the money).

    In the mid-1980s, two scholarly gentlemen — William T. Ziemba and Donald Hausch — attempted to prove that these pool inefficiencies could be exploited for pari-mutuel gain in “Beat the Racetrack,” one of the first and only books on horserace betting that had nothing to do with handicapping.

    Through the application of simple formulas, the “Dr. Z system,” as it came to be known, showed readers how to find overlays in the place and show pools. The system was (and still is) mathematically sound — and it worked… on paper.

    Unfortunately, as I have noted on numerous occasions in the past, overlay betting that relies on small edges is nearly impossible in a pari-mutuel system — for the simple reason that the final odds aren’t known until after the betting stops. According to Gramm and Owens, “only 57 percent of the final pool totals are viewable on the tote board when betting on a race ends.”

    Even worse, this late money, more often than not, erases the more obvious pool discrepancies.

    Take, for example, the third race at Gulfstream Park on Feb. 20, 2012. With a minute to post, this is what the pool percentages looked like:



    (Click on image to enlarge)

    At zero minutes to post, just as the last few horses were being loaded into the gate, the totals looked like this:



    (Click on image to enlarge)

    And, lastly, the final numbers looked like this:



    (Click on image to enlarge)

    While there are still some discrepancies — most notably on the number six horse to show — they are noticeably less pronounced than they were a minute before post time.

    Gramm found this out the hard way.

    In a fascinating experiment, the Rhodes College professor of economics applied the Dr. Z method to 203 races at 34 tracks from February to April of 2003 — in real time.

    I’ll let the good doctor reveal his results:
    Overall, 319 $2 bets were made on 203 races. There were 113 place wagers and 206 show wagers and the net result was a $91.80 loss (-12.8 percent). While these bets looked attractive when made, the late money often lowered their expected return below [what was deemed profitable].

    A $45.70 profit (7.2 percent) would have been made had we received the payouts based upon pool totals when the wager was made. Only 90 of the original 319 wagers meet the criterion both at the time of wager and in the final total. These bets returned $9.30 or 5.2 percent. Using final pool totals, 134 wagers (43 place and 91 show) meet the criteria and [returned] $31.40 or 11.8 percent. The data from the arbitrage experiment exhibited the same favorite-longshot bias found in previous studies when final pool totals are examined. Thus, it is likely that the negative returns were not an aberration.
    So, what does all this mean to a value bettor?

    Really it just reinforces the notion that there are no easy edges out there… yet, one must still be aware of the mathematics. Betting longshots to place, as a general rule, didn’t make sense 30 years ago and it doesn’t make sense today.

    In the example of the 1972 Spinaway Stakes (referenced above), what Davidowitz glossed over is the fact that LaPrevoyante, the winner of the race, paid more to place ($3.00) than she did to win ($2.80) — making her the true (mathematical) overlay in the event.

    Remember, even if you are a great handicapper, it pays to heed the words of legendary gambler “Pittsburg Phil,” who said: “You cannot be a successful horse player if you are going to get the worst of the price all the time.”

    That’s true even in cases when it costs you $15.20.

    Dr. Marshall Gramm on TwinSpires Radio

    Be sure to catch this week’s TwinSpires podcast on Blogtalk Radio, as Dr. Marshall Gramm, co-author of Inefficiencies in Parimutuel Betting Markets Across Wagering Pools in the Simulcast Era (have you ever noticed that academic pieces always have such succinct titles?) joins me on the program. We’ll discuss his test of Dr. Z’s system and other racing-related things he’s doing.

    Click HERE to access the show page.

    The Preps Continue…

    It’s another great weekend for Kentucky Derby preps, as the Risen Star is scheduled to take place at Fair Grounds on Saturday, while the Fountain of Youth is slated for Sunday at Gulfstream Park. Below is a look at the preps run so far from a speed and pace figure perspective:



    (Click on image to enlarge)

    Weekend Win Factor Plays

    Thursday (2/23/12)

    OP1: WIN on 3-Oops And A Half (7/2 morning line odds). 3rd.

    Friday (2/24/12)

    GG6: WIN on 2-Excelling (2/1 morning line odds). 2nd.

    Sunday (2/24/12)

    GP1: WIN on 6-Don't Put It Back at odds of 4-1 or greater. Did not qualify.
    GP3: SHOW on 5-Canaima (if 70% or more of the show pool is controlled by #8). $5.00.

    GP11: From a speed and pace figure perspective, the Fountain of Youth comes down to a couple of horses — DISCREET DANCER, the likely pacesetter, and UNION RAGS, last year’s BC Juvenile runner-up. Hence, for me, it all boils down to price when it comes to betting the Grade II feature. I’m also going to look for value in the exotics with FORT LOUDON, who has shown some ability in the past and may have benefited from a tour around the Gulfstream Park oval in the Holy Bull on Jan. 12. BET(S): WIN on 5 at odds of 5-2 or greater. Did not qualify.

    OP2: BANDINI’S DREAM is a top turn-time contender that dons blinkers for the first time. He’s no Secretariat, but he looks best at a decent price. Despite a tendency to fade, HARLEY MAN scares me in this spot. I think he’s improving and Oaklawn Park is very kind to early speedsters. SHARP ALBERT is a first-time starter with good connections. BET(S): WIN/PLACE on 2. Out.

    OP9: SUNNY STEIN gets a much-needed jockey change following two crazy-fast early efforts in routes (a -23 ESR and -25 ESR on Feb. 10 and Jan. 29 respectively). She’ll be tough on the lead or rating just off the pace. BET(S): WIN on 1.
    2nd.

    PRX9: Not much form to go on; however, KEITH’S ROSE appears to be improving and recorded a very competitive turn-time in her latest. BET(S): WIN/PLACE on 6.
    $4.00 (average).

    Click HERE for more free handicapping reports and selections.
  • 1 comment:

    Anonymous said...

    I've recently, in my older and wussier, persona, made a lot of place bets - if the payoff promises to be not ridiculously low. In a race with a heavy favorite, I'll usually make a small exacta box with the favorite, so if he runs 1st or 2nd it will compensate for the lower place price on my non-favorite horse.

    Some handicapping book - back when - showed that betting odds-on horses to place (over the long run) showed like a 5% profit. (While betting them to win was a losing proposition.) Of course, if one is actually looking to make money on these situations, why not just invest in government bonds? With the state of the world's economy these days, they can be a pretty exciting investment.

    Post a Comment