The Skeptical Handicapper: Using Data and Brains to Win At the Racetrack

horseplayers.

Introduction

A book on handicapping has been rattling around my brain for a long, long time.

I wrote Money Secrets at the Racetrack in 1988. It was a guide to managing your money at the track, but had no advice about handicapping. You do the handicapping; the book then showed you how to make the proper bets.

Oh, there was plenty of advice about handicapping around, and there had been for decades. Much of it was of the make-every-racetrack-your-private-bank variety. All you needed were a few simple rules, or, in later years, a decent computer program, and endless riches would soon be showered upon you.

Sadly, for most people this turned out not to be the case.

You could buy a handicapping book, invest in a system, attend a seminar, or learn a hot new software program. Or you could forget all that and just buy somebody’s selections over the phone or the internet. But steady profits? Those belonged mostly to those who sold the material. Making money at the track on a long-term, steady basis proved elusive for all but a few.

My introduction to horses came via a 1950s television show called Racing From Yonkers, which featured this newfangled thing called videotape. Each night, I’d be treated to harness races where it seemed everybody was named Hanover. I didn’t know anybody that gambled and my parents had zero interest in the sport. But I watched, fascinated. Soon, my elementary-school friends and I were gambling for nickels—we’d make a selection in each race, and whoever had the most winners got paid by the others. One kid, David, kept winning, and before long I realized he was buying a local newspaper that featured a handicapper who analyzed the races, while the rest of us were just playing numbers. It was my introduction to the idea that if you had better information and knew what to do with it, good things could happen.

I toyed with the harness races for awhile, and even wrote a book called Success at the Harness Races, which was published before I was legally allowed to enter the track, although that technicality never stopped me. I enjoyed the races as a hobby, going so far as to buy a few cheap claimers over the years with my earnings as a tennis instructor. Occasionally, I fantasized about gambling for a living. Of course whenever I would mention this to anyone, I would be met with sad, pitying shakes of the head.

Later, I moved to California with the idea of a career in sitcom writing, only to discover that nobody thought I was very funny. By this time I had a wife and child, and there were things like rent and food that required money. To give you some idea of my delusional thinking, I figured that my best chance to acquire said money would be to bet the local harness races. Not only had I never made much money betting harness races in the past, but I knew absolutely nothing about the horses, trainers, and drivers who raced out west. Surprisingly, however, this harebrained idea actually didn’t turn out that badly. I wasn’t getting rich, but the three of us never got evicted, either.

After five years at the harness races, I decided to switch to betting thoroughbreds instead. A key reason for my switch was that my son was starting school, which meant that if I was gone at night (when the harness horses raced), I’d hardly see him at all. But at the afternoon thoroughbred races, I could go after those huge pick 6 scores, I could make large bets without affecting the payoffs, I’d win way more money, and my home life would be a delight. Or so I thought.

Little did I know that thoroughbred racing is to harness racing as basketball is to golf. That is, they’re both sports and there’s a ball involved. Or in the case of racing, there are horses involved. And that’s about it. I had a lot to learn. A very large lot.

In my learning period, it would have been wise for me to bet small amounts or maybe nothing at all. Of course, Mr. Self Confidence started betting large amounts, immediately. And losing said large amounts. I was working every day and night, with nothing to show for it except a thinner wallet.

But I was still determined to pursue my new career, even if there was no evidence that I could succeed. And somewhere during the ensuing year, I began to figure things out.

In 1990, thanks largely to a couple of big scores, I made more than $100,000 at the runners. Finally, all the work was rewarded. And a career, albeit an unusual one, was launched.

When I first got into thoroughbred racing, you had to go to the track to bet. Later came simulcast betting, and then eventually phone and internet betting. For much of my thoroughbred betting career, I worked from my home office. My life was bounded by a television (to see the races live and to watch the replays later), a computer (to compile information and make bets), and a phone (to make more bets).

Exactly what I did in my office was a mystery to most. My own mother knew that somehow I was involved with horse racing, although she was unclear on the details. One day she visited me in Los Angeles and ventured into my office while the races were on. She took a look at my computer monitors, which might display a live toteboard, an exacta matrix, and the odds on two exchanges. She understood none of this. But having overheard me call in a four-figure bet to a rebate shop, she did have one question for me: Was that real money? Yes, Mom, it was.

Daily, I experienced the highs and lows of serious, big-money gambling. It seemed like my body was always on alert, every race an adrenaline rush. This is not a path that leads to good health, and eventually I decided to retire.

Now, my adrenaline doesn’t rush. It moseys. Sure, I still watch horse races every so often. Who could resist the 2016 Breeders’ Cup duels between Songbird and Beholder, or between Arrogate and California Chrome? Yet now I watch as a fan, with no money down. Horses are still wonderful, magnificent creatures, and they’ll be with me forever. Not just as the 4-6 exacta, but as precious, unforgettable memories. I’ll never forget all those Saturdays when I used to bring the neighborhood kids to the morning qualifying races at the local harness track, the whole gang running all over the empty grandstand while I carefully made notes about the on-track pacers. Winning my first big pick 6 at the thoroughbreds, proving to myself that I could actually do this. Watching a 2-year-old colt named Officer win his first five starts in such an effortless, ethereal way that it seemed he just floated around the track. And being at Santa Anita the day in 2009 when the brilliant mare Zenyatta turned in her usual last-to-first performance, only this time it was against the top males in the Breeders’ Cup Classic, the fans’ cheers of appreciation still ringing in my ears all these years later.

Every once in awhile, I make the short drive from my current home to the training center at San Luis Rey Downs and pass a relaxing hour or two watching thoroughbreds go through their paces. Who are they? I don’t care. The sun shines a shimmering yellow, and noble steeds are doing what they do. It is a lovely way to spend a morning, and always will be.

In retirement, I have found I prefer a relaxing day at the pool to working on a ticket for hours in attempt to take down the pool—although occasionally I’ll play a ticket if there’s a gigantic carryover or mandatory payout somewhere, so now and again you’ll find me working diligently on a card from Woodbine or Arlington, tracks I never played when I was gambling daily. Time, previously at a premium, is now available.

Many players had asked me whether I’d be writing a handicapping book. I asked myself that, too. And if I did, what kind of book would it be?

My first thought was to do what many others had done—display some past performances, then explain how I had brilliantly arrived at the winner. Maybe even do a whole card. But the more I thought about it, the less the idea interested me. A typical 7-1 shot wins about one time in ten. Over the course of a season, thousands of these horses would win. It would be no big deal to find a bunch of them and then explain how I would have picked them. Am I not the smart guy? I could even include some that I had actually bet.

Of course, there’d be no way to tell how many losers I picked this way. So yeah, I had that 7-1 shot, but maybe I had just lost on 25 straight previous 7-1 shots, which would not have seemed especially impressive. Maybe the very method I had used to pick 7-1 shots actually hit only one of them every 15 bets, and thus was a horrible long-term strategy. Maybe the 7-1 winner I displayed actually should have been 12-1, so it was a severe underlay and would take away from the central point of professional gambling—which is to seek advantages, not just winners.

My next thought was to do A Year In book. I would record every bet I made all year, with reasons for same. You, the engrossed reader, would be able to follow me through all the ups and downs of a year in gambling, something like Andy Beyer’s classic My $50,000 Year at the Races. You would experience the financial, and emotional, highs and lows as you followed my progress, or lack thereof. I had used such an approach for my book Blackjack Autumn, in which I spent two months playing blackjack in every casino in Nevada. However, that book was more of an adventure travelogue, with places I’d seen and people I’d met relieving the tedium of and then I hit a 5 for a 21. Gambling on horses for me has always been a solitary pursuit. First me alone at the track, then me alone at a simulcast facility, and later me alone in my office. No interesting places, no colorful characters. A potential snoozefest.

Worse, what would happen if I had a bad year? A losing year, or maybe one in which I’d make $2,000 for 250 days of work? Who would buy such a book, other than my enemies?

Besides, in my last few years much of my play had been on overseas betting exchanges. Betfair barred Americans years ago (although in 2016, New Jersey players were reinstated at Betfair, at a far higher takeout than I faced), EHorse no longer offers an exchange, and iBetX and TradeBetX are gone altogether. And I don’t think my adventures using an out-of-country proxy server and mysterious wire transfers from a foreign friend would be relatable to many readers.

Some of my additional action took place with stateside bookies and offshore racebooks, and again this is outside the experience of the typical law-abiding horseplayer. Nobody’s legs ever got broken, and only one time in all those years did I fail to get paid (when a large Curacao book, Aces Gold, closed in 2002). Still, why encourage people to go outside the system?

Then there was my involvement with rebate shops. These places operate within the law, with contracts with U.S. racetracks to accept players as long as the bettor’s handle reaches a certain annual amount. Some take action on Track A but not on Track B, while another might take B but not A. Some are legal in State X but frowned on in State Y. In any event, how much you get back varies depending on such factors as the type of bet, the track, your yearly bet totals, and your own negotiating skills. You might get 3% back on a win bet at Santa Anita and 22% on a trifecta at Penn National. Again, the typical bettor has little interaction with rebate houses.

All in all, when I added the money I bet with bookies, racebooks and rebate shops to the many millions I risked on exchanges, the total dwarfed the amounts I bet directly through racetrack windows, simulcast centers and non-rebate ADWs, although that amount wasn’t insignificant. Thus I was, to some extent, playing on a different field from the vast majority of players. I’m not sure if you’ve ever, in the same race, played a win bet with an offshore racebook in Costa Rica, a bet-against on an exchange in England, and a pick 3 bet with a rebate shop in North Dakota. This I did routinely, for years, although the locations changed from time to time.

What about examples? They offer a clear way to understand a concept. For instance, let’s say I believe that bullet workouts are the key to handicapping success; I would have no trouble coming up with a dozen examples where a winner was sent away at a decent price despite a recent bullet. However, if I thought that bullet workouts were overrated, I would also have no trouble coming up with a dozen examples where a horse was bet heavily primarily due to his workouts, but did nothing. Either way, I could print the appropriate past performances, circle the bullets, and you would easily see what I was talking about.

However, the problem with examples is not that they’re wrong, but that they may not tell us much. Same with anecdotes. I could spin a lovely yarn about how I made a big score using such-and-such an angle. It might even be true. But so what? Without knowing how that angle did over thousands of races, we have no way of knowing if the angle is useful or not. It’s something like feeling a pain in your elbow and seeking Aunt Tilly’s medical advice because once she had a pain in her elbow, too.

Much of what's been wrong with handicapping literature has to do with a failure to understand the difference anecdotes and evidence. A thousand anecdotes equal zero evidence. Just because I bet a horse that dropped two classes and won, and here are fifty other similar winning droppers, doesn't mean betting them all is a good idea. Maybe it is, and maybe it isn't.

To research this question, we'd try to gather as much data as possible and then tally how all the double droppers did. Maybe we might uncover some seemingly successful patterns (such as the horse switched jockeys, the horse came off a layoff, the trainer was ranked in the top 10, the horse was a shipper, double droppers did better on Fridays, etc.). If the number of races that contains the angle is in the thousands, we’re normally in pretty good shape. If it’s only in the hundreds, there’s a much better chance that any positive result was due mostly to luck (of course in small samples, the angle might have been especially unlucky, too; typically, most results featuring low-odds horses lose somewhere around the track take, while angles spotlighting longshots yield slightly worse results).

Here's where handicapping books have let us down. The author either fails to do any statistical research at all, relying on his own experiences, or he simply doesn't understand much about the principles of research. Time after time, authors either fiddle with an original sample and then cite the cherry-picked result as proof of some idea, or else use a woefully inadequate sample size. Or they write stuff which is simply wrong.

So what kind of book would I write? I decided there would be no How I Picked Every Winner, which would be disingenuous. No Year in the Life, which would be irrelevant for most readers. No Study These Examples and You’ll Win, since they wouldn’t be backed up with anything other than my own opinion. But there was one fertile area that had rarely been trodden—the statistical approach to handicapping

In this volume, there will be no day-at-the-races stories with Runyonesque handicappers, no tales of how I won every race at the Breeders’ Cup, and no identify the OO-X reverse pattern and you’ll win. Nothing about angles some long-time horseplayer liked, or some guy at a seminar suggested, or some handicapping author crowed about. I decided for this book to let the data speak. Less opinion, more facts. Thus, this is a book about facts.

I will cheerfully agree that facts, particularly as detailed in the many tables in this book, may bore some—er, most—people. Because facts are rarely as exciting as flashy stories about six-figure winning photos, or tales of colorful characters like Jay the Tool, or the Beast, or Fingers. But facts reveal truths in a way that anecdotes, no matter how memorable, may not. My suggestion: take your time going through these tables, and you may find some rubies that may cause you to reevaluate (or at least modify) your approach.

Facts are objective statements of reality—the typical professional basketball player is taller than the average professional bowler, a desk is heavier than a feather, and favorites win more often than longshots. But if our facts are not really facts, and merely guesses, we’re in trouble.

Certainly, there are other ways to approach the game than what I say here. Some players might read this book and say, This Meadow guy is an idiot. He doesn’t even talk about A and B and C and D, which is what I use to make zillions every year. In fact, the professional gamblers I have met all do something a little different not only from me, but from each other. Some are brilliant programmers who are forever tweaking their computer models, while others handicap with a pen and a Daily Racing Form (also known simply as DRF or the Form). Some bet tracks all over the country, while others play their home circuit and that’s that. Some go to the track every day, and others haven’t been inside an actual racetrack in years. I would never say that a player can’t win if he doesn’t do X, or that he couldn’t win if he does Y. Everybody is different, and talent isn’t equally distributed, either.

And what may be obvious to one person may not be so self-evident to someone else. My brain is not your brain. For instance, you may be a whiz at completing a 2,000-piece jigsaw puzzle. Me? I have a hard time completing the 16-piece puzzles meant for 3-year-olds. So please forgive me if what I say in parts of this volume is obvious to you—it may not be equally apparent to everyone else.

But no matter what anybody’s approach or skill level, it’s useful to start with facts.

I don't believe in the bounce theory, says one player. This horse has a chance if you believe in the Beyers, says another. I like this one because I believe in horses for courses, says a third.

I don't know when human beings first decided they would believe or not believe things for no discernible reason—instead of analyzing the evidence to determine whether the belief is correct or not—but I'm sure it was somewhere around the time somebody looked up and announced, I believe that somebody painted those stars up there.

How do you know that something is true? Every day, people bet their handicapping beliefs. Often they’re based on false assumptions—which is why (along with the high takeout) so many players continue to lose.

Sometimes, beliefs are logical but they're still wrong. Take the popular notion that a first-time starter has an advantage if he trains at the track where he'll be racing. Some reasons: the babies don't have to ship in for the race, they're familiar with the surface and the environment, some trainers stable their best newcomers on track and the lesser ones off track, obscure trainers with 80-1 shots can't get stalls at the track and must ship in from farms which further depresses the off-course stats, etc. So I asked Ken Massa—more on him later—to check out the theory, and he reviewed 63,893 first-time starters nationwide. Did debuters do better on their home field (defined as the track where they had their final workout before making their first lifetime start)? The results:

Home field: 4,662 wins from 44,728 starts (10.4%)

Away field: 1,757 wins from 19,165 starts ( 9.2%)

A slightly better win percentage for the home field debuters, but not appreciably so. Yet you will hear handicappers insist that first-time starters have a major edge if they've trained where they will race.

It is crucial to differentiate between truth and opinions without confusing the two. There is no my truth or your truth, except in the non-verifiable area of faith. In math, 3 + 4 will always equal 7. That’s not what I believe or the guy down the block believes. It’s the truth.

Opinions are different. You can prefer the red shirt to the blue shirt, while your neighbor has the opposite preference. There is no the truth here, just opinions. Two intelligent, well-informed people can make opposing judgments about the shirt.

Racetrack handicapping and betting consist of a mixture of the truth and your opinion. But it’s important not to confuse the two.

The truth might be that a horse was blocked last race. Your opinion might be that the horse should run a better race today.

Your opinion is simply a prediction based on analyzing previous truths. The problem comes when your opinion is based on false assumptions—truths which are not true. Alternative facts and fake news, as a certain U.S. president might tweet.

Let's look at a toteboard assumption for a moment. Let's say we believe that heavy toteboard action on a horse is a positive sign. That will lead us down roads where we may find ourselves recording every toteboard flash, or trying to find the source of the money, or blindly betting the hot horses. However, before worrying about patterns of the toteboard, let us examine the assumption itself—is heavy toteboard action on a horse really a positive sign? Hmmm. Maybe somebody knows something, or maybe somebody just thinks he knows something. Or maybe the owner pounds the horse every start. Or somebody meant to bet #3 and accidentally called #4 by mistake. The reality is that nobody for sure knows or why a horse gets played—but most commonly, a morning-line 8-1 shot who goes off at 3-1 is bet down because the line was awful; any competent handicapper would have made the horse 3-1, which is what the crowd finally did. In any event, a large percentage of the pool shows up in the final few seconds before off time—so even if you knew the money was coming in from an unimpeachable inside source, there isn’t time to do anything about it.

Then there’s the problem of causation vs. correlation. Things happen. Sometimes Thing A causes Result X. Sometimes it is merely one of many possible causes. Sometimes it is associated with the result but doesn't cause it. Sometimes it has little or even nothing to do with that result. A rooster crows, and then the sun rises. Did the rooster’s cry cause the sun to rise?

If a horse won with a new jockey, would have he won just as easily with another jockey? If he added blinkers and raced better, did the blinkers help or was it just a coincidence?

Numbers seem to offer us certainty. After all, if I say a certain sire is good with first-time starters that's one thing, but if I say he's won with 22% of his debuters that sounds much more impressive, doesn't it? Or if I tell you that a trainer wins 18% first off the claim, a handicapping system hits 29% winners, or a jockey's year-to-date win percentage is 19%. But will any of these stats—or others—help your bottom line? Or will they just mislead you?

Even though statistics always beat anecdotes and feelings and memories and opinions, at least when it comes to betting on horses, there is still a place for human judgment. What if you had accurate, meaningful statistics and you had a great feel for the game, understanding when to toss a big last-race speed figure or when to upgrade a horse with seemingly low numbers? Now you’re onto something!

We’ll ask many questions in this book. Each time we come up against a testable assumption, we will examine it using a large computer database. Not everything is testable or has been researched on large numbers of horses. For instance, I haven’t seen any stats on horses who break through the gate at the start of a race; my guess is they probably do worse than horses who don’t, but I don’t know for sure. How about dry horses in the post parade vs. washy ones? Without data, who knows?

But plenty of questions, probably most, can be researched. Is a horse's lifetime win percentage important? If a horse shows early speed in a maiden race and drops into a maiden claimer, is that a gambling positive? Do horses with trouble lines perform especially well the next time out? What about horses getting Lasix for the first time?

Many of us have ideas about these subjects—but unless you've checked them out for thousands of races, they are only ideas. To check on class droppers, we looked at 273,260 of them. By examining which characteristics are more, and less, successful than others, we have a much better handle on what will work and what won’t—as a positive to get us closer to winning, or as a negative for a possible play-against.

At this point, some readers may be saying something like this:

I’m not going to make 273,260 bets on a particular type of horse, so who cares about the results of 273,260 bets? I am going to zero in on my own specialties and my own angles, based on my own experience. Not every 6-1 shot is equally likely to win, and I’m able to figure out which ones are good bets and which ones aren’t. So why are you combining good and bad bets and spitting out totals which have nothing to do with me?

All of us are products of our own background and experiences. If I made some big scores early in my betting career using some particular angle, chances are I’ll keep searching for its reappearance. Maybe the angle worked thirty years ago but, due to the improvement of handicappers generally, it no longer works today. Or, ominously, maybe it never worked very well and I was just lucky.

So we’re going to look at large samples, even though we know the number of horses analyzed in some of these surveys is far greater than any individual would ever bet if he gambled every day until he was 100. Because in large samples there is truth, and in small samples there is luck and volatility.

However, just because the public might be right in the aggregate about certain factors doesn’t necessarily mean the fans are right in every race. Each event is a different conundrum, with the importance of each variable ever shifting.

Decades ago, when simulcasting was in its infancy, each track maintained its own separate pools, and it wasn’t uncommon to find a horse going off at 4-1 at one track and 12-1 at a different track. Both crowds couldn’t have been right. And sometimes the crowd who’s betting today’s fourth race isn’t right, either. Maybe the post positions mean a great deal in today’s second race, but not so much in today’s fifth. Maybe a jockey switch is a key ingredient in the sixth race, but no big deal in the eighth. Evaluating a race differently from how the crowd does gives you the best chance to win, if you’re right often enough.

Heck, I’m probably not any smarter than you. I just paid attention. Kept an eye out for improving and declining jockeys. Realized that replays were not drudgery, but an opportunity for down-the-road profits. Took the time to try to plot the probable pace of every race. Attempted to get inside the minds of trainers, both good and bad.

Our surveys were not designed to find automatically profitable bets. Although a few one-angle plays did just that over thousands of races, nobody can guarantee the angle will hold up profitably into the future. Instead, the studies in this book will show you promising areas to research, or at least to consider when you’re handicapping. And the detailed questions and discussions about many topics will, I hope, steer you onto a path where winning is more than just possible for you.

Maybe, even, probable.

Our Challenge

The handicapping game has changed. Drastically. Not by an evolution, but by a revolution.

The biggest game changer: the rise of well-funded computerized robotic wagering (CRW) teams, also known as rebate teams. Back in ancient times—say, thirty years ago—handicapping software was in its infancy, you couldn’t see races outside your own state (except in Nevada), and there were no rebates. Until then, a professional gambler usually commuted daily to a home track, watched replays at the site of the local races, and made friends with owners and trainers and jockeys to pick up some of his information. His opponents were people who liked to gamble, and not well.

No more. The occasional fans have switched to the casinos and the lottery, the jammed grandstands of yore left as abandoned as a Detroit fixer-upper. Others will no doubt shift their action to legal sports betting, finally permitted by state choice after the Supreme Court ruling in 2018. And thanks to the proliferation of information in the computer age, the remaining players’ skills have improved, markedly.

The rebate teams, which first began to appear in the 1980s and gathered steam by the turn of the century, now account for an estimated 20% of all nationwide thoroughbred handle. Led by such brilliant innovators as Bill Benter, Zeljko Ranogajec, and Peter Wagner, the current incarnations, with input both from programmers and live human handicappers who can tell you who was five wide around the turn last time out, are good. Really, really good. Some of the largest rebate groups play every track in the country, every day.

They work on the same principles explained in Money Secrets at the Racetrack—crunching the data to arrive at accurate betting lines, then playing the discrepancies. More than one racetrack executive has lamented the fact that his own on-track customers are getting pulverized by the CRW teams who can make numerous simultaneous bets on any overlays the software detects. These teams don’t have to win—but if they lose 3% and get a 12% rebate, that’s a $90,000 profit for each million they bet, and some of them bet tens of millions of dollars a year. Some have been able to open their own ADWs, so their only cost is the signal fee to the track, and everything else is gravy.

Their employees are constantly analyzing results and adjusting the software. The advent of rebates created something new in American racing—the well-financed team of computer handicappers. Modeled on what Benter and his late partner Alan Woods had done in Hong Kong, these teams use their own proprietary software to identify overlays—horses or combinations whose odds are higher than they should be (as calculated by the software). The bets are then displayed on various monitors; the computers are configured to emulate betting terminals. Clerks then punch the send button and the bets are made directly into the pari-mutuel pools.

Typical of the team leaders is Dana Parham, a giant player at the biggest rebate shop, Racing and Gaming Services in St. Kitts. Parham’s team routinely bets well north of $100 million per year. Their operation covers every thoroughbred and harness track in the North America, though the amount of the bets is scaled to the size of the pools. Though the overlay may be the same, the program might recommend a $7,000 bet on a race at Saratoga but only a $70 bet in a similar circumstance at an obscure Canadian harness track.

Many people think their job is to pick winners, and this is a fallacy, said Parham in a recent interview. His computer programs don't make selections. Instead, they make a price on every horse—and they play only when there are discrepancies as post time approaches. If you're out trying to pick winners, this will not necessarily make you money, he said. You've got to pick advantages.

By advantages, he gave the example of a horse who's listed on the board at 9-1, but which his computer program identifies as a good bet at 7-1 or higher. At the last possible second, the program calculates the maximum amount to bet without pushing the horse's odds below 7-1. Then the clerk sends the bet.

Winners follow advantages, Parham continued. It's the same with stocks, commodities, baseball, or football. Essentially, the question is, 'Where is the value?’

And that value may vary from pool to pool. What's an overlay in the win pool might be an underlay in the exacta pool, said Parham. Many bettors treat all pools the same, and they're not.

Parham has laid out the master plan to racetrack success, which goes something like this:

• Gather the information.

• Synthesize the information and form an opinion.

• See if there's a discrepancy between your opinion and the crowd's.

• If so, use the board to determine your bet or bets.

• Have the discipline to play only those races that offer an advantage.

The rebate teams are very sharp. So are some lone-wolf individual big players who are great at reading the Ragozin Sheets, or use their own proprietary 50-page computer printouts, or watch each race nine times to make notes on all nine horses. They're betting serious money, and they're not so easy to beat.

OK, let’s go positive now. Yes, you can win. And if I had to pick one word to describe what winning players look for, it's discrepancy.

You cannot find a discrepancy until you check the toteboard, because the toteboard is the crowd's opinion of the race (morning lines don't matter, because nobody pays off on the morning line).

When reviewing the board, ask of each horse whether his odds are about right, too high, or too low. If the odds are about right for everyone, you have no bet. If someone's odds are too high, you might have a play. If too low, you might have a bet-against. And if you do find a discrepancy, ask why your view of the race is more accurate than the public's.

Is your opinion good enough? Good enough to understand why the public is pounding #4, and why #7 is actually a better bet? Good enough to be able to take advantage when the crowd makes #3 the 7-2 second choice, when you feel he shouldn’t be better than 8-1?

The Skeptical Handicapper is here to help—helping you identify factors which may be overbet, or underbet. The old ways of relying on speed figures, or calculating earnings per start, or following the top jockeys, aren’t going to work any more. And that’s good, because many of your competitors remain stuck in the past, using methods that perhaps once were useful, but are now a relic of a former age.

Partially due to the success of the rebate teams and the lone wolves, factors which once may have been profitable have now turned out to be not so. One of the software outputs in Ken Massa’s HTR computer program is one he terms Fraction 1 Velocity, the result of an algorithm that attempts to predict which horse will be in front at the first call. Calculating this number for tens of thousands of races each year landed these results:

FRACTION 1 VELOCITY RESULTS

The rating has chugged along with virtually the same win percentage, but look at how the ROI has tumbled. Massa found similarly unhappy results using Quirin Speed Points; the top-ranked QSP horses’ prices have fallen off the cliff, with no group of six, seven, or eight points (the highest QSP-rated horses) showing an ROI exceeding 0.80. And as further evidence of the decline of early speed as a handicapping factor, we checked the first-call position of every horse in its last start for every U.S. race in the past four years. All surfaces and distances and classes were combined:

LAST-RACE POSITION AT FIRST CALL

This unfortunate fact—you discover something, only to find that other researchers are unearthing the same thing, and your prices collapse—is why you can’t use stats from the Civil War to play today. Too many players are looking at the same data, so what might have proved profitable previously may not be worth much any more.

While sophisticated, ever-changing programs have facilitated the rise of the rebate teams, today’s powerful new software has also been responsible for the creation of new past performance products which can be used by recreational players. Among them: TimeformUS, OptixEQ, STATS Race Lens, and TrackMaster’s BubbleCapper. Many of the newer programs offer not just numbers, but startling visual impressions of varying shapes and colors to help players recognize pace scenarios or statistical trends. They can provide pedigree and trainer-jockey stats that are up-to-today’s-card accurate. And a player can customize some of them in a seemingly endless variety of ways.

These products have become more and more sophisticated. Aimed at the solo handicapper who does his work himself rather than relying strictly on the computer’s suggestions, these commercial programs have helped the loners stay competitive.

Some programs have their own terminology for different factors, and a dedicated student might find his nights filled with arcane discussions of Median Energy Variable, Pace Fulcrum, OptixPLOT Squares, Sustained Match-Up Variegate, Double Razor, Pace Model Index, and Kanga Power Numbers. And once you look at any subject through new eyes, you’re bound to learn things that would never have occurred to you had you been stuck in the same old rut. If you’ve never had a winning year, looking at handicapping in novel ways might be just the thing to kickstart your success. A good program can be a useful tool that will take much of the mathematical drudgery out of your daily handicapping chores, and help you analyze races through an entirely different point of view.

A powerful software program can look at copious permutations and relationships of factors; for instance, with factors A, B, and C, it can look at A, B, C, AB, AC, BC, and ABC. Horse racing may contain hundreds of possible factors (days away can be broken into ranges of days since last in-the-money finish, days since last race, days since last race at this distance, and many more), and thus the relationships climb exponentially—impossible to evaluate for a human, but feasible for a computer program. Horse racing contains enough uncertainty that nobody can win every race. But if a computer program can win a few more races—or, more importantly, identify overlays—it just might push your results closer to the number where rebates can make you a winner. And that’s what the rebate teams do, or at least try to do.

Many companies today are working on advanced artificial intelligence programs, the future promising to change the world. Today, robo-advisors are managing portfolios at a fraction of the cost of human financial planners, and robo-radiologists may soon be interpreting your x-rays. But the future promises much, much more. Forget Siri and Alexa, or even facial recognition—auto accidents might be eliminated once everyone embraces driverless cars, and quantum computers are poised to far outperform all of today’s computers. Yes, computers have beaten the world champions of chess and Go. But now consider this: In 2017, Google’s Alpha Zero program taught itself to play chess through machine learning—and with no human input, and after just four hours of training, it was able to defeat Stockfish 8, which several times had won the world computer chess championship. Compared with these developments, creating a decent horse race handicapping program seems like child’s play.

But if it were cheap or easy, everybody would do it. It is neither. But the fact that some computer groups continue to make strong profits—at the expense of the everyday player—should give us pause, particularly as their action comprises an even more significant percentage of the pools. You are no longer competing against hobbyists who bet on hunches and lucky numbers; instead, you’ve got to beat full-time gamblers who, guided by emotionless algorithms, are intent on stealing your money.

The rebate players are betting big money, and with their bets going in at the last second, they can drastically impact the toteboard. Thus some fans may still be steamed about the 2018 Dwyer Stakes at Belmont, when Firenze Fire, 6-1 with 20 seconds to go before the race began, got hammered to the tune of more than $124,000 and wound up at 5-2. Oh yeah: he won by 9 lengths. But that event pales when compared with the opening race of the Hollywood Park spring meeting in 2002, when a bet of $118,000 on a filly ironically named Global Finance was sent in by a computer group; the on-track toteboard was down, and the wager didn’t appear on the track’s TV monitors until the race was nearly completed. The horse went from 9-2 to 2-5 and won the race; later, it was revealed that the bet was actually a mistake, and should have been the team’s normal four-figure bet. Even computer bettors would sometimes rather be lucky than good.

In the good old pre-computer days, if you had solid speed and pace numbers and little else, you still had a decent chance to win. But those carefree days are long gone. The crowd’s win percentages have gone up, but the prices for the most part have dipped. The public has been getting increasingly adept at picking winners, and it’s been hammering the payoffs. Let’s compare the results of betting every nationwide favorite in 1998 with doing the same in 2008 and 2017. The win percentage is followed by the average $2 payoff.

There appear to be several reasons for the public’s apparent improvement as shown in this comparison over two decades. The expanded use of computerized data, and a jump in the percentage of the handle bet by the aforementioned sophisticated rebate teams, increases win percentages and depresses prices. The plunge in the size of the foal crops (as recently as 2006, the North American foal crop totaled 38,104; in 2017, the number dropped to 22,500) means that fewer horses are available to race, which leads to less competitive fields as the same group of foes frequently faces each other.

To illustrate this, let’s turn to a survey from researcher Oliver Crunk (@o_crunk) regarding 241,462 odds-on favorites from 2000-2017. Look at the percentage of races that featured an odds-on favorite over the years, as listed for the most recent five-year intervals. In 2002, only 23% of races featured an odds-on favorite. Today, that number exceeds 33%. And they go off at lower odds, and win more often. Field sizes are smaller as well—8.1 was the average field size in 2000, while the 2018 figure was less than 7.7. Combine small fields with easy-to-find-by-everybody winners and the opportunities to make money diminish:

PERCENTAGE OF RACES WITH ODDS-ON FAVORITES/RESULTS

Perhaps worst of all, there’s virtually no way to improve the odds-on favorites’ win ROI. Ken Massa checked dozens of factors for the 45,168 horses who went off as odds-on favorites from 2014 through 2017 and found only one—blinkers off, where 291 winners from the 492 odds-on favorites earned a 0.95 ROI—where the ROI exceeded 0.88. Sure, the odds-on horse won often. So what?

In some jurisdictions, the takeout has been raised, which cuts into the amounts returned to bettors. The closing of such casual-player facilities as New York City’s OTB (which once offered 150 parlors) has also played a part in making the game more difficult. Plenty of information available, the least successful players disappearing—the challenge is greater than ever.

Overall, the news isn’t especially good for horse racing. Nationwide handle, which totaled nearly $15.2 billion in 2003, plummeted to less than $11 billion by 2011, and has been bumping along at around that number ever since. From 2007 through 2017, while Nevada’s sportsbooks nearly doubled their handle, the state’s racebooks’ handle fell by more than half. Tracks ranging from major player Hollywood Park to tiny Les Bois Park have closed, and nobody’s building any new facilities. Tracks have slashed the number of racing days, and the number of races on those days. Horses aren’t racing as often; in 1960, the average horse started 11.3 times, while his 2017 counterpart started just 6.2 times.

The smaller number of players who remain have a choice of new handicapping software programs, and the availability of up-to-the-minute stats via such sources as the Daily Racing Form’s Formulator. As a result, bad morning lines get corrected with alarming swiftness. And worst of all for the typical player, the toteboard is becoming less reliable; it’s not uncommon for a horse who’s 4-1 on the board as the horses enter the gate to wind up at 5-2 somewhere around the far turn, which makes it more difficult to predict what a horse’s final odds will be.

One thing hasn’t changed—some players are long-term winners, but the vast majority are not. How many bettors are winning players? No one knows for sure, but the overall news isn’t promising: Bettors, as a group, must lose.

Start a beautiful summer afternoon with fans betting $1 million in the first race, then never replenishing their bankrolls after that. Assuming an overall takeout of 20%, here’s what’s left after each race:

How sad: After 10 races, nearly 90% of the money the bettors have brought with them, full of hope, has disappeared. Thus the constant dips (into wallets), trips (to the ATM), and sips (of alcohol to ease the pain).

Now add in more races (why just bet New York when you can bet Maryland and Florida and California, too?) and more days (why go to the track once when you can go twice a week for a year, and make some phone bets on the other days?), and you can quickly see that, for bettors as a whole, the game is a one-way ticket to bankruptcy.

Most are destined to end up lifetime financial losers, victims of the usurious takeout extracted by the state and track. Many accept this fate, and even joke about it. They like the action, or the camaraderie, or maybe just the chance to escape their regular lives for a little while. That their hobby costs them a few dollars is no big deal.

A small minority, though, doesn't accept the old premise that you can beat a race, but you can't beat the races. This statement makes no logical sense—after all, if you beat enough individual races, you certainly can beat the races—yet it's accepted as a truism by many gamblers. And for them, it is true. They probably won't beat the races.

But why is that? Unlike roulette or craps—two negative-expectation games that cannot be beaten in the long run by any betting system due to their fixed odds—horse racing can be beaten. You simply have to outplay the other gamblers, because they set the odds. You're not competing against the casino—you're battling the other players, many of whom have never watched a replay, compiled a speed figure, or researched a pedigree number.

Many people approach their daily job as if it’s a game. They arrive late and leave early, surf the internet, make long personal phone calls, take leisurely lunch breaks, etc. Their idea is to get through the day without much stress, rather than to provide value for the employer. If they can get something (money) for nothing (little work), they've won the game.

This attitude often translates to the racetrack—and that's good news for serious players. If you are not willing to work, you have virtually no chance to win. If you are willing to work, at least you have a chance. The odds are still against you, but at least the odds are no longer zero.

To succeed, you’ve got to be able to assess horses better than the public at large, and bet only when there’s a discrepancy between your opinion and the crowd’s. The game is mostly not simply about picking winners—it’s about outplaying your fellow bettors. Take any number that produces many winners—the Brisnet Prime Power Ratings, last-race best Beyers, Today’s Racing Digest Comprehensive Performance Ratings, Equibase Speed Figures—and you’ll wind up with something around 30% winners. But you will have a hard time getting the ROI above 0.85, so it has limited value from the standpoint of potential profit.

Not that picking winners is useless. Get home one obvious 4-5 shot and you still might make a score in the pick 5. And at times you’ll handicap a race where you like one horse to win, and you hate at least one of the top three favorites, so you can structure a play-against in bets such as exactas and superfectas.

But usually, finding underbet and overbet horses is more important than picking winners.

Let us get familiar with an irritating fact that will be repeated elsewhere in this volume: When something is obvious to the public, it is usually pounded into unprofitability. Which means we have to look for the not so obvious. Where the public zigs, we must zag. If they’re overbetting Factor A, we need to consider Factor B.

At one time, stagecoach drivers and Morse Code operators and lamplighters were in high demand. But today, nobody thinks of these skills as potential careers. Thus it is in the handicapping world—ideas which at one time might have been useful have been left behind, particularly as the rebate teams have asserted their power. While speed figures and jockey percentages and earnings-per-start calculations may not help you make any money today, there are still plenty of opportunities.

Especially, consider areas where you might have an advantage over the rebate teams—workout notes, body language, track bias, replay observations, deep dives into trainer thinking, race-flow predictions. I had some of my best years at the same time the rebate groups were having theirs. Yes, they’re damn good, but they’re not invincible.

The biggest reasons that most horseplayers don’t win are:

There’s a large takeout, exceeding 25% on some bets at some tracks.

The public is very good