Links
- Handicapping Edge <% if LEN(SESSION("ADMIN"))>0 then %>
- Control Panel
- Wesite Doc
- Blog Editor <% end if %>
Archives
- Sunday, January 25, 2004
- Monday, June 07, 2004
- Thursday, July 01, 2004
- Monday, July 05, 2004
- Thursday, July 15, 2004
- Friday, July 16, 2004
- Thursday, July 29, 2004
- Wednesday, August 04, 2004
- Saturday, August 28, 2004
- Tuesday, August 31, 2004
- Wednesday, September 01, 2004
- Thursday, September 02, 2004
- Friday, September 03, 2004
- Sunday, September 05, 2004
- Monday, September 06, 2004
- Friday, September 17, 2004
- Saturday, October 02, 2004
- Tuesday, October 05, 2004
- Wednesday, October 06, 2004
- Monday, October 25, 2004
- Tuesday, October 26, 2004
- Wednesday, December 08, 2004
- Thursday, January 06, 2005
- Saturday, February 05, 2005
- Monday, February 28, 2005
- Thursday, March 24, 2005
- Thursday, March 31, 2005
- Friday, April 01, 2005
- Monday, April 04, 2005
- Thursday, May 05, 2005
- Friday, May 06, 2005
- Wednesday, May 18, 2005
- Thursday, May 19, 2005
- Friday, May 20, 2005
- Monday, June 06, 2005
- Thursday, June 09, 2005
- Friday, June 10, 2005
- Saturday, June 11, 2005
- Monday, June 13, 2005
- Friday, July 01, 2005
- Saturday, July 02, 2005
- Thursday, July 14, 2005
- Friday, July 15, 2005
- Saturday, July 16, 2005
- Tuesday, July 26, 2005
- Thursday, September 01, 2005
- Friday, September 30, 2005
- Saturday, October 01, 2005
- Sunday, October 02, 2005
- Tuesday, October 04, 2005
- Monday, October 24, 2005
- Wednesday, October 26, 2005
- Thursday, October 27, 2005
- Friday, October 28, 2005
- Saturday, November 12, 2005
- Friday, December 16, 2005
- Sunday, January 29, 2006
- Wednesday, March 15, 2006
- Thursday, April 06, 2006
- Friday, April 07, 2006
- Saturday, April 08, 2006
- Thursday, May 04, 2006
- Friday, May 05, 2006
- Saturday, May 06, 2006
- Friday, May 19, 2006
- Saturday, May 20, 2006
- Saturday, May 27, 2006
- Friday, June 09, 2006
- Friday, July 07, 2006
- Friday, July 14, 2006
- Wednesday, August 09, 2006
- Wednesday, August 23, 2006
- Friday, September 29, 2006
- Saturday, September 30, 2006
- Sunday, October 01, 2006
- Tuesday, October 31, 2006
- Wednesday, November 01, 2006
- Thursday, November 02, 2006
- Friday, November 03, 2006
- Monday, November 13, 2006
- Thursday, January 04, 2007
- Friday, February 16, 2007
- Sunday, April 22, 2007
- Thursday, May 03, 2007
- Friday, May 04, 2007
- Friday, May 18, 2007
- Thursday, June 07, 2007
- Sunday, June 24, 2007
- Friday, August 10, 2007
- Sunday, August 12, 2007
- Friday, August 17, 2007
- Tuesday, September 18, 2007
- Wednesday, September 19, 2007
- Friday, October 05, 2007
- Saturday, October 06, 2007
- Wednesday, October 24, 2007
- Thursday, October 25, 2007
- Friday, October 26, 2007
- Saturday, October 27, 2007
- Monday, November 19, 2007
- Monday, January 07, 2008
- Monday, February 18, 2008
- Wednesday, March 26, 2008
- Friday, March 28, 2008
- Thursday, May 01, 2008
- Friday, May 02, 2008
- Friday, May 16, 2008
- Friday, June 06, 2008
- Tuesday, June 17, 2008
- Monday, July 14, 2008
- Thursday, July 31, 2008
- Friday, August 01, 2008
- Friday, August 08, 2008
- Saturday, August 09, 2008
- Monday, September 22, 2008
- Friday, October 03, 2008
- Saturday, October 04, 2008
- Wednesday, October 22, 2008
- Thursday, October 23, 2008
- Friday, October 24, 2008
- Saturday, October 25, 2008
- Friday, November 21, 2008
- Wednesday, February 25, 2009
- Thursday, February 26, 2009
- Saturday, March 28, 2009
- Wednesday, April 29, 2009
- Thursday, April 30, 2009
- Friday, May 01, 2009
- Thursday, May 14, 2009
- Friday, May 15, 2009
- Wednesday, May 27, 2009
- Thursday, June 04, 2009
- Friday, June 05, 2009
- Thursday, August 06, 2009
- Friday, August 07, 2009
- Saturday, August 08, 2009
- Wednesday, September 09, 2009
- Friday, October 02, 2009
- Saturday, October 03, 2009
- Saturday, October 24, 2009
- Wednesday, November 04, 2009
- Thursday, November 05, 2009
- Friday, November 06, 2009
- Saturday, January 30, 2010
- Friday, July 09, 2010
Mark Cramer's C & X Report for the HandicappingEdge.Com.
Monday, September 22, 2008
C&X
CONTENTS
EDITORIAL
CAN WE CREATE A TIME-SAVING COMPUTER METHOD?
SPIRIT ONE: WHO SAYS WIN OVERLAYS ARE INCREASINGLY DIFFICULT TO FIND?
LINE-MAKING WORKSHOP
C&X CAFÉ:
On Random Betting: Cramer Validated
Class vs. Pars
SOME THOUGHTS ON THE EARNINGS-PER-RACE FACTOR
NEW BOOK ON SYNTHETIC SURFACES
ART KAUFMAN
EDITORIAL
C&X has a long history, having begun as the C&O Report in the early 1990s, thanks to the encouragement and contribution of Bill Olmsted, who was the O. Bill reached a crossroads and decided to move on, even though the publication had been a success. The X today refers to X-rated handicappers.
As anyone who follows the publishing business can guess, print publications are an endangered species, especially newsletters. Dave and I have continued to put out the C&X Report because of mellow inertia: it’s something we like to do, but it’s not a business success. During crunch time, I have often thought of phasing out the publication.
On the other hand, my love for racing keeps growing, and a byproduct of doing C&X is that I am forced to remain a student of the game; this publication could not have survived if we had been recycling old material.
As we near the end of another cycle of issues, I need to decide whether or not I should continue, and if so, in what form and format. For sure, something does have to change, and if it does, we need to make sure it’s for the better.
As you will see from this issue, a continuation of C&X might drift in the direction of more interactive involvement of readers, perhaps some sort of group project.
The lead article of this issue will give you an idea of where my mind is meandering.
Please take note that the next Stakes Weekend will be the Arc de Triomphe, the first Sunday in October, and I’ll try to throw in another race from the USA on the Saturday of the same weekend. If you have not seen my own general and eclectic website before, take a look. www.altiplanopublications.com
Within a few days there should be some new stuff posted there.
mc
CAN WE CREATE A TIME-SAVING COMPUTER METHOD?
During this period I am mulling over what to do with a possible next cycle of C&X, I am trying to find a formula that will build on what we’ve done, a new level that approaches more tangible results on our return on investment.
Ken, one of our dedicated readers and a studious and smart horseplayer, has echoed my sentiments that we need a time-saving short-cut. Of all the people I’ve ever known, Ken is the only one who does not know how to be lazy. He’s a workaholic and he’s ruthlessly detailed in analyzing the races and studying the literature. He’s also the type of devil’s advocate that helps us refine our thinking.
Wise men like Ken are aware that no being from our planet, even the most successful one, can carry on a daily wagering routine without eventually losing concentration, playing catch-up in the information gathering and sorting process, and eventually burning out.
I have been saying something similar for years but hearing it from Ken has convinced me to act more boldly. I’ve seen too many good players reach their half-life and then wilt away from the game.
Before getting to the meat of this article, allow me to indulge in an anecdote that will explain viscerally why the time factor is tantamount. It was the third day of a tournament at Sam’s Town. Mike Helm and I were playing as partners. We had come off a $1,000 prize the previous day for highest profit of the day, thanks to one of my turf-sire bets and one of Mike’s first-time starters, both based on a SECONDARY FACTOR (remember that for later). It was not the first time we had won such a prize. Since the entry fee for the tournament was $500, we were assured of doubling our money, even if we tapped out.
We were in close pressing position, within the top six or seven leaders, but two days and nights of handicapping and decision making were already taking their toll, and going to sleep at 2pm after hours of intense handicapping is not the way to get the sound sleep one needs for a fair level of concentration the next day.
It is especially discouraging to try to sleep only 5 hours before sunrise, when you have not made any handicapping discover for the next day’s cards. The mind is too stimulated to allow for deep sleep. My biggest enemy was not bad handicapping but a lack of that most precious commodity called time.
The next day, groggy from workaholism, we were hard pressed to come up with nine of the required ten bets. As the race cards evoloved, still looking through my hazy eyes at the pps, I came up with a “maybe horse” four our tenth bet. (The tournament required the player to bet all the bankroll each day, ten percent of bankroll on each race. Very bold indeed. Too bold according to the math of Dick Mitchell.
The horse I had found was a cheap claimer stretching out into a route race that looked like the lone speed but had little else in its favor. By the time I had digested my thoughts and had found a couple of other thin reasons for our playing over a $1,000 on this horse, I went up to the special betting tables for tournament players. The horses were at the gate and beginning to load. I handed in the bet ticket.
The lady told me that it was too late for a bet. Only two of the 11 horses had loaded the gate.
“Sorry,” she said, “tournament wagering is closed after the first horse loads.”
Needless to say, our horse took the lead and went wire to wire. He paid over $70 to win. That one horse would have put us in first place. We had missed the opportunity not because of bad handicapping but from a lack of time.
From then on, we were so psychologically devastated, partly from the fatigue of handicapping for so many hours, that it was all downhill, and we tapped out on the last day. It was the most depressing way to make a hundred percent profit in the tournament.
With a 200% profit for the totality of my tournament play, spanning over seven tournaments, it was time to quit. I resolved to never again play in a tournament unless I had discovered the type of automatic bets that would not corrode my time.
Andrew Beyer, a successful workaholic, once said (I’m paraphrasing) that cutting corners, even minor time-saving behavior could mean the difference between a consolation and a huge pick six payoff, just by missing one piece of information.
But for most of us, a comprehensive approach is unsustainable and most of us cannot go at it day in and day out without eventually burning out. Beyer is an exception. He’s still at it. But he is also able to take a vacation, get on his bicycle, and take up a completely different challenge.
On the other hand, if one does the work up front, creating a streamlined model that extracts what is essential, there is a way to bet intelligently and successfully. We want a computer model (or one that can be done by pencil and paper) that will land us on enough longshot winners to sustain a long-term profit.
Today, it’s just as possible to find a longshot winner as it was decades ago when much less information was available. This is contrary to what many experts have alleged. They say that it’s increasingly difficult to outperform the public. I say they are wrong. The average payoff of betting favorites may have declined to some degree but the percentage of favorites has not increased and the number of longshot winners has not declined. This is because the public’s weak point today is not a lack of information but information overload.
Now please follow my logic, because, before I present the hypothesis for a computer model, you need to understand the foundation. Rather than “cutting corners”, why not synthesize, which means extracting what is truly essential, which is really a small part of the available information. That’s not cutting corners, it’s cutting out the fat. We could call it “lean handicapping” but the food industry has already appropriated the term.
Helping us in our endeavor is the research of the very people who believe that what we want to do is impossible. People like my friend Barry Meadow, who has shown with incredibly huge research samples that none of the major handicapping factors yield a positive return on investment.
Meadow’s conclusion: there is no mechanical method.
My conclusion: there is no mechanical method based on primary handicapping factors.
Do you see where I’m going? By cutting out the text book factors like speed, those factors with negative wager value, we save time without losing depth. My research has shown that winners in races where they have not changed distance or surface (to eliminate other variables) increase their Beyer fig by an average of nearly 9 points. Amazing but true. You win by projecting change and not predicting continuity.
So, primary or conventional factors which predict continuity can be diminished in importance and we can base whatever computer model we can construct on contrarian factors, some of which are recognized as “secondary” factors by the experts and others which remain under the radar of public consciousness.
All right, so that’s the foundation. Now comes the invitation. Even if I have not had the pleasure of communicating with many of our readers, I feel we have the nucleus of a potential community, what is referred to in academia as “collegiality”. We should also understand that a community is based to a certain extent on sharing but not so much as to deny individuality, which is important for maintaining the dynamism of the whole. A real community finds the right balance between collective consciousness and rugged individuality. This balance could help us come up with the right model for either a computer program or pen-and-paper method.
Over the years, C&X readers have participated by asking good questions, sending opinions, sending criticism, and even sending independent research. There are some talented researchers among us, some of them skilled in computer programming. My call is to you, that some of you volunteer to help out with this project and that it be shared with our community, because, as an old saying goes, “If you can’t share it, you don’t have it.” This is a potential format for continuing C&X. Certainly our collective analytical skill is superior to that of any individual among us, as proven empirically in the book The Wisdom of Crowds, which also recognizes that the group, even the most sophisticated group, can be overcome on occasions by groupthink, thus enabling an informed minority of one (or few) to find the right path.
My first step is to propose a foundation for a potential computer program whose purpose is to save lots of time without cutting corners. Much of what I include in this proposal is based on previous research, of mine or of colleagues, with other parts based on lesser but still important anecdotal evidence.
As I’ve noted, perhaps, in looking for a formula for another cycle of C&X (because intellectually I need a higher level or I won’t see the reason for continuing) we could engage in this group project in search of a time-saving computer method, pooling our diverse types of knowledge and awareness.
My awareness is conceptual but not technological. The more technically oriented among us are invited to study this model and reflect on whether it will fit in a “program”. In the end, we may discover that a computer is not even necessary but that our model can be done with pen and paper, still saving time from comprehensive handicapping but gaining in pari-mutuel force. So here goes.
Concept:
One of the dilemmas of handicapping each race is identifying which factors should take precedence. So the computer must choose which three factors among the many will be used. We extract the essential, and a three factor model allows us to triangulate between the factors.
Once the computer has selected each of the three factors, if more than three horses qualify by a particular factor, the computer eliminates that factor from consideration. (The three-horse parameter can later be modified according to number of horses per field:
Between 7 and 9 horses, no more than three qualifying horses for a factor to kick in.
Between 10 and up, no more than four qualifying horses for a factor to kick in.
Between 5 and 6 horses, no more than two qualifying horses.
Once the method of interaction between the three factors is determined, we will then decide whether the model fits more for straight win bets of for exotics. I suspect the latter, for in my own pen-and-paper experimentation with this model, I have not been able to whittle it down to a single horse. Not yet.
(2) Once the computer designates the three most significant factors, the next step is triangulation. Horses that qualify by all three factors get priority over those qualifying by two. A horse qualifying under only one factor is NOT eliminated, but gets less priority.
However, a horse may only qualifies under one factor BUT if this horse is an only qualifier under this particular factor, then this may count more than a horse that is a double or triple qualifier among three horses in a category.
Here are the factors I have selected as either “secondary” or contrarian.
Trainer 1. Horse shows at least one profitable category in DRF trainer specialty stats for this particular race, and this profitable caterory must show at least 18% winners, so that we know that the positive roi did not come from a fluke longshot. Remember that if more than three horses qualify by this parameter, the computer will reject the parameter and move on to the next one.
Class. Horse must not have lost more than once AT TODAY’S CLASS LEVEL OR VS. LESSER (a proven loser). This goes for maiden special weight, maiden claiming, claiming and the allowance “non-winners of” levels. Races at other levels, such as starter allowance, mid-level stakes or graded stakes do not apply to this rule because, each for a different reason, they are too difficult to objectify. (I have anecdotal evidence that stakes races, for example are often at a different level than their class designation, because of poor fields.) If more than three horses in the field are not proven losers, then the computer will reject this parameter and move on to the next.
Course. Horse has won 30% or more races at this track/surface. Easy to notice: it’s right in the performance box. If more than three horses qualify, the computer rejects this parameter.
Trainer 2. Trainer has a 18% or more overall win record this year (or this year plus last year if we are within the first two months of the calendar year). Again, if more than three horses meet this criterion through their trainer, the computer rejects this factor.
Surface (“turf” or “dirt-fast”). Horse has won 30% or more on the surface of today’s race.
Overachiever. Horse shows at least three running lines in pps in which it has EITHER won at 5-1 or up, finished in the money at 9-1 or up, or a combination of both. Or, horse shows two such finishes in its last four races. Once more, the computer will reject this factor if more than three horses qualify as overachievers.
I say the computer, but a smart programmer may be able to design a pen-and-paper way to use these six parameters. They are easy to identify with a quick scan of the pps.
The computer is now supposed to select the three categories in which there are qualifiers. I can imagine using two categories as well, if none or the others qualify, and I can also envision using four. If we have five or six working factors, then we either (a) have to choose the best three from them, or (b) use them all with some unique cross-tallying method. Even so, the concept of triangulating, or balancing three sides, needs to be explored and refined.
Now here, computer programmers, please heed: we are looking for an interaction of factors that goes more deeply than simply counting points. We want something with balance of strength, which means, perhaps, that qualifying by more than one factor is equally important as being an only qualifier under one factor.
If only one factor emerges, we still will work with that factor, considering that the rest of the race is a hodgepodge of factors, a chaos race, and we want to extract the one thin slice that makes one horse truly different from the others.
Anywhere beyond a single parameter may be more applicable to exotics than to win betting, but that’s only an initial impression. Since these are all secondary factors, it may be possible that one horse towers above all the rest and still goes off at 10-1.
The dilemma.
I have been studying this methodology anecdotally with both American and French races. In races where there are three qualifying factors, sometimes one factor trumps the others, and other times there’s a complex interrelationship between these factors.
About a month ago, a French TV documentary crew accompanied me to Chantilly race course on a pre-scheduled filming. My back was to the wall, as these were lesser-of-evil races. The filmmaker knew nothing about handicapping, so if I lost, he would assume, anecdotally, than it was all hype and beating the races was impossible. If I won, he would at least consider that human thoughtfulness, analysis and creativity can be a deciding factor in making a profit at the races.
I zoomed in on one of the race where it looked as if the payoff would be good, and I used this method and triangulated three factors, trainer 2, class and course. My class factor here was different from the above model because the class levels are not the same as we know them, so I used “average earnings per race for the year” and required 3,000 per race in order to qualify. I played the French version of the superfecta, which is really a 4 horse quinella: you win no matter what the order.
No horse was an only qualifier under any of the factors. However there were some horses that qualified by two factors and others by one. I wheeled the dual qualifiers with single qualifiers. In other words, if two single qualifiers came in the winning combo, I wouldn’t collect. One combination came in, and that was thanks to a longshot single qualifier that qualified only by the course criteria, a horse I would have eliminated based on comprehensive handicapping. This showed me that the method picks horses that the public is less likely to pick. I thus salvaged my reputation with a generous payoff in front of the camera.
Let me highlight the fact that if I had done comprehensive handicapping of the same race, I would have eliminated my single qualifier and ended up with 3 of 4. Since that afternoon I have noticed that my automatic criteria pick in-the-money finishers that I would not have used in exotic bets based on conventional handicapping. And yet, this unconventional method is based, fundamentally, on meaningful statistics
But to show you the dilemma, on another of these mechanical bets, more recently, I triangulated three factors but one of the factors (trainer 2) completely trumped the other two, and the payoff was huge. In other words, the question our computer needs to answer is: when does one contrarian factor eclipse all the others and when is a triangulation of factors called for?
I’ve practiced this hypothesis with American races for a much longer time. The French experience only corroborates my American records: but again, the big question that we want the computer to decide (if possible) is: when will one factor trump the other two.
Once the computer discards the non-factors, it has to make a decision as to whether or not all three qualifying factors (or both qualifying factors if there are only two) should be integrated or if one factor should drown out the other(s).
I have some ideas as to how to do this, but they are based on anecdotal evidence.
I don’t believe any other computer method is close to the area I’m covering. I have eliminated the three most classic factors: speed, pace and current form, on the notion that they are much too overbet by the public.
I also have a relatively simple method for the computer to decide which factor(s) should be used, based on the need for one to three horses to qualify within a factor, no more and no less. If four horses qualify by the horse-course factor, for example, then we eliminate that factor. And on the other end, if no horses qualify, then more obviously it becomes a non-factor.
I did not come up with this idea in a vacuum. I spent long hours with Ed Bain, watching him play his automatic trainer method, and he showed me that often if there were more than two qualifying trainers, the winner would come from elsewhere and not from the qualifiers.
One of the big advantages of this hypothetical structure for a computer method is that I have eliminated the need for choosing a running line.
My research on trainer and class shows that these two factors have validated wager value.
The course/surface seems anecdotally valid.
So please contact me if you have computer experience and think there’s a relatively simple way to come up with a program. My biggest concern is the time it would take for data entry. Would that data entry time be better spent in just doing it by pen and paper?
Even if we end up with no computer system, we may be on to a time-saving mechanical way of handicapping that does not sacrifice depth of analysis.
Sorry for a certain degree of repetition. I want to make sure this different approach is clear to all.
SPIRIT ONE: WHO SAYS WIN OVERLAYS ARE INCREASINGLY DIFFICULT TO FIND?
During the past decade, a recurring cliché in the pari-mutuel milieu has emerged: because so much information is now available, it is increasingly hard to find an overlay in the win pool. Therefore, we need to look to the exotic pools as our only hope for a winning return on investment.
First and foremost, it should be noted that when we have more information available, it becomes increasingly difficult to manage that information. The C&X Report has often dealt with this dilemma, offering alternative ways of sifting through information and separating the good stuff from what is banal. “Thin slicing”, for example, or picking out one particular secondary factor that trumps all the others, is one of our approaches.
In the case of the Arlington Million, all the information one needed to know was right in the past performances. In the C&X Stakes Weekend website analysis we noted that there were “comparables” that could be used when appraising the wager value of the three Euro horses: Archipenko, Mount Nelson and Spirit One. The legitimate but overbet favorite, Archipenko (6/5) had defeated a horse named Balius by a length and _ in April. Spirit One had defeated Balius by 3 in March, and Balius was six lengths better than the rest of the field so his race was not a clunker.
Then we saw that Mount Nelson had been defeated by a horse named named Athanor that had been defeated by the French filly Cicerole, suggesting that Mount Nelson’s competition was not the classiest. So my comment was that even if these three Euro horses could be rated in a similar bracket, because of the hype, Spirit One figured to have the highest odds. We also learned, directly from the DRF workout comments that Spirit One had looked exceptionally fit in his Arlington workout. And finally, perhaps less evident in the racing pages was the fact that Spirit One’s early speed, a negative factor in the European bias scene, would become positive in American turf racing. I have been writing about this factor for a number of years: Euro horses moving from against-the-bias racing to with the bias.
Perhaps less evident to the American betting public was the fine record of trainer Philippe Demercastel, and the personal anecdote I had about the fact that a top American trainer had wanted Ioritz Mendizabal to ride for him.
Okay, so we had some privileged information, but any handicapper knowing the value of information, knowing that information equals pari-mutuel power, could have seen that Spirit One had been the pace setter in a key race whose top three finishers had come back to win, and it was clear from DRF articles that the winning horse in that field, Duke de Marmalade, towered over the Arlington Million contenders. Spirit One had finished only 2 _ behind the Duke.
The result of the Arlington Million was not deterministic. On another day, if Archipenko does not get boxed in by Murtagh on Mount Nelson, he might catch Spirit One.
The point is that there was an odds differential between the three Euro horses that should not have been. With Archipenko at nearly even money, Mount Nelson’s odds were pushed up to a hard 7/2. But how does one explain the 13-1 odds on Spirit One?
Simply stated, the betting public got it wrong, and this betting public, deemed so accurate by the naysayers (those who say that win overlays are hard to come by) was not even close.
Fortunately, we sometimes have a way of measuring these things. The leading handicapper in the Paris-Turf, the French version of the DRF, listed his choices for the Arlington Million as Archipenko, Mount Nelson, Einstein, Stream Cat and only then Spirit One, so there was no national preference in favor of the French horse among the experts. But in their separate pool, French players made Spirit One 6-1.
The occasional existence of separate pools proves that betting publics can still get it wrong, even in the information age. On some of these rare occasions, I have the luxury of choosing my betting pool. In the Beverly D the French public had a similar view of the eventual winner Mauralakana (the horse I named as most-likely winner) but Communique, one of my contenders, was 30-1 in France and only 12-1 in the USA, so I used Communique with Mauralakana in one of my quinella bets and got back three times the odds that an exacta box paid in the USA.
Until now, you might argue that the pari-mutuel inefficiency is restricted to foreign horses, but this is not the case. First, the American public did make Archipenko the heavy favorite, so there was absolutely no prejudice against the foreigners. There was, however, a tendency to follow the hype: what I’ve often called “the publicity horse”.
Archipenko was the publicity horse, and that usually has an impact on the odds. This leads me to a second area where all the information in the world will still not make the betting public efficient or accurate. In any stakes races where stars are participating, the hype has an extraordinary influence on the odds.
I had made this point on the C&X Stakes Weekend website, for the March 29 Bonnie Miss Stakes at Gulfstream, where I noted that the two favorites would draw excessive action because of the classy company they kept. Each of those two had negative trainer stats. So I had no privileged information beyond what was found in the DRF.
Hype cannot overcome the combination of bad pace scenario and negative trainers stats, so I went with She’s All Eltish on top (a hard 4-1) and Robbie’s Girl as the place horse, and that’s the way they finished in the exacta. It was not so much choosing the top two horses but recognizing an inefficiency in the pari-mutuel market because of the presence of two publicity horses.
Robbie’s Girl indicates yet another facet of betting public inefficiency. Robbie’s Girl had finished in the money at high odds on several occasions, and did so again in the Bonnie Lass. Certain horses, call them “un-hype” horses, have a record of being consistently underbet by the public. I call these horses “overachievers”. They are horses that regularly win or finish in the money at high odds. I had identified Robbie’s Girl as an overachiever in the place hole. Overlays are still found in the straight pools when overachievers can be identified.
Yet another opportunity for overlays in the win hole is found in the realm of shippers. The natural human tendency is to embrace what is already known. Given the choice of an underlay restaurant that is already known or a potential overlay restaurant on the same corner that may or may not end up delighting, most people fear venturing into new territory and willingly accept the underlay.
Shippers are the new guy in town and some of them may offer overlay potential, if the player does a little homework on the trainer.
These are just a few areas where win overlays are still to be had. The betting public may have become more efficient in picking low-priced horses but in the middle or upper odds ranges, live horses may remain untapped by even the smartest players.
LINE-MAKING WORKSHOP
We’ve gone through many issues of C&X with no line-making workshops and I fear some of the newer subscribers may scratch their heads when I make certain references to a “personal odds line”. Therefore, I’d like to revisit the process with one of the races we did for the Stakes Weekend. Those of you who already are aware of this process cannot be hurt by this refresher. Let’s look back at the Claiming Crown Tiara on the turf. Here you see the line I posted on the Stakes Weekend site:
Simply Run 7-2 (need 5-1 for the overlay). 5-1 ML
Inhonorofjohnnie 4-1 (need 6-1). 4-1 ML
House Quest 4-1 (need 6-1). 7-2 ML
Allnightdance 5-1 (need 7.5-1). 5-2 ML
Couple Whiles 6-1 (need 9-1). 6-1 ML
The other two at 15-1.
As you can see, this was a deeply contentious race, and even the other two horses not included as contenders were given a chance as part of the line.
First, please notice that my line odds are based on a 100% probability that there will be a winner: no more, no less. Program odds, what we call the morning line, use approximately 120 percentage points. By adding percentage probability for each horse they are lowing the odds of each horse, (a) in order to reflect the track takeout, which also essentially lowers the odds of each horse, and (b) to encourage more betting by creating an illusion that toteboard odds are “fair”. More horses will “seem” like overlays when their odds are compared to a 120% morning line, and fewer horses will really be going off above their line odds when the a personal odds line based on 100% is set side to side with the morning line odds.
Second, please note that I have two numbers. My first number is my perceived intrinsic probability of the horse to win, based on the 100 percentage points available for all horses. These odds are based on a balance between objectivity (stats, figs, etc.) and subjectivity, my (or your) personal analysis and synthesis of that objective information. It is possible, but much more complex than you can imagine, to make a truly objective line. If only one factor mattered, say trainer percentage, then if the trainer win rate of a horse is 20% I could give the horse an automatic 4-1, the equivalent of 20%. But as we all no there are too many other stats, some of which contradict each other, and thus, the line maker needs to make subjective appraisals.
My second number is what I need in order to make a play. What I need is 50% more than the line odds. Those 50% refer to a betting advantage. For example, if I make a horse 2-1 and he’s going off at 2-1, that means no advantage. The only way to make a long-term profit at the races is to play only when you have an advantage. If your line is relatively accurate (it does not have to be perfect), then you should be comfortable in not playing any horse that is not clearly above your line odds.
Any horse that you make above 6-1 is considered by you as a non-contender. In the higher-odds range there is less of a difference in percentage points and therefore a greater chance for error. The difference between 2-1 (33%) and 5-2 (28%) is 5 whole percentage points. The difference between 15-1 (6%) and 30-1 (3%) is only 3 points even though the gap seems greater. Therefore, if you make a horse 15-1 in your line, you are only 3 percentage points off the mark if that horse should have been 30-1. For this reason, knowing that precision in the higher odds range is impossible, when your 15-1 goes off at 30-1 it is not a real overlay. Any horse whose odds you make at above 6-1 is therefore not a contender no matter what the post odds end up being. Nevertheless, if you choose to make a line, you should reflect carefully on the odds of the non-contenders because that will affect the odds of your contenders.
The same horseplayers who rebel against this line making procedure are willing to drive 70 miles to get an extra half point on a football wager. Lines for football and no lines for horse racing? That seems contradictory. The same line-making process is used when purchasing real estate, as it is by value stock pickers. But for some reason, horseplayers just want to play their most likely winner, even if the odds are so low that in the long run they could not possible grind out a profit on such horses.
The third point here is that when we make a personal odds line, we are simply translating probabilities into odds. Let’s look back at the same line:
Simply Run 7-2 (22%) (need 5-1 for the overlay)
Inhonorofjohnnie 4-1 (20%) (need 6-1)
House Quest 4-1 (20%) (need 6-1)
Allnightdance 5-1 (16%) (need 7.5-1)
Couple Whiles 6-1 (14%) (need 9-1)
The other two at 15-1. (6% each)
The total percentage adds up to 104%, slighty above my hundred percent requirement. I usually stay within 5% and make late adjustments to the precise 100% when I see track conditions and other late developments.
Looking back at this line, my most-likely winner, Simply Run, was going off as the heavy favorite, way below the 7-2. I had no problem resisting a wager. He was 7-2 on my line because of a certain potential weakness, especially in distance, so you should understand that I did not “pick” this horse, even if he was my most-likely winner. On the contrary, by incorporating his distance uncertainty in the odds, my line effectively eliminated him as a bet. C&X readers who played Simply Run because I had him on top missed the point of the line.
If he had won, I would not have been ruffled. By tallying results of lines, the handicapper validates correct handicapping procedures and discovers flaws in others. If too many of my most-likely winners are actually winning the races, then I would learn to adjust my lines downward. But this has not happened.
My second most likely winner was going off at 4-1, same as my line odds. No problem passing this one because there was absolutely no edge. I needed 6-1. The third and forth choices went off clearly below my line odds, so it was a no-brainer to pass them up.
My fifth choice, 6-1 in my line, was going off above 9-1. I was happy to play this one. The odds of 9-1 represented an edge. My logic on Couple Whiles was the fact that high-odds local horses like this one were historically successful in Claiming Crown turf races and that this trainer had a significant flat bet profit in all pertinent categories, including turf. Essentially, I was weighing the trainer and course factor [which follows my thinking in the previous computer model article] more than primary handicapping factors.
In deep stretch, Couple Whiles looked like a winner, only to get caught at the wire and lose by a scant neck to one of my non-contenders. Most C&X readers probably looked at this one as another loser and went on to the next race. I saw it as an example of how a personal odds line points me to horses of value, helps me pass numerous races where there is no value, and brings out the best in longshot betting, since most very live longshots are not one’s top choice. I also accumulated one more piece of evidence on a possible methodology based on integrating secondary factors.
Postscript. Even on occasions when a line is too difficult to make, the player who has gone through the process will be able to make better wagering decisions than the one who has no idea of the line making process. An example can be seen in the Beverly D, a race we did for a Stakes Weekend. There were two favorites in the race: Precious Kitten and Mauralakana. In my Stakes Weekend posting I gave a slight edge to Mauralakana. At the track, Precious Kitten went off at 1.9 while Mauralakana was a hard 2-1. The distinction could have led to a win bet, but even if it didn’t, it would say that Mauralakana was of greater value as a key horse in exotics than Precious Kitten.
ODDS/PERCENTAGE CONVERSION TABLE (Rounded off)
1-5 83% 3-1 25
1-2 67 7-2 22
3-5 62 4-1 20
4-5 55 9-2 18
1-1 50 5-1 16
6-5 45 6-1 14
7-5 41 8-1 11
3-2 40 10-1 9
8-5 38 12-1 8
9-5 35 15-1 6
2-1 33 20-1 4
5-2 28 30-1 3
Please note that there are abbreviations to the line making process but when first practicing, it is good to establish line odds for every horse, including non-contenders. I do it by starting with odds and then checking to see how much the percentages total. If the total is not very near 100%, I go back and make fair adjustments, making sure that the irregularity in my line is not caused by a flaw in my handicapping. Other line makers begin with percentages, asking for each horse: “If this race were run 100 times, how many times would this horse win?” Then they translate the percentages into odds. I use the first way but there are good arguments for the other way as well.
The big obstacle is that the line making process is time consuming, and that it cannot be forced. If you do not “understand” a race, making a line becomes an abstract illusion. When you make a line, you are feeding objective information into the mix and then collecting a subjecting impression. It is part science and partly artistic.
The line making process has helped me to play fewer races. If I see off the bat that I do not have enough useful information for a race (meaningful discoveries), I recognize that my line will come out flawed, bypass the line making, and pass the race.
On the other hand, the process has helped me to gain a sense of value, to the extent that I can now often “wing it” and bypass the line-making process, though I use it conceptually in deciding whether a horse is offering sufficient odds for a wager.
Just because a horse is 99-1 does not make him an overlay. In fact, more underlays are found in the higher odds ranges. That’s why anything above 6-1 in our line is a non-contender. The exception would be a race like the Kentucky Derby, where, with 20 horses, many with some chance to win, the most likely winner might be 4-1 on a line and 8-1 line odds might mean a horse is a contender. But Derby type races are rare.
C&X CAFÉ
ON RANDOM BETTING: CRAMER VALIDATED
Dan writes:
So I have in front of me the results of a table of 111,700 horses and the net return is 1.52 (for each $2 bet), so it looks like random betting would lose 24 cents on the dollar, while
betting favorites lose 18 cents on the dollar (net 1.64). Then how come the track take is
more like the favorite loss than the random loss?
Only a third of the money goes on favorites so where is that missing money going?
I guess I'm not up on what the track take really is?
Dan M. (former C&X subscriber)
Dan,
I have written about this in C&X. Amazingly, without your computer, I came up with a similar figure of approximately 24% loss just by tallying several samples of race results, not even large samples.
The reason the loss greater than the track take is that when you bet randomly, you are playing more underlays than overlays. In order to get a loss that equals the track takeout, you would have to play horses that were neither overlay nor underlay, or a balance of the two. Favorites are closer to that balance.
I’m surprised that this erroneous “percentage of random loss = the track take” has been repeated by so many experts and still has not been corrected, at least not in the public consciousness.
Does this answer your question?
Mark
CLASS VS. PARS
When we look at whether the horse is a proven loser or not, we also consider its class when it was running in Maiden and compare it to its non-maiden running lines? Without a set of par tables (which I have), it would be very hard to determine if the class he ran in at straight maiden was higher or lower than a low level claiming race. Please let me know, and I can’t IMAGINE how I could EVER have any more questions about this.
Ken
Cramer responds:
You have brought up a "philosophical" issue. Is class nothing more than speed? If you mix the class variable with the speed variable, then you are dissolving the class factor into a meaningless appendage.
There are handicappers who believe that class is really speed. My good friend Dick Mitchell and I disagreed on this one. I am convinced they are wrong. And even if they were mostly right, they are still wrong because speed is overbet and class is underbet at this moment in parimutuel history.
The animal behaviour expert I have mentioned in my upcoming novel is a real person, and he explained how fighting fish in Thailand, when they feel they are outclassed, will swim away from the fight. A horse that is outclassed (in the pecking order) will give up and run slower than he is capable of, so that his par time is useless. That's why horses dropping from straight maiden to maiden special weight often increase their speed rating notably.
The original handicappers had it right: class is an independent factor.
SOME THOUGHTS ON THE EARNINGS-PER-RACE FACTOR
I suppose William Quirin used this factor in his original research in the late 1970s because it is so easy to calculate, and certainly does reflect something about the class of a horse. When I tested the Quirin computer method, I found that the weakness of the earnings-per-race factor resided in the fact that it often reflected past class for a horse that was currently off form.
By using “earnings per race this year” you minimize the past class flaw. Furthermore, in today’s racing, the factor may be enhanced in its effect as a result of the increasing numbers of trainers who prime a horse to win after a layoff, thereby reducing the impact of current form.
I have been studying and sometimes betting the factor with the harness races for the simple reason that it is easier to apply. Eventually I plan on developing a way to use the factor automatically with the Tbreds. My recent discoveries apply a fuzzy logic to the factor, as opposed to the simply adding up the earnings. In other words, I establish an earnings benchmark, and look for situations where most horses are below the benchmark and only one or two are above it.
In harness racing, low purses that help me use simple numbers. Sometimes the horses are so cheap that my benchmark is $1,000 per race in earnings. I like to find a field where all the horses have earned less than a thou and only one or two are clearly above the thousand mark, with a significant gap.
Here’s an example of a race from Balmoral, where one horse emerged as a qualifier. (C&X readers are invited to send examples where the earnings-per-start worked or did not work, with your analysis of what went right or wrong. This will help us to come up with the right filters.)
Balmoral, 6 September, Race 14
1 Real Monet, 34 races, $12,641 earnings, approximately $300 per race
2 Art’s Ticket, 27 races, 34,396, $1,273 per race
3 You Rock, 5 races, 3,316 earned, much less than a thousand per race
4 Four Starzzzz EC, 31 races, 18,727 earned, a little more than $500 per race
5 Wake Up Mike, 30 races, 17,856 earned, just above $500 per race
6 Trigger Fish, 8 races, 2,808 earned, less than $500 per race
7 Time Alone, 30 races, 20,9499 earned , much less than a thousand per race
8 Western Gambler, 14 races, 13,264 earned, slightly less than a thousand per race
9 Bangkok Cruiser, 25 races, 25,377 earned, $1,015 per race
10 Highland Pride, 29 races, 30,096 earned, $1,034 per race
Without looking at any past performance factor other than earnings, you could also see in the past performances that Art’s Ticket showed two wins in his last twelve races, for purses of $9000 and $8910 whereas the next highest purse victory by any other horse in the field was for $7500. This tells us that Art’s Ticket did not accumulate his highest earnings per start from nibbling at place and show finishes, and that on his best night, he is capable of popping up and outclassing this field.
As often happens when searching for an automatic bet, two horses in this field were cause for confusion. The 9 and the 10 were ever so slightly above the $1,000 benchmark. Nevertheless there was a significant gap between those horses and the earnings of Art’s Ticket. I ask computer programmers among us what would happen if we had set the benchmark at a thousand. Can the computer be programmed to discern that Art’s Ticket qualified by the spirit of the law even if not by the letter of the law?
To me this is a key question when developing a computer model. This is vital to know because the essential reason for using these secondary factors which are underbet by the public is that it forces us to bet the horse no matter what else we see in the past performances.
Poor current form
For example, if we had handicapped this race in a “normal” way, we would have seen that Art’s Ticket had poor current form with several back-of-the-field finishes. Without the earnings factor, we would not have been prompted to look beyond the surface for positive nuances:
In his top running line, most recent race, he had shown some degree of new-found early speed, before eventually backing up badly. In that running line, he was in third place first over (sucking air while wide rather than drafting behind another horse) for at least two calls before calling it quits, while in his previous races, he had followed the field like a merry-go-round horse: no moves. His most recent race looked like a mild wake-up. But again, this nuance could have only seemed important in the context of the earnings ranking.
Another asset for Art’s Ticket was the fact that the second, third and fourth highest earnings horses, all of whom around the thousand-per-race mark, were in the poor outside posts 8, 9 and 10.
But the trick of using any mechanical factor such as earnings per race is to NOT depend on interpreting vague handicapping factors and instead to trust the stat we are using.
The fact that Art’s Ticket won this race is not as significant as his final payoff: $79.00. The lesson we learn is that the betting public scorns the average-earnings-per-race factor, and thus it becomes a viable factor. I am convinced that in the Tbreds, average-earnings-per-start-this-year is also an underbet factor. (I suppose that near the end of the year, we run a greater risk of embedding a past-class distortion in the statistic because we are going too far back in the horse’s cycles.)
No one is saying that the average-earnings-per-race by itself, with no filters, will be a deterministic factor, only that it should be respected as one of the wager-value factors. Currently it is not respected in the pari-mutuel financial market.
The message is simply that in any form of investment, lesser-used meaningful information carries more value, and that we will be following up on this factor.
Where I suspect that average earnings per start might be most useful is a scenario not so far from the harness races: cheap thoroughbred tracks where proven losers or dull horses are facing each other and none seem like a possible winner. In the past I was able to use the factor with moderate success at Charles Town. But now, I’m looking for long-term research.
NEW BOOK ON SYNTHETIC SURFACES
BETTING SYNTHETIC SURFACES
Conquering Racing’s Newest Frontier
By Bill Finley
Price: $24.95
Hardcover - 176 pages
Publisher: DRF Press
I have not had the chance to read the whole book yet but have gotten my hands on a couple of chapters, finding that Finley’s experience as a journalists enriches both the substance and the style. He offers numerous statistics to back up his assertions, and places the numbers within a crisp narrative style, backed by opinions from interviewed insiders.
For the first time this fall, the Breeders’ Cup will be run on a synthetic surface. We have seen with trainer stats that some trainers have improved with these new surfaces while others have declined. The universal speed bias has been obliterated but during certain strange interludes, sometimes perhaps because of moisture in the track and other times because of adjustments in the riders’ mass psychology, the speed bias returns with a vengeance, and then, poof, it’s gone.
Finley is the first author to confront in a methodical way the apparently necessary handicapping adjustments and angles, though he admits that the synthetic situation is still evolving, as tracks make adjustments after too many bizarre early results. There are parts of this book that are not pure fun to read, for example, the differences between Polytrack, Cushion Track and Tapeta, and it seems like a daunting escalation of handicapping variables.
If you haven’t read Bill Finley, you’re in for a treat. He handles the English language with craft and precision. He has been covering horse racing for a quarter of a century. He’s also a fan, and that adds flavor to his writing; he’s visited 67 North American race tracks and never missed a Kentucky Derby.
ART KAUFMAN
With Art Kaufman's recent passing, the collective sadness of those who knew
him is enough to move the tides. Art had to use the name Tomlinson because his mutual fund colleagues frowned on his interest in horse racing, though in today’s economy, the Tomlinson ratings are performing better than the stock market.
I used his Tomlinson ratings religiously and one day was moved to phone him to thank him for the best handicapping tool I had ever used. During that period in the early and mid-1990s, before the Tomlinson ratings were published by the DRF and before trainers used the ratings for spotting their horses, this was my automatic bet, my source for signers, and my way of zooming in on one single automatic method.
We got to know each other well and I learned the definition of generosity and warmth. I accompanied Art on his Euro racing tours, which seemed like a business from the outside but was a mere labor of love for Art. I knew this because whenever Art had to make a decision between potential profit for the business or comfort and pleasure for his customers, he chose that latter. He won by breaking even.
Art brought happiness to anyone near him, including my elderly and frail Aunt Ada,
whose heart he warmed with his optimism. Ada was already in her late 80s at the time, and every time she pronounced the word “Arthur” it probably added a few days to her life. That’s because he made all of us feel like truly important people.
For one Breeders' Cup at the Albany OTB, Art noticed that I could not interest my then 14-year son in accompanying us. So Art handed him a racing form and said, "If you are willing to study this, I'll give you $2 to bet for each race ... but only if you study!" My son decided to come along with us, read the form carefully after some instruction from Art, and collected on Reraise in the Sprint.
But when the day was over, my son realized that playing the horses for real required an extraordinary amount of study and persistence, and he decided that it was too big a task for him. But thanks to Art my son had learned to respect the avocation of his old man.
Our condolences go out to his wife Jackie and the rest of the family, and also to the magnificent people who knew and admired him.
GEORGE BANGO
If you’ve played Turf Paradise or Canterbury Park, then sooner or later you would have had to handicap a George Bango horse. Eventually you would have noted that many of his horses were overachievers. Bango operated a small hands-on stable. His stable was neat and clean, his grooms worked in as pleasant a situation as possible, and he loved his job. He was the opposite of a “designer trainer”. He was accessible and always ready with a good one-liner for the media and for any fan who asked a question.
Racetrack managers saw him coming and knew that he’d be griping lyrically about one thing or another, the condition book or the track surface, and after he spoke you wished you had a tape recorder to catch his newest quip.
“You wait and see,” he once said after the Claiming Crown post position draw about one of his longshots. “Superman Can is gonna kick ass. Superman Can can!” And two days later the horse went on to win at huge odds.
Trainers like Bango hold the game together and prevent it from becoming just one more consumer product. Bango was serving as Vice President of the Board of Directors of the Arizona Horsemen’s Benevolent and Protective Association.
George Bango has passed away after a long bout with cancer. Our condolences to his family, his many friends, and his longtime groom Monserrat.
CONTENTS
EDITORIAL
CAN WE CREATE A TIME-SAVING COMPUTER METHOD?
SPIRIT ONE: WHO SAYS WIN OVERLAYS ARE INCREASINGLY DIFFICULT TO FIND?
LINE-MAKING WORKSHOP
C&X CAFÉ:
On Random Betting: Cramer Validated
Class vs. Pars
SOME THOUGHTS ON THE EARNINGS-PER-RACE FACTOR
NEW BOOK ON SYNTHETIC SURFACES
ART KAUFMAN
EDITORIAL
C&X has a long history, having begun as the C&O Report in the early 1990s, thanks to the encouragement and contribution of Bill Olmsted, who was the O. Bill reached a crossroads and decided to move on, even though the publication had been a success. The X today refers to X-rated handicappers.
As anyone who follows the publishing business can guess, print publications are an endangered species, especially newsletters. Dave and I have continued to put out the C&X Report because of mellow inertia: it’s something we like to do, but it’s not a business success. During crunch time, I have often thought of phasing out the publication.
On the other hand, my love for racing keeps growing, and a byproduct of doing C&X is that I am forced to remain a student of the game; this publication could not have survived if we had been recycling old material.
As we near the end of another cycle of issues, I need to decide whether or not I should continue, and if so, in what form and format. For sure, something does have to change, and if it does, we need to make sure it’s for the better.
As you will see from this issue, a continuation of C&X might drift in the direction of more interactive involvement of readers, perhaps some sort of group project.
The lead article of this issue will give you an idea of where my mind is meandering.
Please take note that the next Stakes Weekend will be the Arc de Triomphe, the first Sunday in October, and I’ll try to throw in another race from the USA on the Saturday of the same weekend. If you have not seen my own general and eclectic website before, take a look. www.altiplanopublications.com
Within a few days there should be some new stuff posted there.
mc
CAN WE CREATE A TIME-SAVING COMPUTER METHOD?
During this period I am mulling over what to do with a possible next cycle of C&X, I am trying to find a formula that will build on what we’ve done, a new level that approaches more tangible results on our return on investment.
Ken, one of our dedicated readers and a studious and smart horseplayer, has echoed my sentiments that we need a time-saving short-cut. Of all the people I’ve ever known, Ken is the only one who does not know how to be lazy. He’s a workaholic and he’s ruthlessly detailed in analyzing the races and studying the literature. He’s also the type of devil’s advocate that helps us refine our thinking.
Wise men like Ken are aware that no being from our planet, even the most successful one, can carry on a daily wagering routine without eventually losing concentration, playing catch-up in the information gathering and sorting process, and eventually burning out.
I have been saying something similar for years but hearing it from Ken has convinced me to act more boldly. I’ve seen too many good players reach their half-life and then wilt away from the game.
Before getting to the meat of this article, allow me to indulge in an anecdote that will explain viscerally why the time factor is tantamount. It was the third day of a tournament at Sam’s Town. Mike Helm and I were playing as partners. We had come off a $1,000 prize the previous day for highest profit of the day, thanks to one of my turf-sire bets and one of Mike’s first-time starters, both based on a SECONDARY FACTOR (remember that for later). It was not the first time we had won such a prize. Since the entry fee for the tournament was $500, we were assured of doubling our money, even if we tapped out.
We were in close pressing position, within the top six or seven leaders, but two days and nights of handicapping and decision making were already taking their toll, and going to sleep at 2pm after hours of intense handicapping is not the way to get the sound sleep one needs for a fair level of concentration the next day.
It is especially discouraging to try to sleep only 5 hours before sunrise, when you have not made any handicapping discover for the next day’s cards. The mind is too stimulated to allow for deep sleep. My biggest enemy was not bad handicapping but a lack of that most precious commodity called time.
The next day, groggy from workaholism, we were hard pressed to come up with nine of the required ten bets. As the race cards evoloved, still looking through my hazy eyes at the pps, I came up with a “maybe horse” four our tenth bet. (The tournament required the player to bet all the bankroll each day, ten percent of bankroll on each race. Very bold indeed. Too bold according to the math of Dick Mitchell.
The horse I had found was a cheap claimer stretching out into a route race that looked like the lone speed but had little else in its favor. By the time I had digested my thoughts and had found a couple of other thin reasons for our playing over a $1,000 on this horse, I went up to the special betting tables for tournament players. The horses were at the gate and beginning to load. I handed in the bet ticket.
The lady told me that it was too late for a bet. Only two of the 11 horses had loaded the gate.
“Sorry,” she said, “tournament wagering is closed after the first horse loads.”
Needless to say, our horse took the lead and went wire to wire. He paid over $70 to win. That one horse would have put us in first place. We had missed the opportunity not because of bad handicapping but from a lack of time.
From then on, we were so psychologically devastated, partly from the fatigue of handicapping for so many hours, that it was all downhill, and we tapped out on the last day. It was the most depressing way to make a hundred percent profit in the tournament.
With a 200% profit for the totality of my tournament play, spanning over seven tournaments, it was time to quit. I resolved to never again play in a tournament unless I had discovered the type of automatic bets that would not corrode my time.
Andrew Beyer, a successful workaholic, once said (I’m paraphrasing) that cutting corners, even minor time-saving behavior could mean the difference between a consolation and a huge pick six payoff, just by missing one piece of information.
But for most of us, a comprehensive approach is unsustainable and most of us cannot go at it day in and day out without eventually burning out. Beyer is an exception. He’s still at it. But he is also able to take a vacation, get on his bicycle, and take up a completely different challenge.
On the other hand, if one does the work up front, creating a streamlined model that extracts what is essential, there is a way to bet intelligently and successfully. We want a computer model (or one that can be done by pencil and paper) that will land us on enough longshot winners to sustain a long-term profit.
Today, it’s just as possible to find a longshot winner as it was decades ago when much less information was available. This is contrary to what many experts have alleged. They say that it’s increasingly difficult to outperform the public. I say they are wrong. The average payoff of betting favorites may have declined to some degree but the percentage of favorites has not increased and the number of longshot winners has not declined. This is because the public’s weak point today is not a lack of information but information overload.
Now please follow my logic, because, before I present the hypothesis for a computer model, you need to understand the foundation. Rather than “cutting corners”, why not synthesize, which means extracting what is truly essential, which is really a small part of the available information. That’s not cutting corners, it’s cutting out the fat. We could call it “lean handicapping” but the food industry has already appropriated the term.
Helping us in our endeavor is the research of the very people who believe that what we want to do is impossible. People like my friend Barry Meadow, who has shown with incredibly huge research samples that none of the major handicapping factors yield a positive return on investment.
Meadow’s conclusion: there is no mechanical method.
My conclusion: there is no mechanical method based on primary handicapping factors.
Do you see where I’m going? By cutting out the text book factors like speed, those factors with negative wager value, we save time without losing depth. My research has shown that winners in races where they have not changed distance or surface (to eliminate other variables) increase their Beyer fig by an average of nearly 9 points. Amazing but true. You win by projecting change and not predicting continuity.
So, primary or conventional factors which predict continuity can be diminished in importance and we can base whatever computer model we can construct on contrarian factors, some of which are recognized as “secondary” factors by the experts and others which remain under the radar of public consciousness.
All right, so that’s the foundation. Now comes the invitation. Even if I have not had the pleasure of communicating with many of our readers, I feel we have the nucleus of a potential community, what is referred to in academia as “collegiality”. We should also understand that a community is based to a certain extent on sharing but not so much as to deny individuality, which is important for maintaining the dynamism of the whole. A real community finds the right balance between collective consciousness and rugged individuality. This balance could help us come up with the right model for either a computer program or pen-and-paper method.
Over the years, C&X readers have participated by asking good questions, sending opinions, sending criticism, and even sending independent research. There are some talented researchers among us, some of them skilled in computer programming. My call is to you, that some of you volunteer to help out with this project and that it be shared with our community, because, as an old saying goes, “If you can’t share it, you don’t have it.” This is a potential format for continuing C&X. Certainly our collective analytical skill is superior to that of any individual among us, as proven empirically in the book The Wisdom of Crowds, which also recognizes that the group, even the most sophisticated group, can be overcome on occasions by groupthink, thus enabling an informed minority of one (or few) to find the right path.
My first step is to propose a foundation for a potential computer program whose purpose is to save lots of time without cutting corners. Much of what I include in this proposal is based on previous research, of mine or of colleagues, with other parts based on lesser but still important anecdotal evidence.
As I’ve noted, perhaps, in looking for a formula for another cycle of C&X (because intellectually I need a higher level or I won’t see the reason for continuing) we could engage in this group project in search of a time-saving computer method, pooling our diverse types of knowledge and awareness.
My awareness is conceptual but not technological. The more technically oriented among us are invited to study this model and reflect on whether it will fit in a “program”. In the end, we may discover that a computer is not even necessary but that our model can be done with pen and paper, still saving time from comprehensive handicapping but gaining in pari-mutuel force. So here goes.
Concept:
One of the dilemmas of handicapping each race is identifying which factors should take precedence. So the computer must choose which three factors among the many will be used. We extract the essential, and a three factor model allows us to triangulate between the factors.
Once the computer has selected each of the three factors, if more than three horses qualify by a particular factor, the computer eliminates that factor from consideration. (The three-horse parameter can later be modified according to number of horses per field:
Between 7 and 9 horses, no more than three qualifying horses for a factor to kick in.
Between 10 and up, no more than four qualifying horses for a factor to kick in.
Between 5 and 6 horses, no more than two qualifying horses.
Once the method of interaction between the three factors is determined, we will then decide whether the model fits more for straight win bets of for exotics. I suspect the latter, for in my own pen-and-paper experimentation with this model, I have not been able to whittle it down to a single horse. Not yet.
(2) Once the computer designates the three most significant factors, the next step is triangulation. Horses that qualify by all three factors get priority over those qualifying by two. A horse qualifying under only one factor is NOT eliminated, but gets less priority.
However, a horse may only qualifies under one factor BUT if this horse is an only qualifier under this particular factor, then this may count more than a horse that is a double or triple qualifier among three horses in a category.
Here are the factors I have selected as either “secondary” or contrarian.
Trainer 1. Horse shows at least one profitable category in DRF trainer specialty stats for this particular race, and this profitable caterory must show at least 18% winners, so that we know that the positive roi did not come from a fluke longshot. Remember that if more than three horses qualify by this parameter, the computer will reject the parameter and move on to the next one.
Class. Horse must not have lost more than once AT TODAY’S CLASS LEVEL OR VS. LESSER (a proven loser). This goes for maiden special weight, maiden claiming, claiming and the allowance “non-winners of” levels. Races at other levels, such as starter allowance, mid-level stakes or graded stakes do not apply to this rule because, each for a different reason, they are too difficult to objectify. (I have anecdotal evidence that stakes races, for example are often at a different level than their class designation, because of poor fields.) If more than three horses in the field are not proven losers, then the computer will reject this parameter and move on to the next.
Course. Horse has won 30% or more races at this track/surface. Easy to notice: it’s right in the performance box. If more than three horses qualify, the computer rejects this parameter.
Trainer 2. Trainer has a 18% or more overall win record this year (or this year plus last year if we are within the first two months of the calendar year). Again, if more than three horses meet this criterion through their trainer, the computer rejects this factor.
Surface (“turf” or “dirt-fast”). Horse has won 30% or more on the surface of today’s race.
Overachiever. Horse shows at least three running lines in pps in which it has EITHER won at 5-1 or up, finished in the money at 9-1 or up, or a combination of both. Or, horse shows two such finishes in its last four races. Once more, the computer will reject this factor if more than three horses qualify as overachievers.
I say the computer, but a smart programmer may be able to design a pen-and-paper way to use these six parameters. They are easy to identify with a quick scan of the pps.
The computer is now supposed to select the three categories in which there are qualifiers. I can imagine using two categories as well, if none or the others qualify, and I can also envision using four. If we have five or six working factors, then we either (a) have to choose the best three from them, or (b) use them all with some unique cross-tallying method. Even so, the concept of triangulating, or balancing three sides, needs to be explored and refined.
Now here, computer programmers, please heed: we are looking for an interaction of factors that goes more deeply than simply counting points. We want something with balance of strength, which means, perhaps, that qualifying by more than one factor is equally important as being an only qualifier under one factor.
If only one factor emerges, we still will work with that factor, considering that the rest of the race is a hodgepodge of factors, a chaos race, and we want to extract the one thin slice that makes one horse truly different from the others.
Anywhere beyond a single parameter may be more applicable to exotics than to win betting, but that’s only an initial impression. Since these are all secondary factors, it may be possible that one horse towers above all the rest and still goes off at 10-1.
The dilemma.
I have been studying this methodology anecdotally with both American and French races. In races where there are three qualifying factors, sometimes one factor trumps the others, and other times there’s a complex interrelationship between these factors.
About a month ago, a French TV documentary crew accompanied me to Chantilly race course on a pre-scheduled filming. My back was to the wall, as these were lesser-of-evil races. The filmmaker knew nothing about handicapping, so if I lost, he would assume, anecdotally, than it was all hype and beating the races was impossible. If I won, he would at least consider that human thoughtfulness, analysis and creativity can be a deciding factor in making a profit at the races.
I zoomed in on one of the race where it looked as if the payoff would be good, and I used this method and triangulated three factors, trainer 2, class and course. My class factor here was different from the above model because the class levels are not the same as we know them, so I used “average earnings per race for the year” and required 3,000 per race in order to qualify. I played the French version of the superfecta, which is really a 4 horse quinella: you win no matter what the order.
No horse was an only qualifier under any of the factors. However there were some horses that qualified by two factors and others by one. I wheeled the dual qualifiers with single qualifiers. In other words, if two single qualifiers came in the winning combo, I wouldn’t collect. One combination came in, and that was thanks to a longshot single qualifier that qualified only by the course criteria, a horse I would have eliminated based on comprehensive handicapping. This showed me that the method picks horses that the public is less likely to pick. I thus salvaged my reputation with a generous payoff in front of the camera.
Let me highlight the fact that if I had done comprehensive handicapping of the same race, I would have eliminated my single qualifier and ended up with 3 of 4. Since that afternoon I have noticed that my automatic criteria pick in-the-money finishers that I would not have used in exotic bets based on conventional handicapping. And yet, this unconventional method is based, fundamentally, on meaningful statistics
But to show you the dilemma, on another of these mechanical bets, more recently, I triangulated three factors but one of the factors (trainer 2) completely trumped the other two, and the payoff was huge. In other words, the question our computer needs to answer is: when does one contrarian factor eclipse all the others and when is a triangulation of factors called for?
I’ve practiced this hypothesis with American races for a much longer time. The French experience only corroborates my American records: but again, the big question that we want the computer to decide (if possible) is: when will one factor trump the other two.
Once the computer discards the non-factors, it has to make a decision as to whether or not all three qualifying factors (or both qualifying factors if there are only two) should be integrated or if one factor should drown out the other(s).
I have some ideas as to how to do this, but they are based on anecdotal evidence.
I don’t believe any other computer method is close to the area I’m covering. I have eliminated the three most classic factors: speed, pace and current form, on the notion that they are much too overbet by the public.
I also have a relatively simple method for the computer to decide which factor(s) should be used, based on the need for one to three horses to qualify within a factor, no more and no less. If four horses qualify by the horse-course factor, for example, then we eliminate that factor. And on the other end, if no horses qualify, then more obviously it becomes a non-factor.
I did not come up with this idea in a vacuum. I spent long hours with Ed Bain, watching him play his automatic trainer method, and he showed me that often if there were more than two qualifying trainers, the winner would come from elsewhere and not from the qualifiers.
One of the big advantages of this hypothetical structure for a computer method is that I have eliminated the need for choosing a running line.
My research on trainer and class shows that these two factors have validated wager value.
The course/surface seems anecdotally valid.
So please contact me if you have computer experience and think there’s a relatively simple way to come up with a program. My biggest concern is the time it would take for data entry. Would that data entry time be better spent in just doing it by pen and paper?
Even if we end up with no computer system, we may be on to a time-saving mechanical way of handicapping that does not sacrifice depth of analysis.
Sorry for a certain degree of repetition. I want to make sure this different approach is clear to all.
SPIRIT ONE: WHO SAYS WIN OVERLAYS ARE INCREASINGLY DIFFICULT TO FIND?
During the past decade, a recurring cliché in the pari-mutuel milieu has emerged: because so much information is now available, it is increasingly hard to find an overlay in the win pool. Therefore, we need to look to the exotic pools as our only hope for a winning return on investment.
First and foremost, it should be noted that when we have more information available, it becomes increasingly difficult to manage that information. The C&X Report has often dealt with this dilemma, offering alternative ways of sifting through information and separating the good stuff from what is banal. “Thin slicing”, for example, or picking out one particular secondary factor that trumps all the others, is one of our approaches.
In the case of the Arlington Million, all the information one needed to know was right in the past performances. In the C&X Stakes Weekend website analysis we noted that there were “comparables” that could be used when appraising the wager value of the three Euro horses: Archipenko, Mount Nelson and Spirit One. The legitimate but overbet favorite, Archipenko (6/5) had defeated a horse named Balius by a length and _ in April. Spirit One had defeated Balius by 3 in March, and Balius was six lengths better than the rest of the field so his race was not a clunker.
Then we saw that Mount Nelson had been defeated by a horse named named Athanor that had been defeated by the French filly Cicerole, suggesting that Mount Nelson’s competition was not the classiest. So my comment was that even if these three Euro horses could be rated in a similar bracket, because of the hype, Spirit One figured to have the highest odds. We also learned, directly from the DRF workout comments that Spirit One had looked exceptionally fit in his Arlington workout. And finally, perhaps less evident in the racing pages was the fact that Spirit One’s early speed, a negative factor in the European bias scene, would become positive in American turf racing. I have been writing about this factor for a number of years: Euro horses moving from against-the-bias racing to with the bias.
Perhaps less evident to the American betting public was the fine record of trainer Philippe Demercastel, and the personal anecdote I had about the fact that a top American trainer had wanted Ioritz Mendizabal to ride for him.
Okay, so we had some privileged information, but any handicapper knowing the value of information, knowing that information equals pari-mutuel power, could have seen that Spirit One had been the pace setter in a key race whose top three finishers had come back to win, and it was clear from DRF articles that the winning horse in that field, Duke de Marmalade, towered over the Arlington Million contenders. Spirit One had finished only 2 _ behind the Duke.
The result of the Arlington Million was not deterministic. On another day, if Archipenko does not get boxed in by Murtagh on Mount Nelson, he might catch Spirit One.
The point is that there was an odds differential between the three Euro horses that should not have been. With Archipenko at nearly even money, Mount Nelson’s odds were pushed up to a hard 7/2. But how does one explain the 13-1 odds on Spirit One?
Simply stated, the betting public got it wrong, and this betting public, deemed so accurate by the naysayers (those who say that win overlays are hard to come by) was not even close.
Fortunately, we sometimes have a way of measuring these things. The leading handicapper in the Paris-Turf, the French version of the DRF, listed his choices for the Arlington Million as Archipenko, Mount Nelson, Einstein, Stream Cat and only then Spirit One, so there was no national preference in favor of the French horse among the experts. But in their separate pool, French players made Spirit One 6-1.
The occasional existence of separate pools proves that betting publics can still get it wrong, even in the information age. On some of these rare occasions, I have the luxury of choosing my betting pool. In the Beverly D the French public had a similar view of the eventual winner Mauralakana (the horse I named as most-likely winner) but Communique, one of my contenders, was 30-1 in France and only 12-1 in the USA, so I used Communique with Mauralakana in one of my quinella bets and got back three times the odds that an exacta box paid in the USA.
Until now, you might argue that the pari-mutuel inefficiency is restricted to foreign horses, but this is not the case. First, the American public did make Archipenko the heavy favorite, so there was absolutely no prejudice against the foreigners. There was, however, a tendency to follow the hype: what I’ve often called “the publicity horse”.
Archipenko was the publicity horse, and that usually has an impact on the odds. This leads me to a second area where all the information in the world will still not make the betting public efficient or accurate. In any stakes races where stars are participating, the hype has an extraordinary influence on the odds.
I had made this point on the C&X Stakes Weekend website, for the March 29 Bonnie Miss Stakes at Gulfstream, where I noted that the two favorites would draw excessive action because of the classy company they kept. Each of those two had negative trainer stats. So I had no privileged information beyond what was found in the DRF.
Hype cannot overcome the combination of bad pace scenario and negative trainers stats, so I went with She’s All Eltish on top (a hard 4-1) and Robbie’s Girl as the place horse, and that’s the way they finished in the exacta. It was not so much choosing the top two horses but recognizing an inefficiency in the pari-mutuel market because of the presence of two publicity horses.
Robbie’s Girl indicates yet another facet of betting public inefficiency. Robbie’s Girl had finished in the money at high odds on several occasions, and did so again in the Bonnie Lass. Certain horses, call them “un-hype” horses, have a record of being consistently underbet by the public. I call these horses “overachievers”. They are horses that regularly win or finish in the money at high odds. I had identified Robbie’s Girl as an overachiever in the place hole. Overlays are still found in the straight pools when overachievers can be identified.
Yet another opportunity for overlays in the win hole is found in the realm of shippers. The natural human tendency is to embrace what is already known. Given the choice of an underlay restaurant that is already known or a potential overlay restaurant on the same corner that may or may not end up delighting, most people fear venturing into new territory and willingly accept the underlay.
Shippers are the new guy in town and some of them may offer overlay potential, if the player does a little homework on the trainer.
These are just a few areas where win overlays are still to be had. The betting public may have become more efficient in picking low-priced horses but in the middle or upper odds ranges, live horses may remain untapped by even the smartest players.
LINE-MAKING WORKSHOP
We’ve gone through many issues of C&X with no line-making workshops and I fear some of the newer subscribers may scratch their heads when I make certain references to a “personal odds line”. Therefore, I’d like to revisit the process with one of the races we did for the Stakes Weekend. Those of you who already are aware of this process cannot be hurt by this refresher. Let’s look back at the Claiming Crown Tiara on the turf. Here you see the line I posted on the Stakes Weekend site:
Simply Run 7-2 (need 5-1 for the overlay). 5-1 ML
Inhonorofjohnnie 4-1 (need 6-1). 4-1 ML
House Quest 4-1 (need 6-1). 7-2 ML
Allnightdance 5-1 (need 7.5-1). 5-2 ML
Couple Whiles 6-1 (need 9-1). 6-1 ML
The other two at 15-1.
As you can see, this was a deeply contentious race, and even the other two horses not included as contenders were given a chance as part of the line.
First, please notice that my line odds are based on a 100% probability that there will be a winner: no more, no less. Program odds, what we call the morning line, use approximately 120 percentage points. By adding percentage probability for each horse they are lowing the odds of each horse, (a) in order to reflect the track takeout, which also essentially lowers the odds of each horse, and (b) to encourage more betting by creating an illusion that toteboard odds are “fair”. More horses will “seem” like overlays when their odds are compared to a 120% morning line, and fewer horses will really be going off above their line odds when the a personal odds line based on 100% is set side to side with the morning line odds.
Second, please note that I have two numbers. My first number is my perceived intrinsic probability of the horse to win, based on the 100 percentage points available for all horses. These odds are based on a balance between objectivity (stats, figs, etc.) and subjectivity, my (or your) personal analysis and synthesis of that objective information. It is possible, but much more complex than you can imagine, to make a truly objective line. If only one factor mattered, say trainer percentage, then if the trainer win rate of a horse is 20% I could give the horse an automatic 4-1, the equivalent of 20%. But as we all no there are too many other stats, some of which contradict each other, and thus, the line maker needs to make subjective appraisals.
My second number is what I need in order to make a play. What I need is 50% more than the line odds. Those 50% refer to a betting advantage. For example, if I make a horse 2-1 and he’s going off at 2-1, that means no advantage. The only way to make a long-term profit at the races is to play only when you have an advantage. If your line is relatively accurate (it does not have to be perfect), then you should be comfortable in not playing any horse that is not clearly above your line odds.
Any horse that you make above 6-1 is considered by you as a non-contender. In the higher-odds range there is less of a difference in percentage points and therefore a greater chance for error. The difference between 2-1 (33%) and 5-2 (28%) is 5 whole percentage points. The difference between 15-1 (6%) and 30-1 (3%) is only 3 points even though the gap seems greater. Therefore, if you make a horse 15-1 in your line, you are only 3 percentage points off the mark if that horse should have been 30-1. For this reason, knowing that precision in the higher odds range is impossible, when your 15-1 goes off at 30-1 it is not a real overlay. Any horse whose odds you make at above 6-1 is therefore not a contender no matter what the post odds end up being. Nevertheless, if you choose to make a line, you should reflect carefully on the odds of the non-contenders because that will affect the odds of your contenders.
The same horseplayers who rebel against this line making procedure are willing to drive 70 miles to get an extra half point on a football wager. Lines for football and no lines for horse racing? That seems contradictory. The same line-making process is used when purchasing real estate, as it is by value stock pickers. But for some reason, horseplayers just want to play their most likely winner, even if the odds are so low that in the long run they could not possible grind out a profit on such horses.
The third point here is that when we make a personal odds line, we are simply translating probabilities into odds. Let’s look back at the same line:
Simply Run 7-2 (22%) (need 5-1 for the overlay)
Inhonorofjohnnie 4-1 (20%) (need 6-1)
House Quest 4-1 (20%) (need 6-1)
Allnightdance 5-1 (16%) (need 7.5-1)
Couple Whiles 6-1 (14%) (need 9-1)
The other two at 15-1. (6% each)
The total percentage adds up to 104%, slighty above my hundred percent requirement. I usually stay within 5% and make late adjustments to the precise 100% when I see track conditions and other late developments.
Looking back at this line, my most-likely winner, Simply Run, was going off as the heavy favorite, way below the 7-2. I had no problem resisting a wager. He was 7-2 on my line because of a certain potential weakness, especially in distance, so you should understand that I did not “pick” this horse, even if he was my most-likely winner. On the contrary, by incorporating his distance uncertainty in the odds, my line effectively eliminated him as a bet. C&X readers who played Simply Run because I had him on top missed the point of the line.
If he had won, I would not have been ruffled. By tallying results of lines, the handicapper validates correct handicapping procedures and discovers flaws in others. If too many of my most-likely winners are actually winning the races, then I would learn to adjust my lines downward. But this has not happened.
My second most likely winner was going off at 4-1, same as my line odds. No problem passing this one because there was absolutely no edge. I needed 6-1. The third and forth choices went off clearly below my line odds, so it was a no-brainer to pass them up.
My fifth choice, 6-1 in my line, was going off above 9-1. I was happy to play this one. The odds of 9-1 represented an edge. My logic on Couple Whiles was the fact that high-odds local horses like this one were historically successful in Claiming Crown turf races and that this trainer had a significant flat bet profit in all pertinent categories, including turf. Essentially, I was weighing the trainer and course factor [which follows my thinking in the previous computer model article] more than primary handicapping factors.
In deep stretch, Couple Whiles looked like a winner, only to get caught at the wire and lose by a scant neck to one of my non-contenders. Most C&X readers probably looked at this one as another loser and went on to the next race. I saw it as an example of how a personal odds line points me to horses of value, helps me pass numerous races where there is no value, and brings out the best in longshot betting, since most very live longshots are not one’s top choice. I also accumulated one more piece of evidence on a possible methodology based on integrating secondary factors.
Postscript. Even on occasions when a line is too difficult to make, the player who has gone through the process will be able to make better wagering decisions than the one who has no idea of the line making process. An example can be seen in the Beverly D, a race we did for a Stakes Weekend. There were two favorites in the race: Precious Kitten and Mauralakana. In my Stakes Weekend posting I gave a slight edge to Mauralakana. At the track, Precious Kitten went off at 1.9 while Mauralakana was a hard 2-1. The distinction could have led to a win bet, but even if it didn’t, it would say that Mauralakana was of greater value as a key horse in exotics than Precious Kitten.
ODDS/PERCENTAGE CONVERSION TABLE (Rounded off)
1-5 83% 3-1 25
1-2 67 7-2 22
3-5 62 4-1 20
4-5 55 9-2 18
1-1 50 5-1 16
6-5 45 6-1 14
7-5 41 8-1 11
3-2 40 10-1 9
8-5 38 12-1 8
9-5 35 15-1 6
2-1 33 20-1 4
5-2 28 30-1 3
Please note that there are abbreviations to the line making process but when first practicing, it is good to establish line odds for every horse, including non-contenders. I do it by starting with odds and then checking to see how much the percentages total. If the total is not very near 100%, I go back and make fair adjustments, making sure that the irregularity in my line is not caused by a flaw in my handicapping. Other line makers begin with percentages, asking for each horse: “If this race were run 100 times, how many times would this horse win?” Then they translate the percentages into odds. I use the first way but there are good arguments for the other way as well.
The big obstacle is that the line making process is time consuming, and that it cannot be forced. If you do not “understand” a race, making a line becomes an abstract illusion. When you make a line, you are feeding objective information into the mix and then collecting a subjecting impression. It is part science and partly artistic.
The line making process has helped me to play fewer races. If I see off the bat that I do not have enough useful information for a race (meaningful discoveries), I recognize that my line will come out flawed, bypass the line making, and pass the race.
On the other hand, the process has helped me to gain a sense of value, to the extent that I can now often “wing it” and bypass the line-making process, though I use it conceptually in deciding whether a horse is offering sufficient odds for a wager.
Just because a horse is 99-1 does not make him an overlay. In fact, more underlays are found in the higher odds ranges. That’s why anything above 6-1 in our line is a non-contender. The exception would be a race like the Kentucky Derby, where, with 20 horses, many with some chance to win, the most likely winner might be 4-1 on a line and 8-1 line odds might mean a horse is a contender. But Derby type races are rare.
C&X CAFÉ
ON RANDOM BETTING: CRAMER VALIDATED
Dan writes:
So I have in front of me the results of a table of 111,700 horses and the net return is 1.52 (for each $2 bet), so it looks like random betting would lose 24 cents on the dollar, while
betting favorites lose 18 cents on the dollar (net 1.64). Then how come the track take is
more like the favorite loss than the random loss?
Only a third of the money goes on favorites so where is that missing money going?
I guess I'm not up on what the track take really is?
Dan M. (former C&X subscriber)
Dan,
I have written about this in C&X. Amazingly, without your computer, I came up with a similar figure of approximately 24% loss just by tallying several samples of race results, not even large samples.
The reason the loss greater than the track take is that when you bet randomly, you are playing more underlays than overlays. In order to get a loss that equals the track takeout, you would have to play horses that were neither overlay nor underlay, or a balance of the two. Favorites are closer to that balance.
I’m surprised that this erroneous “percentage of random loss = the track take” has been repeated by so many experts and still has not been corrected, at least not in the public consciousness.
Does this answer your question?
Mark
CLASS VS. PARS
When we look at whether the horse is a proven loser or not, we also consider its class when it was running in Maiden and compare it to its non-maiden running lines? Without a set of par tables (which I have), it would be very hard to determine if the class he ran in at straight maiden was higher or lower than a low level claiming race. Please let me know, and I can’t IMAGINE how I could EVER have any more questions about this.
Ken
Cramer responds:
You have brought up a "philosophical" issue. Is class nothing more than speed? If you mix the class variable with the speed variable, then you are dissolving the class factor into a meaningless appendage.
There are handicappers who believe that class is really speed. My good friend Dick Mitchell and I disagreed on this one. I am convinced they are wrong. And even if they were mostly right, they are still wrong because speed is overbet and class is underbet at this moment in parimutuel history.
The animal behaviour expert I have mentioned in my upcoming novel is a real person, and he explained how fighting fish in Thailand, when they feel they are outclassed, will swim away from the fight. A horse that is outclassed (in the pecking order) will give up and run slower than he is capable of, so that his par time is useless. That's why horses dropping from straight maiden to maiden special weight often increase their speed rating notably.
The original handicappers had it right: class is an independent factor.
SOME THOUGHTS ON THE EARNINGS-PER-RACE FACTOR
I suppose William Quirin used this factor in his original research in the late 1970s because it is so easy to calculate, and certainly does reflect something about the class of a horse. When I tested the Quirin computer method, I found that the weakness of the earnings-per-race factor resided in the fact that it often reflected past class for a horse that was currently off form.
By using “earnings per race this year” you minimize the past class flaw. Furthermore, in today’s racing, the factor may be enhanced in its effect as a result of the increasing numbers of trainers who prime a horse to win after a layoff, thereby reducing the impact of current form.
I have been studying and sometimes betting the factor with the harness races for the simple reason that it is easier to apply. Eventually I plan on developing a way to use the factor automatically with the Tbreds. My recent discoveries apply a fuzzy logic to the factor, as opposed to the simply adding up the earnings. In other words, I establish an earnings benchmark, and look for situations where most horses are below the benchmark and only one or two are above it.
In harness racing, low purses that help me use simple numbers. Sometimes the horses are so cheap that my benchmark is $1,000 per race in earnings. I like to find a field where all the horses have earned less than a thou and only one or two are clearly above the thousand mark, with a significant gap.
Here’s an example of a race from Balmoral, where one horse emerged as a qualifier. (C&X readers are invited to send examples where the earnings-per-start worked or did not work, with your analysis of what went right or wrong. This will help us to come up with the right filters.)
Balmoral, 6 September, Race 14
1 Real Monet, 34 races, $12,641 earnings, approximately $300 per race
2 Art’s Ticket, 27 races, 34,396, $1,273 per race
3 You Rock, 5 races, 3,316 earned, much less than a thousand per race
4 Four Starzzzz EC, 31 races, 18,727 earned, a little more than $500 per race
5 Wake Up Mike, 30 races, 17,856 earned, just above $500 per race
6 Trigger Fish, 8 races, 2,808 earned, less than $500 per race
7 Time Alone, 30 races, 20,9499 earned , much less than a thousand per race
8 Western Gambler, 14 races, 13,264 earned, slightly less than a thousand per race
9 Bangkok Cruiser, 25 races, 25,377 earned, $1,015 per race
10 Highland Pride, 29 races, 30,096 earned, $1,034 per race
Without looking at any past performance factor other than earnings, you could also see in the past performances that Art’s Ticket showed two wins in his last twelve races, for purses of $9000 and $8910 whereas the next highest purse victory by any other horse in the field was for $7500. This tells us that Art’s Ticket did not accumulate his highest earnings per start from nibbling at place and show finishes, and that on his best night, he is capable of popping up and outclassing this field.
As often happens when searching for an automatic bet, two horses in this field were cause for confusion. The 9 and the 10 were ever so slightly above the $1,000 benchmark. Nevertheless there was a significant gap between those horses and the earnings of Art’s Ticket. I ask computer programmers among us what would happen if we had set the benchmark at a thousand. Can the computer be programmed to discern that Art’s Ticket qualified by the spirit of the law even if not by the letter of the law?
To me this is a key question when developing a computer model. This is vital to know because the essential reason for using these secondary factors which are underbet by the public is that it forces us to bet the horse no matter what else we see in the past performances.
Poor current form
For example, if we had handicapped this race in a “normal” way, we would have seen that Art’s Ticket had poor current form with several back-of-the-field finishes. Without the earnings factor, we would not have been prompted to look beyond the surface for positive nuances:
In his top running line, most recent race, he had shown some degree of new-found early speed, before eventually backing up badly. In that running line, he was in third place first over (sucking air while wide rather than drafting behind another horse) for at least two calls before calling it quits, while in his previous races, he had followed the field like a merry-go-round horse: no moves. His most recent race looked like a mild wake-up. But again, this nuance could have only seemed important in the context of the earnings ranking.
Another asset for Art’s Ticket was the fact that the second, third and fourth highest earnings horses, all of whom around the thousand-per-race mark, were in the poor outside posts 8, 9 and 10.
But the trick of using any mechanical factor such as earnings per race is to NOT depend on interpreting vague handicapping factors and instead to trust the stat we are using.
The fact that Art’s Ticket won this race is not as significant as his final payoff: $79.00. The lesson we learn is that the betting public scorns the average-earnings-per-race factor, and thus it becomes a viable factor. I am convinced that in the Tbreds, average-earnings-per-start-this-year is also an underbet factor. (I suppose that near the end of the year, we run a greater risk of embedding a past-class distortion in the statistic because we are going too far back in the horse’s cycles.)
No one is saying that the average-earnings-per-race by itself, with no filters, will be a deterministic factor, only that it should be respected as one of the wager-value factors. Currently it is not respected in the pari-mutuel financial market.
The message is simply that in any form of investment, lesser-used meaningful information carries more value, and that we will be following up on this factor.
Where I suspect that average earnings per start might be most useful is a scenario not so far from the harness races: cheap thoroughbred tracks where proven losers or dull horses are facing each other and none seem like a possible winner. In the past I was able to use the factor with moderate success at Charles Town. But now, I’m looking for long-term research.
NEW BOOK ON SYNTHETIC SURFACES
BETTING SYNTHETIC SURFACES
Conquering Racing’s Newest Frontier
By Bill Finley
Price: $24.95
Hardcover - 176 pages
Publisher: DRF Press
I have not had the chance to read the whole book yet but have gotten my hands on a couple of chapters, finding that Finley’s experience as a journalists enriches both the substance and the style. He offers numerous statistics to back up his assertions, and places the numbers within a crisp narrative style, backed by opinions from interviewed insiders.
For the first time this fall, the Breeders’ Cup will be run on a synthetic surface. We have seen with trainer stats that some trainers have improved with these new surfaces while others have declined. The universal speed bias has been obliterated but during certain strange interludes, sometimes perhaps because of moisture in the track and other times because of adjustments in the riders’ mass psychology, the speed bias returns with a vengeance, and then, poof, it’s gone.
Finley is the first author to confront in a methodical way the apparently necessary handicapping adjustments and angles, though he admits that the synthetic situation is still evolving, as tracks make adjustments after too many bizarre early results. There are parts of this book that are not pure fun to read, for example, the differences between Polytrack, Cushion Track and Tapeta, and it seems like a daunting escalation of handicapping variables.
If you haven’t read Bill Finley, you’re in for a treat. He handles the English language with craft and precision. He has been covering horse racing for a quarter of a century. He’s also a fan, and that adds flavor to his writing; he’s visited 67 North American race tracks and never missed a Kentucky Derby.
ART KAUFMAN
With Art Kaufman's recent passing, the collective sadness of those who knew
him is enough to move the tides. Art had to use the name Tomlinson because his mutual fund colleagues frowned on his interest in horse racing, though in today’s economy, the Tomlinson ratings are performing better than the stock market.
I used his Tomlinson ratings religiously and one day was moved to phone him to thank him for the best handicapping tool I had ever used. During that period in the early and mid-1990s, before the Tomlinson ratings were published by the DRF and before trainers used the ratings for spotting their horses, this was my automatic bet, my source for signers, and my way of zooming in on one single automatic method.
We got to know each other well and I learned the definition of generosity and warmth. I accompanied Art on his Euro racing tours, which seemed like a business from the outside but was a mere labor of love for Art. I knew this because whenever Art had to make a decision between potential profit for the business or comfort and pleasure for his customers, he chose that latter. He won by breaking even.
Art brought happiness to anyone near him, including my elderly and frail Aunt Ada,
whose heart he warmed with his optimism. Ada was already in her late 80s at the time, and every time she pronounced the word “Arthur” it probably added a few days to her life. That’s because he made all of us feel like truly important people.
For one Breeders' Cup at the Albany OTB, Art noticed that I could not interest my then 14-year son in accompanying us. So Art handed him a racing form and said, "If you are willing to study this, I'll give you $2 to bet for each race ... but only if you study!" My son decided to come along with us, read the form carefully after some instruction from Art, and collected on Reraise in the Sprint.
But when the day was over, my son realized that playing the horses for real required an extraordinary amount of study and persistence, and he decided that it was too big a task for him. But thanks to Art my son had learned to respect the avocation of his old man.
Our condolences go out to his wife Jackie and the rest of the family, and also to the magnificent people who knew and admired him.
GEORGE BANGO
If you’ve played Turf Paradise or Canterbury Park, then sooner or later you would have had to handicap a George Bango horse. Eventually you would have noted that many of his horses were overachievers. Bango operated a small hands-on stable. His stable was neat and clean, his grooms worked in as pleasant a situation as possible, and he loved his job. He was the opposite of a “designer trainer”. He was accessible and always ready with a good one-liner for the media and for any fan who asked a question.
Racetrack managers saw him coming and knew that he’d be griping lyrically about one thing or another, the condition book or the track surface, and after he spoke you wished you had a tape recorder to catch his newest quip.
“You wait and see,” he once said after the Claiming Crown post position draw about one of his longshots. “Superman Can is gonna kick ass. Superman Can can!” And two days later the horse went on to win at huge odds.
Trainers like Bango hold the game together and prevent it from becoming just one more consumer product. Bango was serving as Vice President of the Board of Directors of the Arizona Horsemen’s Benevolent and Protective Association.
George Bango has passed away after a long bout with cancer. Our condolences to his family, his many friends, and his longtime groom Monserrat.