*This interview with Roxy Roxborough is from 1999. Since then, Roxy has retired from LVSC and now devotes his time to his syndicated odds column America's Line which currently appears in over 125 newspapers across the United States and Canada.
Stern: Thank you for agreeing to the interview. Could you start by describing your profession.
Roxy: I'm the president of Las Vegas Sports Consultants. Las Vegas Sports Consultants have been oddsmakers to Nevada casinos since 1982.* We make the opening odds and point spreads for the casinos and advise them of things that may change the odds and point spreads in the course of events, such as injuries, weather, etc. We also have as clients several of the Canadian provincial lotteries and sports books in Mexico, Australia, Austria, England, Ireland, Antigua, Curacao, Russia, and probably some other countries that I can't remember right now.
I read somewhere that you were a professional gambler before becoming an odds maker.
Yes, I was a professional gambler in Las Vegas for about five or six years, playing poker, betting on sports, and betting on some fixed odds horse racing. Those are games that it is possible to make a living at. In poker you are playing against other players, in sports it's your opinion versus the house's opinion (although, of course, you pay some vigorish in order to play the game), and fixed odds (rather than pari-mutuel) horse racing is again a matter of the bettor's opinion versus oddsmakers.
Were you making a living that way?
Yes.
Then what led you to go into oddsmaking?
Well, in 1981 I had a chance to make odds with a club in Reno. A new sports book manager up there noticed that I had a particular expertise in betting baseball over-and-unders and he asked me if I would help set the betting lines for them while I was vacationing in Reno. It's like a Businessman's holiday; Las Vegas people go to Reno for some reason for their vacation. When I went back to Las Vegas he asked me if I would help make odds on other sporting events and I agreed. Then I got more clients in Reno, and in December of 1983 the Stardust Hotel in Las Vegas asked me to make their opening line for football bets. They are known in Las Vegas for having the betting line available first. It became quite obvious to me that betting and making odds was a conflict of interest, so I stopped betting and started making odds. I didn't do it because I thought that there was going to be a tremendous future in it. There was no way of knowing that every casino would have a sports book and that there would be rapid growth around the world in sports betting. I did it because I was looking for a change. Betting for a living is very stressful, and I thought it was time to move on and do something else, and 16 years later the sports betting business worldwide has grown tremendously and so has our company.
I read Jimmy "the Greek" Snyder's autobiography. Was he the first person to hold the kind of position you have now?
Well, he had better public relations agents than most of the guys that set the odds. The line maker who did it before us was a guy named Bob Martin; he was the Las Vegas oddsmaker from about 1973 until about 1980 when he retired. I think Jimmy the Greek was more of a publicity kind of a person; he was a good front man. Actually he did a lot for sports betting with his television appearances because no one after that has been able to get on television and talk about gambling like he was able to.
Do you have any formal training in mathematics or statistics?
No, not really. I went to the University of Nevada, Las Vegas, and I took an elementary probability course, which I flunked twice. One of the reasons why I flunked the course though is because I didn't take the test.
Yes, we count those.
I couldn't be bothered. I always thought from my standpoint that the important thing was to learn how to ask the right questions. If you ask the question properly, there are a lot of people on the planet who can give you the answer.
So you have made use of people who are trained statisticians.
Yes, we do. The hard part is being able to ask the question. We have people that have statistical backgrounds that work for us from time to time. Also, one of the younger members of our staff went to the University of California at Berkeley, where he was a statistics major and did extensive studies in probability theory. He is an oddsmaker, plus we run a lot of ideas and questions by him.
Setting the Odds for Football
I imagine that one of the most common questions that you are asked concerns the process that is used to set the point spread or betting line on a football game. How do you come up with that initial line?
We like to think it's a little art and a little science. We blend computer technology with subjective analysis. We use computers to do all our datakeeping. At one time when I first started in the business all the datakeeping was done in loose-leaf notebooks. Today our sister company, Computer Sports World, keeps the data for us and displays it the way the oddsmakers want to see it. Now we also have computerized power ratings for all the teams. We blend the computerized power ratings and the experience that our oddsmakers have. Three of the oddsmakers that work with me have worked here 11 years, another one 7 years, another one 8 years, so we have a lot of experience and it would be a shame not to use that experience.
How does the blending work? Would you use the computerized power ratings as a starting point and then add other subjective factors?
Well, actually, we usually do the subjective analysis and computer work independently, and then we look at the point spread suggested by the power ratings and at what the oddsmakers come up with. If we get too far apart, it's usually one of the two needs tweaking. In pro football the computer power ratings are not that important. They are much more important for college football because sometimes teams don't have common opponents. In the professional sports all the teams play in a closed universe; not only do they always have common opponents but they keep playing each other numerous times and as a result I don't think power ratings are that important. In some sports there are things that are more important than the total amount of individual talent on the teams. By and large, however, when one is handicapping sports, talent is the overriding factor in determining the point spread. Things like injuries, home-field or home-court advantage, and weather are some of the ingredients that go into the recipe, but they are certainly overshadowed by talent. Talent is the most important ingredient.
Follow the Money
When we talked about doing this interview, the first thing that you said is that statistics play a surprisingly small role in oddsmaking. As a starting point, could you explain why the goal of the odds that you quote is not necessary to accurately estimate the outcome of the game?
The idea sometimes has been quoted that the oddsmaker's goal is to split the money with the line so that equal amounts are wagered on both sides of a bet because then bookmakers will use losing bets to pay the winners while keeping the vigorish as profit. The odds don't necessarily agree with how we think the teams will perform. By and large, however, the two approaches become the same. The odds that split the betting action on the game actually coincide with how the teams perform. However, we may be a little slow to move on certain trends because we follow the betting public rather than the teams' performance. If the betting doesn't follow a new trend, then we may not change. I'll give you an example: If everybody is betting on the San Francisco 49ers but the San Francisco 49ers never cover the point spread, then this is not necessarily going to change the way we handicap the game. We don't look at just the results; we like to handicap games by looking at the results for about 50% of our information and the betting patterns for about 50% of it. There is a good reason for doing this because the lines move when informed people, professional bettors, bet a lot of money. We really aren't interested in what the small bettors think or what the uninformed public is doing; we're really concerned about what the professional bettors do. If the professional bettors are always betting on the San Francisco 49ers even though they aren't covering the spread, there's something they see about the 49ers that we don't see. Now they might be wrong at that time, but our experience is that over a long period of time professional gamblers have a reason for why they bet their money.
Do sports books know when the money is coming in from serious, knowledgeable bettors?
Yes; it's not necessary to actually know who they are; just the fact that the betting line moves is significant enough.
Because that means a big enough sum of money has been bet.
Right. There's another reason too. You want to incorporate the way the betting line moves into the way you're making future betting lines. One of the reasons is that it can help if the sports books are losing money. When people bet on sports, people are either betting against one of the teams or they are betting on the other. Those are the only two things that can make someone want to bet. And it becomes quite apparent after one or two games, whether they are betting on a certain team or against a certain team. For example, in college football the knowledgeable bettors may keep betting on North Carolina State one month. The reason you want to incorporate what professional bettors are doing into your line-making process is because, if they ever discover a form of betting "Holy Grail," perhaps some master computer program, then whatever they are getting from their statistical analysis is going to show up in your line. So the door will be closed on them eventually. That's why there is no such thing as the Holy Grail for beating professional sports; not only are sports volatile and changing all the time but oddsmakers are incorporating what the serious money is doing into our analysis, so even if we don't know why they like a team or they don't like a team, it's going to end up in our ratings.
This means that the odds are a moving target if you are trying to build a profitable betting system.
It is a moving target, and not only that, the nature of the sport can change too for reasons that are totally out of everyone's control. For example, they may change the rules from one year to the next, which may make all the previous work you have done moot. That happens particularly in over/under bets.
Yes, like if they change the rules on passing in football and produce higher scoring games or something like that.
They also change the rules in basketball all the time.
You talked before about the possibility of bettors doing something clever with computers. Do you know of syndicates that successfully apply sophisticated approaches in professional sports?
Yes. They can be successful in the short term. We know they are not wildly successful in the long run because with any type of effective betting they would get to the point where they would be betting so much money that the sports books wouldn't be able to take it anymore. But yes, there are short-term successes. There is one thing I have never seen though, I've never seen anybody capable of beating all sports. I have seen people who are pretty proficient at certain sports; I've never seen anybody that could beat them all.
Uses of Statistics
Do sports books themselves use statistical techniques? One important use of statistics in industry is quality control. In sports betting, for example, a sports book might check at the end of the month or the end of the quarter to see if they did well on a particular type of bet, say over-under bets in basketball. Then the sports book might come back to you and say, "we didn't do as well as we have in the past." Does this kind of thing happen?
We get those reports every month by calling all the sports books and asking how they are doing. Usually we have a good idea before we ask them because in sports betting you get instant feedback. When you put up a line at 8:30 in the morning, or at 8:00 in the morning, you're getting feedback on what people think of the line just by the way they are betting. And this feedback comes all day long. And we know how sports books do every day, so by the end of the month when we gather the data we pretty well know how they have done or haven't done and we usually know what they are doing well in. Suppose they are not doing well in a particular thing, like basketball over/unders or they are not doing well in hockey or baseball, whatever it is. Then we have to decide if it's something that we are doing wrong in compiling the odds or if it is a statistical anomaly. If the point spread is supposed to make things equal, then half the time people will win and half the time people will lose. But, just like if you flipped a coin 100 times, it's not going to come up 50/50 each time. There are going to be some streaks where there's going to be a lot of heads and there are going to be some streaks where there's going to be a lot of tails. Sometimes people just pick winners, and that's the nature of the game. Sometimes they pick too many losers too. We've had seasons where we have won more than we should have.
Do you ever use statistics to check whether a hypothesis is true? For example, I saw a study sometime in the last year or two about the poor performance of West Coast baseball teams traveling east. If you saw such a study, would you go back and look at data to see if it was true?
Yes. I can think of some examples. One of them was that a pro basketball team playing its fourth game in five nights does not play very well, particularly if the games are on the road. That is already factored into the line now. We have also looked at basketball teams playing in Denver and playing again the next night because of the effect of the altitude factor. The things we are talking about now are called situational analysis, and what that means is that the situation can be more important than the talent. It happens mostly in professional basketball and often revolves around travel because the travel schedule is so brutal in professional basketball that it can actually equalize the talent. So if we have an idea or we have a theory about something that may be affecting how the game is played, we'll go and take a look at it. We might need to make an adjustment or we might not.
Types of Betting
One of the things that I find interesting in sports betting is that point spreads are used in football but odds are used in baseball. Do you have an explanation for the two types of betting?
The reason you can't use point spreads in baseball without also having odds associated with the bet is that the game is so even that even half a run is worth too much. Also, using odds for baseball is better because the game of baseball is priced by pitching rather than on team scoring strength. About 70 to 80% of the game is priced by who the pitcher is. Here is something you might be interested in. I worked 10 years ago on a great experiment. I thought that the bettors had pitching overrated and that really it was only worth about 50 or 60%, and we started making odds based on the teams standings. We'd use pitching for just about 60% and the teams' talent for 40%. The lines got adjusted and moved by the bettors so much that after a week I had to stop the experiment. They weren't interested in whether the odds were right or not; every game betting line was moving after we opened it, indicating that the bettors were convinced that pitching was between 70 and 80% of the game and not the 50 or 60% that I thought. We just had to go back and switch.
Several of the things you have said today bring analogies with Wall Street to mind. There is an analogy to your baseball experiment. People use the Black-Scholes formula as a guide to valuing stock options. This is an academically developed formula that people know isn't quite right, but everyone uses it and it begins to drive the market. If you were correct, then in theory you would have won many bets. Did you give up because it is more important to split the betting than to "be right?"
Well, there are two problems. First, it's pretty brave to experiment with other people's -- namely, the sports books' -- money. That's one problem. I have a piece of a bookmaking operation out here too, so people might say that I could have done that with my own place. But the problem then is you would have too many arbitrageurs in the business (bettors taking advantage of the discrepancy in odds) and the arbitrageurs would end up, even if I was right, with the money because they would just take too much money out of the game. So you know, sometimes you have these ideas, and as long as the betting market is convinced the idea isn't right you have to go along with it. One of the real big betting groups ? they aren't in business now but they were for five or six years, and all they did was play baseball; they always bet on the best pitcher.
Technology
Here is an opportunity to compliment yourself. Do you think the initial lines have gotten better over time? The first line that is put out on football games each week -- is it more accurate now than it might have been 15 years ago?
Yes, we know it's a lot more accurate because of what has happened with teaser bets. We know the opening lines are much more accurate, but that doesn't mean we are making any more money. One of the reasons they are much more accurate is because the people betting have become more proficient, so the odds have become more proficient. It reminds me of card counting in blackjack. I counted cards beginning in 1970. By 1975 I was infinitely better at counting cards than in 1970, but the opportunity to make money was better in 1970. The casinos adjusted to card counters by shuffling more often, then the count started getting better, and then casinos would adjust the shuffling techniques or the rules again. You have to react to the market. So yes, our odds are much better now than they were 10 years ago and they are much better now than they were 5 years ago, but that's because the market has driven us that way. My guess is that if no one with an IQ of over 100 ever made a sports bet we probably would still be using a hand calculator and loose-leaf notebooks. There would be no reason to be better.
Technology affects you just like every other industry.
The pro football lines have become such accurate barometers that we have the catch-22 situation of the lines being so efficient that the sports books often lose money on teaser wagers.
A teaser bet is where the bettor has the opportunity to use some number of points and to change the lines in their favor.
Right, they can take a seven-point favorite and make them a one-point favorite. Then you need to have at least two correct selections in order to win the bet. The more accurate the point spread the bigger the advantage to the bettor in a teaser bet. Sports books have had to lower the payout on these bets.
Given the increased accuracy of the odds, have you ever seen results that appeared so unusual you suspected point shaving or game outcomes being fixed.
Occasionally, if the betting is too extreme then we will stop betting on the game. This happened for an Arizona State-Washington college basketball game for which people recently pled guilty to point shaving. Just based on results, you can never assume a game is fixed. Results are so random in sports that you can never look at a result and say it must be fixed.
That seems like a perfect sentence to end on. Thank you.
