5. COMPUTER SIMULATIONS

 

WHERE ARE WE GOING?
You now have some idea of how to evaluate the skills and abilities of the various players. But what do you do with this information?

While the players' skill levels vary, we know that it is the job of the Matchmaker to try to equalize these differences by handicapping the players through assignment of post positions, partners and opposition. We also know that the Matchmaker is fairly successful overall, because a look at the player statistics in the program reveals that most of the players win about the same proportion of their games played, approximately one in eight.

Since the Matchmaker sees the players play every day, we can hardly expect to know as much about their skills as he does, right? Right! Some people say you can't win at jai-alai because you can't outsmart the Matchmaker. They are right, but wrong. You can't outsmart him, but you don't have to in order to be a winner. What you have to do is understand him, specifically how he handicaps the players.

No Matchmaker assigns the best players the worst post positions and partners in every game. He varies everyone's assignments from game to game and program to program. On the average, the better the player the more heavily he is handicapped, but "on the average" is not every time.

Furthermore, each game offers an assortment of good players and not-so-good players. The Matchmaker mixes them up to make the games interesting. If you or I have a reliable line on the players, then what we need to know next is what the effects of the various handicaps are. If we can assign numerical skill indices to players, we need some way to modify those ratings to reflect handicaps.

Once we have adjusted ratings (i.e., skill ratings adjusted for handicaps) we need some way to translate those numbers into odds for or against winning in a given game against opponents of various levels.

Odds determination is an interesting exercise. Successful gamblers develop the ability to size up a betting proposition and intuitively tell whether the odds are fair or not. You don't have to be a "pro" to recognize that if someone offers you 2 to 1 that you won't flip a head on the toss of a coin, you should take the bet since the correct odds are 1 to 1. But when it gets more complicated than that, you have to be a mathematician or have a lifetime of gambling experience to go on. Determining the true odds (as opposed to the betting odds) in jai-alai is complicated.

Fortunately, there is a way around this problem. It involves something called a computer simulation. It is a technique used to solve complex scientific, business and military problems. It is a "natural" to solve the jai-alai odds problem.

 

SIMULATIONS
Suppose you wanted to predict the outcome of an election as early as possible, the way the big TV networks do. What you would do is hire some mathematically competent people to build a model of the election process, including all of the characteristics, past voting patterns, etc., of the electorate. They would describe all of these things in the form of mathematical equations and program a computer to process the information.

If they knew what they were about, this computer model would accurately describe voter behavior. Then, from early returns on election day, you would be able to tell early on who was going to win.

Wonderful! But how can the networks be so confident that their models are correct? The answer is that they have simulated previous elections, feeding in actual data to see if the computers would have made correct "predictions" if they had been used in the past. If they would have, then they can feel confident that the model is valid as long as underlying conditions don't change.

In jai-alai, the underlying conditions are constant. These are the rules of the game. Instead of describing how voters vote, a jai-alai simulation model describes how players play (i.e., it assumes that they win or lose points in same relation to their skill indices). If our assessment of relative skill indices is correct, then the model will be validated (i.e., proven to. be correctly representing the real world) if we feed in actual historical records and produce results close to what actually happened.

Not everyone has access to expensive high-speed computers. And not everyone knows how to program simulation models. Therefore, while a jai-alai model is a simple one, relatively speaking, most people's chances of building one are small.

I have been incredibly fortunate in having access to output from not one, but two such models. One was built at Georgia Tech as part of a graduate thesis; the other was built to my specifications by an investment consultant in Maryland who agreed to do the computer work if, in return, I would bet for him too.

To date, we have simulated aver 8,000,000 games of jai-alai on giant Contral Data computers. * The models cross-check each other and are validated ** by actual historical records of over 15,000 games.


* For those interested in such things, at Georgia Tech, one million games chews up between a half-hour and an hour of CPU time.

** Kolmogorov-Smirnoff test, 99+% confidence.


 

KING SOLOMON'S MINE
Have we struck the Mother Lode? Do we win all the time? No and No.

Let's back up. Smart gamblers look far an edge. If they could predict outcomes with certainty, there would be no gambling, or at least not far long.

The best players drop balls. The ball takes funny bounces. I've seen the weakest team in a game win without ever making a good play. Their opponents overserved, dropped routine shots, hit the ceiling, etc., while the winning team did little but accumulate points. The fans boo and yell "Fix," but they fail to realize that these things happen. No player in any sport makes no mistakes. A typical professional player handles the ball between 10,000-20,000 times a year during the course of 1,000 or more games. Expecting flawless play would be like expecting a golfer to hole every putt. I once saw one of the top players in the world serve the ball over the top of the front wall in two consecutive games where his team was at Game Point. Both times a victory on that point would have enabled me to cash a Perfecta ticket. I wanted to beat him to death with his cesta. I had to remind myself that he is only human and that without his slashing play on earlier points his team never would have reached Game Point.

You can't win all the time or even most of the time. If you want to win, you must play the percentages. This means betting on teams where the chances of winning are greater than the odds offered. If you do this over and over you will win money in the long run, despite the 18% take, despite luck, despite the Matchmaker's handicaps. But you won't win every game and you won't win every night. Don't expect to.

I am amused when I take someone to jai-alai, and they know I have a "system." They bet exactly what I tell them for the first two games. If I haven't won by then, they begin to get skeptical. If I don't win the third game, they invariably say something like "I think I'll bet on my own in the next game. Maybe that way one of us will win." That's all right with me. I'm just playing percentages. My methods give me a fertile field to plough, not a gold mine to plunder.

 

EVERYBODY'S DOIN' IT
If everyone uses the approach to betting I suggest in this book, will they all win? Or will the method be self-defeating since everyone is using it? Again the answers are No and No.

What I have done is tell you how to go about rating the players. What I will tell you in the next two chapters is how to use those ratings in your betting. Each person who applies what this book teaches will come up with a unique set of ratings. Different persons will apply the techniques differently.

Each person's relative degree of success will depend on how well he evaluates the players and how well he applies the techniques. Most people will not bother, others will give up too soon, still others will be poor at evaluating the players. Many will try shortcuts or otherwise attempt to "improve" the methodology. Theoretically, if everyone did the same thing and made the same judgments, the "bargains" would disappear (but, of course, others would take their place). Practically, however, this cannot happen.

Another fair question is to ask whether the past may be taken as a valid guide to the future. The answer is yes, as long as the structure of the game does not change. The individual ratings will change, however, as players improve, mature, and decline. The difference between long-term success and failure will be the bettor's ability to spot player performance changes which are temporary (i.e., hot streaks and cold streaks) versus those which are permanent.

 

SYSTEM BUILDING
Everyone loves a "system." A system is a set of rules that tells you what to bet and when. It is simple, requiring little effort beyond mechanical application of the rules.

There are two types of systems. "Numbers systems" involve betting certain post positions in specific situations. "Player systems" involve betting players (or teams) with superior ratings at times when their chances are better than the odds being offered.

Numbers systems all suffer a common flaw. They fail when everyone is on to them. For example, one of the first things any jai-alai bettor learns is that the High-Low combination generally accounts for more than half of the Quinielas involving any two numbers. For example, the 7-5 combination in the win-place positions accounts for two-thirds of the 5-7 Quiniela while the 5-7 combination in the win-place position accounts for only one-third. Therefore, you might conclude that betting the 7-5 Perfecta would be a better strategy than betting the 5-7 Quiniela, on the assumption that the odds on a 7-5 Perfecta will be double the 5-7 Quiniela odds. This produces excellent results unless the bettors bet the 7-5 Perfecta twice as heavily as the 5-7 Perfecta. Sometimes they do, and sometimes they don't. You can't just bet blindly; you have to watch the odds.

Similarly, the 1-2-3 Quiniela Box pays off in about one out of every six games. Because it is a popular Box, the average payoff is low. You will generally lose if you bet the 1-2-3 Quiniela Box regularly. However, if you only bet this Box in the game immediately following any game in which it scores, the average payoff is significantly higher. The reason "for this is that the fans shy away from it immediately after it hits, on the theory that it is no longer "due." This is a false assumption that produces a temporary betting aberration on which the observant bettor may capitalize. Eventually, however, more and more people will become aware of this phenomenon and bet accordingly, thus negating its previous advantage.

For this reason, while numbers systems are fun to seek out and develop, I will not dwell on them here since they all are self-defeating in the long run. Player systems are another matter, since each fan will evaluate the players differently.

There are two types of player systems. The first is based on the assumption that each player will win a certain percentage of his games. While details vary, these systems call for betting on players who are "due." These systems are doomed to failure for the same reason that you can't win at roulette playing numbers or colors that have not come up in awhile. Each game is independent of the previous ones. A player's chances of winning do not increase when he's in a losing streak. In the long run you will not have a higher-than-average winning percentage or an above-average payoff amount if you use any of the variations of this type of system.

The other type of player systems involves rating all teams, selecting the highest rated, and making some sort of bet if the odds are acceptable. This is the way I approach systematic betting. For example, I tested a system in which I bet only the highest rated team in each game, coupling it in Perfectas with the two highest-rated teams among the three post positions which most frequently come in second when the high-rated post position wins. That is not as complicated as it sounds.

To illustrate, suppose Post Position 8 had the highest-rated team. The three most frequently winning Perfectas with Post 8 on top are 8-3, 8-4 and 8-5. I would couple Team 8 in two Perfectas, one with each of the two highest-rated of Teams 3, 4 and 5.

This System produced a profit of 51% in a test covering 50 separate programs at one fronton. It scored with one Perfecta out of every 24 bet (which is just one winning ticket per program if you bet two Perfectas every game) at an average price of $109. * Although the average payoff was not as high as most Perfectas, this was more than made up by the high win frequency.


* The profit of 51% is calculated by dividing the average profit per program
of $109 - $72 = $37 by the total amount bet of $72 = (12 games x $6 per game).

I tested it at another fronton and, while the results were still positive (15% profits), they were not as satisfactory since I was less sure of the player ratings that I was using.

A system of betting based on player evaluations should never be mechanical, however. The three keys to successful betting on jai-alai are:

(1) to properly evaluate the players, teams and post positions in relation to the odds;
(2) to properly manage your money; and
(3) to know how the Matchmaker at your fronton sets up his program.

We have already talked about player/team evaluation and money management. The computer will enable us to adjust the player skill ratings to reflect the various post position handicaps. The adjustments are "correct" because when we apply them to thousands and thousands of games, they reproduce the same patterns of wins, places and shows that we see in the real world.

Each fronton is different, not only in its physical characteristics, but also in what it takes in order to be able to win there over an extended period. In Milford, Pedro Bari pairs various players more frequently than is true at the other frontons. Therefore, team records are all important at Milford.

In Bridgeport, Ricardo Sotil puts his strongest team in post 5 rather than post 7, as is true elsewhere. He never plays his Superstars in posts 1, 2 or 3. And he never pairs his top frontcourt stars (Ondarres and Egurbi) with his two backcourt greats, Churruca and Chimela.

In Hartford, the changes of players and Matchmakers can cause confusion. But the World Jai-alai Matchmakers' willingness to mix players of different classes and their frequent placing of stronger players in early post positions affords the astute handicapper with many opportunities for a betting coup. Also, since all of the players come from World Jai-alai's Florida frontons, being aware of the tremendous rivalry between the Miami and Tampa players can provide an edge, particularly in the early part of the season. Generally speaking, an "all-Miami" or "all-Tampa" team may have an advantage over a "mixed" team in the early part of the year, due to the fact that they were recently playing together in Florida. It can pay to know who came from where!

World Jai-alai actually has four "classes" of players, rather than the usual three. In the later part of the Hartford season, when many of the top players return to their homes in Spain and France for a vacation, World Jai-alai will use a high percentage of younger, less experienced players. Many of them are recent amateur champions, or graduates of World Jai-alai's extensive training school system. While they can be exciting to watch, their play is extremely erratic. This makes them quite difficult to handicap. For this reason, long odds on any bet are more important than when betting on the more predictable players.

In the next two chapters we will go into these elements of handicapping in more detail as we are now ready to talk about how to tie together everything we have learned in order to begin betting intelligently.