Educating the uninitiated, an idea that’s countercultural and downright ludicrous, has been the superficial solution of port-brained operatives who refuse to accept that most people prefer knowing to learning. Like cavemen teaching apes to use pieces of stone to write on the walls of their shelters, these proponents of organized instructions for prospective fans have made little progress at anything and now represent a species endangered. If nothing else, Watson gave proof that machines can be endowed with the mental agility of Homo sapiens. Someday soon, they’ll be doing their work.
“I, for one, welcome our new computer overlords,” wrote game show contestant Ken Jennings on his Jeopardy drawing board after suffering a beating by the room-sized computer. Up until Wednesday, Jennings had emerged on top in 74 straight Jeopardy bouts against humans. Facing Watson, he seemed to be slow with his answers and lacking knowledge. The machine, on the other hand, was quick to recall all that was lodged in its “brain” and able to untangle convolutions. Watson was, in effect, solving problems. That’s what betting on horses requires.
One reason the challenge can never be met is because horseplayers are dealing with information about events that have happened. What has happened has only some bearing on what will happen next. No matter how sophisticated the data available is, it will be only somewhat predictive. Horseplayers rarely admit that they lose because they’ve looked backwards instead of ahead – it’s always the high rate of takeout or the fault of another or some other injustice that screws them. Doesn’t that tell you how strong their belief in the unknown is and how powerful their confidence gets?
Given the tendency for people to believe in themselves, the computers of the future could produce better wagering results than those achieved by an experienced handicapper. Artificial intelligence that's delivered without burdening the horseplayer with the task of analysis could be delivered in a variety of ways. By tapping an app on the face of a telephone and keying in questions to Watson, the patron could summon up a variety of information without having to know how to search for it. What would make this delivery of service appealing is that it would provide options, not conclusion, unlike conventional tip services that start at the end, destroying the fun of the puzzle.
In mere seconds, Watson could analyze the data that advanced players pour over for hours. It could sort out the probabilities associated with each horse in relation to the odds so that bettors know the underlay and overlays in comparison to factual history. The limitation to having an instantaneous answer for any question is merely a function of budget. If the sport locked up the intelligence provider with a license to manage the program’s distribution, such delivery of information could provide racecourses an opportunity to offer patrons something they can’t find at simulcast outlets or online.
Nevertheless, handicapping by having a machine do your research would be so popular with fans that there’d be no restricting it. For the game to refrain from capitalizing on the enormous bounties that would flow from providing the uninitiated instantly with knowledge that takes years to acquire would be difficult at best. Moreover, who’s that organization or person to work on behalf of the many and determine what’s best for the sport? The simplest of things are damn near impossible to organize.
What makes these possibilities even more science fictional is that horse racing doesn’t have the money to take on a project of this scope, even if it possessed the vision. Racecourse owners have arms too short to reach into the bottom of the bags filled with items that need funding already. The industry’s disorganization interferes with its abilities to tackle challenges. Its structure, based on 30 different states determining 30 sets of rules, is antithetical to teamwork.
It would be interesting to know what emotions are being felt in the hearts of the strategic business planners at a publisher of data such as the Daily Racing Form or organizations such as the Thoroughbred Racing Association and The Jockey Club with their Equibase Company asset - entities that derive their purpose in being to fact collection. The unprocessed information that horseplayers receive now from these sources are headed for obsolescence. As a reference tool, the information used for handicapping horse races will be as out-of-date as casebooks for lawyers and medical journals for doctors soon. A de-emphasis of the importance placed on reading the archaic language of past performances will lead to a need for being more observant of proceedings.
Ironically, an information delivery revolution might be just what is needed to turn horse racing’s fortunes. Theoretically, the sport’s opportunists can turn to Watson to open wide a fascinating game to audiences that heretofore couldn’t understand it. Once they do, all the impatient, attention deficit-plagued, give-me-the-baby-without-the-pain types may become interested in gambling the cool way.
Vic Zast posts his ideas about horse racing regularly on Facebook.com and Twitter.com. Please visit him there.
21 Feb 2011 at 05:34 am | #
Handicapping has been more of an art than science. And while results have always been black/white, the process is more like a million shades of grey and not very suitable for flow charts and computer code.
On a brighter note, skill and labor specialization have gotten mankind this far. But if we start outsourcing our minds to machines, it will get us right where we come from - into the caves.
21 Feb 2011 at 06:25 am | #
The new capability of computers will allow them to store the data found in the past performance sheets plus more and then summon that information up at the handicapper’s command. There still will be need to apply the art of interpretation, but you won’t have to ponderously compute the relative value of data and analyze what some people find to be hieroglyphics. In face, you won’t even have to know the data.
Advanced handicapping programs from Brisnet already highlight reasons why each horse is playable or not in the subtext of conventional historical lines. The computers of the future will sort the horses by any factor to enable you to compare their achievements. In the end, because you’re betting on the future, you’ll have to apply your vision and hope for luck. One big difference will be that you’ll be doing your research by asking questions to your phone instead of reading a newspaper.
Experienced handicappers won’t find making the switch to the new way easy. They’ll feel that they know more from seeing the running lines. But people who haven’t learned to master the current handicapping process will accept computer intelligence resourcing with the same ease that truck drivers who used to read road maps have become reliant on GPS.
21 Feb 2011 at 07:48 am | #
How do you think the big computer teams do it?
Bottom line though is if too many players use the exact same bets overlays become underlays in a hurry and some former underlays will become overlays.
In horse racing it is all about probability versus the takeout.
If a super computer came up with picks that everyone used, Horseplayers would still lose at the blended rate of 21%.
21 Feb 2011 at 08:10 am | #
Vic,
Your first paragraph contains the following thesis: “The length of time that it takes for people to learn how to read the past performances has forever been viewed by the sport as a detriment to making new fans.”
That’s not correct.
Prior to the emergence of legal gambling at venues other than racetracks or the state of Nevada, racing had no problem making new fans—some of whom eagerly embraced reading PPs.
Widespread legalized gambling created alternatives to horse betting, but it wasn’t the reason the complexity of PPs became “a detriment to making new fans.”
That began, roughly, in the mid-1980s. For want of a better description, the short attention span of Americans created during the so-called MTV generation changed how we process information.
Books, newspapers, concerts where people actually listed to music, contemplative thought in general, had been the norm. Those things have been replaced by the quick take.
From the 3-minute videos of MTV to present day tweets and smart phones, the evidence is clear. The art of reasoned thought is dying, if not dead. Just look at what passes for debate in our political capitals.
MTV’s first presentation was “Video Killed the Radio Star.” And just as surely as that prophecy proved correct, thought has been supplanted by reaction.
In short, it wasn’t the game which changed, it was us. And not for the better.
21 Feb 2011 at 08:19 am | #
Nick, the game changed in the 80’s. Enough so as to lose the new generation, as the older generation began going less because they were coming home with less.
Simulcast wagering gave players more plays per day to lose money on. They came home with less money.
Triactors (with higher takeouts) became more popular, and this gave players more plays. They came home with less money.
As blended takeout rates increased, and more tracks were added to the simulcast menu, there were a lot less players breaking even or winning long term.
By 1989 when Beyers were added to the form, the playing field became more level, and those with an edge using speed figs, now had their edge virtually eliminated.
The buzz that the game could be beaten was now gone.
And then came poker, a game that allowed for the odd 20 year old kid to become a Super Star.
Horse racing needs visible winners. But at a 21% blended takeout rate, that just isn’t going to happen.
21 Feb 2011 at 08:20 am | #
You make some excellent points, Nick Kling. I stand corrected. Would you accept a re-write that went, “The length of time that it takes for people to learn how to read the past performances has long been viewed by the sport...,” to be more accurate?
21 Feb 2011 at 08:57 am | #
I got outbuzzed by Watson, but he (it?) might answer yes.
21 Feb 2011 at 08:58 am | #
I believe in “No kid left behind.” Daily Racing Form as mandatory first grade curriculum.
TTT
21 Feb 2011 at 09:04 am | #
Maury,
I’m a graduate of the Barry K. Schwartz school of takeout. That is, when your product is not selling you lower the price, not raise it.
In racing’s case, a lower price benefits the customer even when she or he doesn’t realize it.
I’ve often said at seminars that (given current levels of takeout) newer bettors have two options.
1) If you want to end the year with more money than when you started, bet to win.
2) If you want a chance at that 1 in a million life-changing score, bet exotics. Just realize you’ll probably lose.
Unfortunately, most choose option 2.
21 Feb 2011 at 09:55 am | #
So Watson is going to analyze, interpret, digest, anticipate, and project a horse a) having a headache, b) stumbling out of the gate, c) jockey falling off, d) lousy jock ride, e) blocked, f) carried wide, g) bumped, h) dwelt, I) rushed, J) squeezed, k) crowded, l) sore, m) hurt and for sale, n) bobbled, o) broke down, p) blown turn, q) checked, r) steadied, s) fractious, t) ran off, u) broke in air, v) eased, w) drifted, x) ducked in, y) between horses, and z) simply tired.
21 Feb 2011 at 11:36 am | #
No. Watson is going to do none of that. You are talking about predicting future events, wmcorrow. Read the column, please. Watson is going to compile the information from past performances so it can be easily retrieved, even by people who can’t read.
21 Feb 2011 at 11:49 am | #
Good to know - Watson is not a trip handicapper.
21 Feb 2011 at 02:46 pm | #
Wonderful. Mr. Zast, you are commenting that Watson cannot deal with future events like stumbling out of the gate; so Watson is going to do the thinking for me (based on prior data); the computer is going to come up with the selection for the next race; a selection that would have taken me probably 10 minutes to select a probably winner.
Watson’s future in doing the thinking, reasoning, and decision making is terrific for Doctors, Lawyers, and everone else. Why not? The computer can store much more data subject to quick recall than the human brain; the wave of the future for mankind (what is the need to educate oneself? merely know now to turn on the computer).
Two things are in the way, though, the numerous variables involved in a horse race, and what if the plug becomes pulled out, or the electricity goes off (say, two minutes to post)? Perhaps there is a third ‘thing’: Watson cannot anticipate what is going to happen one minute from now - it is consumed by historical data fed to it by brilliant human computer experts.
Watson will never make it in the arena of handicapping thoroughbreds. Why? Because Watson does not know if his selection has a a) headache, et cetera, poor guy!
21 Feb 2011 at 03:01 pm | #
Watson will never be able to look at a horse in the Post Parade and determine how the horse is acting. Watson might be able to decipher “who is the best horse” but NOT “who is the best horse THAT DAY”
And I think I am agreeing with WMCORROW. And that scares me.
21 Feb 2011 at 04:42 pm | #
Mr. Zast: In reading you commentary again, I note some comments by you:
1. ‘Horseplayers enjoy the mind game that handicapping offers’. What empirical data to you possess to support this comment?
2. ‘The gambler believes he can beat an impossible game’. What empirical data to you possess to support this comment?
3. ‘What has happened has only some bearing on what will happen next’. I conclude that this statement makes Watson obsolete already.
4. ‘Horseplayers rarely admit that they lose because they’ve looked backwards instead of ahead’. This comment has left me scratching. Watson also is left impotent as he only looks back.
Mr. Zast, open an ADW account, go to an OTB or the track, and test your theories; then you should be able to write about us degenerate gamblers from experience.
21 Feb 2011 at 08:09 pm | #
Thanks for a thoughtful and thought-provoking post.
As noted by many of the commenters, racing represents a formidable challenge to Watson. Yet it would be interesting to see what might turn out should the IBMers be turned on to this challenge.
Saturday’s (Feb. 19) races at Gulfstream are a good example of just what Watson would have to cope with. Of the 11 races, 6 were won by horses from the 6 post with 1 second and 2 thirds from there also. Of the 6 winners, 5 were noted by the caller as tracking in the 3 path before making their run. The 6th winner was just noted as being outside before starting his move. Watson’s ability to assimulate and digest this kind of information successfully and make the necessary adjustments for handicaping future races would be scintillating indeed.
22 Feb 2011 at 02:50 pm | #
If I may, I would like to kick Watson around some more. All selections by horseplayers serious about coming up with a winner are determined from an analysis of past performances, or going with so-called expert handicappers offering their ‘pick’ via newspapers or internet sites.
At one time we serious ‘cappers used the track variant and speed rating (which I still do); then, Ol’ Andy introduced the Beyer numbers. Who has made money playing the Beyer numbers? If they were of value, wouldn’t the top Beyer number win most of the races? Try ‘em, you won’t like ‘em.
So now, Watson is going to analyze all existing racing data on any and all thoroughbreds, giving us perhaps numerous bits of information as to speed, post position, distance, weight, class, et cetera, et cetera and the selection should win according to Watson. Unfortunately, the selection won’t win in the majority of races, as Watson can’t assimulate how the nag is as they load into the gate.
Mr. Kling, in a comment above, wrote that if one wants to end the year with a profit, bet to win.
At one time, just a few years ago, Mr. Kling was correct; however, today with the double and pick three available as wagering options, they are the best (double always the best from year one).
Most serious horseplayers will agree that it is not difficult to come up with three horses that could win a horse race; thus, three horses are played in doubles and pick threes, giving a ‘capper an excellent opportunity to ‘beat this game’ (right, Mr. Zast?).
Watson will try to select the three possible thoroughbreds, but he will do no better than Ol’ Andy’s Beyer figures, which simply do not deliver winners.
BTW, I keep reading, continually, how difficult it is to read the past performances; and because of this, newbies are turned off. Are you kidding?
Damn shame that the thousands of slot players are totally unaware of what they could be winning by wagering on the ponies. But they all know about the horses, trained by Pletcher, Asmussen, Mott, Zito, Baffert, Dutrow, et cetera have in training for the Kentucky Derby.
And the beat goes on ....
22 Feb 2011 at 02:57 pm | #
More nonsense from you Wm. The idea of making money isn’t about winning the majority of races. You are seriously OTL when it comes to probabilities.
As for slot players, they have no idea about the trainers you named. They have no interest in learning how to read the pps because there are no visible winners to motivate them.
22 Feb 2011 at 03:02 pm | #
Another thing. Regarding a super computer and how a horse looks. If someone adds this info prior to post time and the computer takes it into consideration, there is no reason why the resulting selections/bets won’t be superior.
There is nothing that says that new info can’t be inputted between 20 MTP to post time.
22 Feb 2011 at 03:15 pm | #
Maury Ezra: You comment, “The idea of making money isn’t about winning the majority of races.” Golly, I win only at about a 33% rate. Who wins the majority of races? You?
As to slot players, numerous slot players are very aware of the trainers and Kentucky Derby horses, as there is nothing else to read if they happen to note a column in a newspaper or ‘catch’ a comment at a web site (thanks to the tunnel vision of turf writers). Remember, they are gamblers also. BTW, can you name a visible winner at slots? yet, slots are very popular, aren’t they?
As to what Watson will generate, you absorb the output - good luck!
22 Feb 2011 at 06:29 pm | #
“Ironically, an information delivery revolution might be just what is needed to turn horse racing’s fortunes.”
Outstanding piece, VZ. It revived my respect for your work following a run of misguided rantings against racing’s yet-to-be-revered customers. Maybe there’s an activist inside you after all. Your comment #2 was also insightful as the time spent to perform the actual handicapping process, once learned, can be as much an obstacle as the time it takes to learn how to do it. Not that you showed us any more respect this time, but you certainly shed new light on the process that so intrigues us.
“The limitation to having an instantaneous answer for any question is merely a function of budget.”
Far from being Science Fiction, the instantaneous answer for horseplayers is available now, though at a prohibitive cost for most recreational bettors. You almost inspired me to jump up and purchase Jeff Platt’s “JCapper” software and connect it to the Dragon “Naturally Speaking” voice-recognition program on a new “Quad Core” computer, switch to the fastest Internet Service Provider, start a subscription to a suitable data source, and then translate my insights and strategies into software robots (a newbie could go with the default robots and probably do better than many experienced handicappers).
I could then sit back and verbally instruct my computer to immediately isolate, rank, and justify the contenders for any given race or some defined subset of all the races to be run on a given day, i.e., the pick six at a Southern California track, or perhaps those at any track in which a generated “Performance Potential” rating for a single horse exceeded those of its fellow entrants by a specified margin.
The program numbers of the contenders surviving my visual inspection in the paddock and post parade could then be passed on to my cellphone wagering app with access to an on-line tote board and my ADW account. The app would send my wagers to my ADW in appropriate amounts and types based on available bankroll to optimize my return when the outcome matched the computer’s projections and my personal observations. No more exhausting past performance perusals or last minute pressure.
The fly in the ointment besides cost is transferability of one’s insights. As it happens, on some weekends I do race rundowns with a human computer named Vern using DRF “Classic” past performances. After a 10-15 minute analysis of a race, he’ll come up with three groups of entrants to play in the superfecta: 1) those to play in the top four slots, 2) those to play in the bottom three slots, and 3) those to play in the bottom two slots. He will pass on about half of the races. Even without late adjustment for actual odds, strong favorites, and alternate trifecta value, his results are impressive. The man is an incredible handicapper and instinctive gambler, but as far as either of us can tell, not all of what he does can be replicated by a computer; at least not until a computer itself is capable of translating his instincts into robots.
Since I believe my own less subliminal (and, so far, less rewarding) analyses can be successfully converted, I might take a shot at it if I also had access to a historical data base which I could verbally query to automatically build and test robots without extensive keying and visual analysis of test results. I don’t know to what extent the various racing data base software packages can provide that capability at this time, but I’m certain they’ll eventually reach that point. The only question is whether the game will still be around by the time they can.
Until a consortium of racetracks can deliver Watson-like functionality, maybe HANA could become a host for a historical data base its members could cost-effectively query to conduct research and/or build their robots, and only have to pay full-price for ongoing data following their initial entry into live handicapping. Perhaps HANA could fund some portion of its activities by operating such a data base.
23 Feb 2011 at 08:27 am | #
Indulto, enjoyed #21. I was always hoping that the Daily Racing Form would provide such a database, with software far more maleable than “formulator,” but it seems they have decided for their financial purposes to relegate us to linear calculations, and restrict us from accessing and manipulating their proprietary information as we see fit; I guess they have to serve the majority. Personally, I’m not interested in the percentage of winners Todd Pletcher has had in the month of November with 4-year olds on the turf in claiming races over the past 5 years, etc., but there are those that are, and I am in the minority. One thing I do not do is criticize another man’s method of handicapping, and I never argue with success. What a game, heh?
TTT
24 Feb 2011 at 04:00 am | #
TTT,
I would think the DRF is best positioned to offer subscribers a natural language interface to a database query/handicapping robotic capability. At the current price for the Print Edition with a limited number of tracks, wouldn’t many customers be willing to pay that for a pass by a custom robot against all tracks for a day? DRF could still service the existing majority and after a while, they would become the minority.