THE BOOK--Playing The Percentages In Baseball

...on individual games.

I’m still of the belief that Vegas is quite beatable on team W-L totals.  Not all of them… but if you look at the most egregious difference between Vegas lines and some common, publicly-available sabermetric projections, I think you can win that bet more than 60% of the time in the long run.

Two years ago, that team was the Devil Rays.  Last year, I think it was the Mariners.  In both cases, I believe these teams beat their season lines by more than 10 wins.

I’ve yet to wager on this, but I’m eagerly awaiting this year’s lines (and projections) to see if there’s a similar team that looks like a good bet for 2010.

I always find myself wanting to make bets on the pre-season win totals and end up kicking myself later when I don’t.  I wanted to bet the over on the Cardinals, Angel, Yankees, Nationals, Rangers, and Rockies.  I’m pretty sure only the Nationals were a loss from that group.  I wanted to bet the under on the Mets, Braves, Cubs, Dodgers, White Sox,and Blue Jays.  I didn’t do as well on that group…

I know that you’re talking about game by game outcomes though…

Mike, the thing is that most bettors don’t look at those sabermetric projection systems, and many people who do look at the sabermetric projections either don’t bet or don’t have the stones to bet on a team whose projection looks totally out of whack (like the 2008 Rays).

And if there is an inefficiency in the lines, Vegas isn’t going to correct it unless they stop making money on it

Chris doesn’t really show enough of his work to fully evaluate his study, but Steven Levitt has shown this not to be the case for the NFL, historically. 

http://pricetheory.uchicago.edu/levitt/Papers/LevittWhyAreGamblingMarkets2004.pdf

He shows that betting on home underdogs was exploitable for a profit, due to biases held by the general betting public. 

I haven’t kept up with Vegas Watch lately (check the blogroll) but his blogging origins are from a contrarian sports betting forum.  He would be a good source to verify whether these sorts of exploitable biases still exist in MLB or not.  It wouldn’t surprise me too much if people had caught on, but I would hypothesize that they still exist.

Very interesting topic, but I am not completely sold on its results and the article title and premise are a little misleading.

1) First off, you don’t want to bet on every game nor do you want to bet the same amount on every game that you bet on.  Betting the same number of units on every single game is going to make your ROI crap.  You need to develop a model that gives you a win expectancy for each game, then another model (Kelly system?) which tells you whether or not to bet on that game and if so, how much.  Just like in blackjack, if you are able to gain an advantage by counting cards you bet more win the odds are more favorable and if you have a workable model for predicting games then you know when the odds are more favorable.  Without a good model the point is moot.

2) As the author points out there are more than just Money Line bets.

3) Which Vegas lines did the author use?  Opening or closing lines?  From which book?  What was their juice?  Not all books operate their juice the same way.  Obviously, you’d need to take that into account.  It’s more difficult to bet against closing lines and easier against opening lines, but your bet amounts are limited on opening lines.  If you have a good model, you can bet early against the opening lines in modest amounts.

4) Vegas only sets the opening line, forces from the betting public move the line (up or down).  So, if you are using the closing lines (which I would assume the author was doing) it isn’t so much that Vegas is getting smarter (they may be) but what would be killing the average bettor is that the betting public is getting smarter.  With the advent and popularity of sabermeterics there are likely more people betting that can take advantage of and move a poorly set opening line.

5) One season and even two seasons seems like it would raise a small sample size red flag.  The margins on baseball are very slim.  A 66% win probability in a baseball game is considered a lopsided game but in reality a one in three chance is not all that difficult to overcome in one sample.  Any system that could accurately predict winners of baseball games vs Vegas would need to be tested against more than one or two years worth of games in order to prove a winner or loser.

Just some thoughts.
vr, Xei

Bill L, about a year ago Levitt showed that home dogs were just about 50-50 in 2007, and were way under .500 covering the spread in 2008 (don’t know about 2009).

He guessed that it was just luck, but it’s possible that the “home underdog inefficiency” was arbitraged away

I was in an office pool that kept running standings of records against the spread, and I know that underdogs as a whole finished 133-123, but I don’t remember what the breakdown was for home underdogs…

Just went to Vegas and looked at some of the odds.  Mariners were 100 to 1 odds to win the World Series at the end of last offseason.  Now they are 25 to 1 to win the WS.  At 100 to 1 in a weak AL West that would have looked like a great bet to me.

This article and its conclusions are terrible. This is like saying blackjack isn’t beatable because one cannot beat it playing flat bets with basic strategy at a table that pays 6-5 on blackjacks.  It is true that you will lose money if you bet on every underdog (or favorite) every day at a book charging full juice, but no intelligent bettor is doing this.

I wonder if Chris also thinks it is impossible to make money short selling stocks, since someone who shorts every stock on the NYSE is a big loser in the long run.

Actually what the author is saying is that the baseball lines in that year were very good.  I don’t know what methodology he was using, but basically he is saying that the variance between the line and the outcome of the game (I’m not even sure what that means - is he using RMS between the the line and the outcome, where the outcome is either 1 or 0?) is as small or smaller than it would be if the books knew the exact probability of each game.  There is nothing wrong with what he did, as far as I can tell, although I would like to know what he means by the “variance between the line and outcome of the games” or however he worded it.  If the lines are “perfect” then he is right - they cannot be beat by definition.  Obviously the lines cannot be perfect, but according to his methodology, they were darn close for that year.  As he indicates in the article, because of the one-year, small sample he is working with, the certainty of his calculations and conclusions are not great.  But he makes an excellent point, IMO, one that apparently has not been understood by a lot of the readers.  It would have have been much better if he “showed his work” and explained his methodology a bit better (actually at all).

Examples of perfect or near perfect lines by the sports books would be:

1) If they had one team or another a “-infinity” favorite (i.e., they were 100% certain that 1 of the two teams would win) and in fact each of those favored teams won.

2) They had a different line for every game, there were maybe 300 or so unique lines (-110, -111, -112, etc.), and if we look back on all of their -110 lines, those teams won 52.6% of the time, their -123 teams won 55.2% of the time, their -228 teams won69.5% of the time, etc.  These lines would essentially be unbeatable.

An example of a beatable line would be if they had the home team as a -117 favorite (the “no juice” line) for every game.

Basically there are 2 things that tell you how good or beatable a betting line potentially is:

1) The variance of the lines themselves.  In my example of a very beatable line, the variance is zero - every line is -117 for the home team.

2) How accurate each line is in the aggregate.  In my second example of the unbeatable line, the variance is high AND each line is accurate in the aggregate.

If both of things are very high/good, the lines become unbeatable. If one of these things is bad, the lines are beatable.  Number 2 is almost never bad - if it were, that would suggest a bias in the line somewhere - like there USED TO be with respect to favorites/dogs in the NFL.  The thing that makes some lines beatable is number one.  The less the spread of the lines, the more beatable they are.  Another way to look at it is if they have a lot of games lines at -150, for example, obviously it is not true that in ALL of those game, one team is a 60% favorite. If you can figure out in which of those games the favored team is really a 55% favorite and in which of those games they are a 65% favorite, then you can “beat” that line.  However, if they have very few games at -150, and very few at -145, and very few at -155, etc. (a high variance in the lines), and all of those categories in the aggregate are accurate, as I mentioned above, then you CAN NOT beat the line essentially.  How much variance should there be if the lines are unbeatable.  That is what I suspect the author was trying to figure out.  You can do that by estimating the variance in talent for all teams and then applying that to a simulated season of x number of games.  For example, if all teams were exactly even then the perfect line would be -117 for the home team (to account for HFA), and those lines would be unbeatable.  If half the teams were true 55% teams and half were true 45% teams, then you can simulate a season and see what the variance of the true probabilities of each team winning is for each game.  If the variance is equal to the variance in the lines then the lines are perfect (again, assuming that each line in the aggregate is accurate).

In the example I give, where half the teams are true 55% teams and half are true 45% teams, then 1/2 of the games would have a line of even money (without the HFA), when the 45% teams play the 45% teams and the 55% teams play the 55% teams, and 1/2 will be whatever the log 5 is of a 45% teams playing a 55% team.  If the lines reflected those same percentages, then we can assume that the lines are near prefect (we can never assume they are perfect because they can still screw them up and have the same variance as the “prefect” line would have and be accurate in the aggregate) or at least very good.

Although the author does not show his work, and I don’t know what he did, I think he essentially determined that the variance on the lines was so high that he suspected that they could not be beat.  He may be right. He did say that it could be a one-year anomaly or it could be that the lines are just getting much tougher, or some combination.

My feeling is that with all the sabermetric information publicly available on the internet and in books and articles, the baseball lines have become essentially unbeatable, at least after the public has a chance to bet on them.  BTW, for anyone who says that the public determines the lines 100%, that is nonsense.  A line is a combination of the public betting and the oddsmakers’ handicapping ability both of the true probability of the game and of the public. And all of those things are inextricable.  And the oddsmakers put out better starting lines as their knowledge and information gets better.

That is also true of the pre-season over-unders.  In fact, even more so.  With several sabermetric sites posting their pre-season team win/loss projections in March, the lines you see in the sports books for the season wins/losses are very good.  That was not true 5 years ago or even 2 years ago.  You will NEVER see a line like a TBA at 73 or 74 wins like you did 2 years ago.  That was the tipping point.  There is no way that the sports books are going to post a line of 73 wins when all of the sabermetric sites have that team pegged at 87 or 88 wins.  And if they did, it wouldn’t last more than a day or so. Sorry folks, but the quality of betting lines in this day and age is a function of the technology that is widely available on the internet and in other sources.  Betting sports is a whole new ballgame these days…

Ah, this explanation I get.  Ok, so let’s say we have that, that we have perfect knowledge of 15 teams being “true” .550 and 15 teams being “true” .450.

By “true”, this means that the good teams, when they play the other 14 good teams, will win .500, and the good teams, when they play the other 15 bad teams, will win .59667.

(14*.5 + .59667*15)/29 = .550

We know that the binomial tells us that when two equal teams face each other, after 81 games, and if you do this for all the teams, the standard deviation of each team’s OBSERVED win% (or observed win% minus expected-true win%, which is .500 for all the teams in this case) will be 1 SD = .056.  If we get that, then we know the bookies made the perfect call.

And for the games where it’s a good v bad team, 1 SD = .054 for the observed win% minus the expected-true win%.

So, if I am getting this right, MGL is saying to look at each team against the implied odds, and see their observed win% against the implied-expected-true win%.  And if the standard deviation of those differences, after 162 games, is 1 SD = .039, then we know the bookies put out the perfect line, and hence are unbeatable (by the general public as a group).  Obviously, any one person can luck into winning.

Did I get that right MGL?

So, what were the implied-expected-true win% of each team in 2009?

I’ve posted a reply to the original article which you can click my name to read.

Quick reply to mgl on over/unders: The very season before the 2008 TB bombshell, much noise was made about PECOTA’s 72-win projection for the White Sox, whose over/under was as high as 89.5 at some books.  They went on to win 72 games.  Shouldn’t this have tipped off the books?  Did they really need the 2008 Rays to hammer the point home?

You’re sort of right when you say “it wouldn’t last more than a day”.  TB opened at 68.5 wins in 2008, and were up to 71.5 later that day.  But the line closed at 76.5, so there was still plenty of movement to go.

#10. How does one come up with the “implied-expected-true win%”?  If this was known, then we’d have the perfect model, right?  Also, aren’t there infinite permutations to come up with a SD of 0.39?  Are they all equal?  And lastly, if one had the “perfect model”, they’d obviously make money betting a game like baseball. 

And there are so many other factors not discussed in the original article.  Namely, how vig can be reduced by playing multiple books against each other.

vr, Xei

Here would be my guess as to his logic.

1) Let me look at every game and the closing line. 

2) The closing line implies a win percentage and some range of error in which the “house” will win

3) Bucket the games based on the win percentage into relatively tight buckets.  50% win, 50.5% win, 51.0% win, etc.

4) Calculate what the actual win percentage was and was this win percentage within the limit that implies that the house wins.

5) My guess is his logic says that in every bucket the actual win percentages were within the house limits of wins.

But as many pointed out, what one SHOULD do is develop some model that predicts the win percentage.  Probably 90+% of games will fall within the limits where you lose money.  However, there is probably a small percentage where you prediction is outside of those limits set by a sportsbook.  Only those are the ones that you bet on.  And as Xei says the further your model’s win percentage is outside the limits of the Vegas book, the more one should be willing to bet.

On a slightly different point, I’m curious as to how far the ban on betting on baseball?  Obviously players and managers, and I would assume owners/general managers are technically banned from betting on baseball. What about someone like a Dan Fox? Is it all employees of the ball club including let’s say a ticket taker or usher?  Curious to wonder if the ban would affect someone like Tangotiger as a consultant to a front office.  Just curious if anyone knows.

I’ve gotten into arguments with people on here in the past, but I still think there’s enough “dumb money” floating around the sportsbooks that the bookies don’t necessarily have to set the line at what they think the true odds are.

I mean, let’s take the case of the team over/unders. How many people do you really know that look at PECOTA projections or anything of the like? I know hundreds of sports fans, and maybe 5 of them look at any sort of sabermetric preseason projections. And yet, I know lots and lots of people that bet on sports, some of whom wager thousands of dollars every night, without any advanced knowledge of statistics whatsoever.

Sabermetrically-inclined people are still a small minority of the entire population, and most likely make up a small percentage of gamblers.

Sure, there’s probably more “smart money” being put down on games than in times past, making it harder for Vegas to shade the lines, but I still think there’s plenty of dumb money showering too much money on favorites and big-name teams. I think the public as a whole still displays betting tendencies that the bookies can and do take advantage of on occasion. Maybe not all the time, and definitely not as much as they used to, but probably from time to time.

I dont know. Maybe I’m wrong, but it just kinda seems to make sense to me…

The bookie though doesn’t have to set the line so that he’s got equal number of bets on each side of the line.

Suppose the Cubs and Yanks are in the World Series, and suppose they are true equals.  But, with people’s biases what they are, they expect lots of action for the Yanks.  They are not going to necessarily set the line so that they have equal bettors, especially if they “know” they’ve got a coin flip on their hands.

I presume they’ll move the line in such a way as to try to minimize their losses (perhaps ride their vig)?

So, I’d say it’s the bookies who are the ones who will take advantage of the “dumb money”.

#14.  Prop bets usually have much lower limit on how much you can win/bet.

#15. Exactly!  Vegas’ “true objective” is NOT to alter the line to split the money evenly on both sides.  Vegas’ true objective is to maximize their profit.  The two have a lot of overlap, but are not the same thing.  They maximize their profit by doing what you described when need be.

vr, Xei

Tango, I definitely agree with what you’re saying. The bookie wants to maximize his expected return. Usually that means setting a “true” line. Sometimes, it means getting even money on both sides. Sometimes it means setting a different line all together.

If there’s a lot of dumb money going one way, the bookie will definitely try to take advantage if it can. And an astute gambler MIGHT be able to simultaneously take advantage. I emphasize “might”, because it would probably be difficult to identify a line that the bookie has purposely altered

The bookie does not maximize his return by setting a line which (presumably) splits the public action.  By doing that he minimizes his fluctuation only.  Well, he also reduces the chance that a wise public is going to get the best of him as well, if he should put out a bad line and public knows it.

However, he reduces his return by setting a line that splits the public action, assuming that he can set a better line than the betting public, which may or may not be true.  For example, let’s say that he thinks that the Yankees should be a 2-1 favorite (-200 for the a “no-juice” line) in today’s game, but he thinks that the public is going to like the Yankees a lot such that 60% of them will beet on the Yankees at a -200 line.  Basically, the public thinks that the Yankees should be a -220 (or whatever) favorite but they are wrong.  Again, he is confident in his own assessment of the game (-200).

If he sets the line such that he gets balanced money on both sides with no fluctuation (he makes the juice no matter who wins the game), he is going to make his vig (1 or 2% in baseball) and that’s it.  In order to do that, he has to set the line at -220, or whatever the public thinks it should be.

However, if he makes the line -210 or so, and more of the public still bets on the Yankees, then assuming that the true odds are 2-1, then the bookie makes more than the vig.

In most sports the public likes the favorites and the bookie knows this.  He slants the line a little toward the favorite but not so much that he gets balanced action.  He wants more action on the favorite at a bad line (one that is skewed toward the favorite) so that he can make more than his vig.  The bookie always wants to encourage the public to bet on the “wrong side” of a game.  He is not necessarily looking for balanced action unless he is extremely risk-averse.  These days, bookies do not operate independently in terms of the lines they set.  So bookies generally cannot get away with having (more than slightly) different, independent lines - at least the large ones that is.

If everyone had the Yankees at -210 in order to get more action on the Yankees on a bad line, and you stubbornly had them at -220 because you wanted to get balanced action, you wouldn’t get balanced action!  Everyone would bet the dog at your book since they could get better odds on the favorite at another book.  So the only reason one bookie might have a different line than everyone else is to encourage action on one side or another (or they are a small bookie with clients who do not have much access to other bookies).  Again, bookies have various reasons why they want to encourage action on one side or another.  One reason is to keep the action somewhat balanced (they don’t want to have huge fluctuations and they are afraid that they might have a bad line - that the public knows something that they don’t).  The other reason is, as I said, to try to encourage people to bet on what they consider the wrong side of an intentionally bad line.

Keep one thing in mind.  Bookies are very good at setting lines these days (at least the ones who do set the lines - most bookies just follow other bookies).  Of course there is at least one company in Las Vegas who has a large staff and is paid to set the lines.  They really want to try and at least put out what they think is a correct line on most games and then shade it towards that they think the public will do.  The says of putting out a line that splits the sucker action is gone.  Why is that?  Because there are way too many sharp betters with large bankrolls.  If they put out a line which split the sucker public action but it was a really bad line (which would happen all the time, especially in high profile games), they would get hammered by the sharpies.

Anyway, back to the question at hand.

What I would do is first estimate the spread of talent among the teams.  Then I would simulate a season.  For example, let’s say that in baseball, I think there are only 3 true talents in any given year:  .500 teams, .450 teams, and .550 teams.  Now there are going to be 2430 games, right.  So first game, I randomly choose 2 teams.  Obviously each team has a 1/3 chance of being a .500, .450. or .550 team.  Say I get (after choosing randomly) a .450 versus a .450 team on my first game.  The true line for that game is even money of course.  Then I do the same thing for day 2.  When all is said and done, I compute the variance of my lines.  That should be the variance you get if you made perfect lines.

Now, I look at the bookie’s lines.  The first thing I do is to test his lines in the aggregate to see if they are accurate, which they should be.  Even a bad bookie who puts out bad lines should be accurate with his lines in the aggregate.  If they are, then he passes Test I.  The second Test is to look at the variance of his lines.  If it is less than the variance we came up with in our perfect line experiment, then he is exploitable.  That means that when he puts out a line of even money, sometimes the line is really less and sometimes it is really more than that, even though on the average it is even money. If we made perfect lines, we would be able to exploit his lines.

If his variance is greater than the perfect line variance, then he is taking a game that is, for example, even money, if we had a perfect line, and sometimes making it more and sometimes less, even though on the average he is right.  That is exploitable as well.

I THINK that is the method I would use to test a bookie’s lines.  As I explained that requires a knowledge of or an estimate of the distribution of true talent among the teams, which I think we can do with a database of season records, by using the formula variance of true talent + binomial random variance = observed variance.

I don’t know if this is the same or different from what Tango talks about above and I don’t know if this is the same or different from what the author of the original piece did.

This is a troubling thread, given the talent of the contributors.  (MGL’s comment about the lines being “near-perfect” one season is truly laughable).  It’s really not hard.  The odds available for a given game represent the money-weighted opinion of what the chances are for each outcome.  Nothing more than that.  There is absolutely no reason to think that this money-weighted opinion is going to be the very best opinion.  Smart people develop valuation models to predict outcomes of games.  Whenever a model is sufficiently superior to the money-weighted opinion, it will generate profit.  If any one person can create a model than consistently wins over a large sample, the market is by definition not 100% efficient.  In reality, thousands of people generate consistent profits (as in, make money every year) betting baseball.  I.e., the market is nowhere near efficient.

In my own experiments, some simple models have been positive when back tested from 2007-2010, so I’m inclined to believe that its not *impossible* to beat the moneyline, but I don’t question that its a small majority (if any) that consistently do.

Here are roi’s for various “naive” strategies, aka betting on the fav every time, the dog every time etc… where an roi of -.02 means you lose two cents on every dollar bet, on average. These are on lines from 2007-2010. Note that home dogs did alright. Has anyone else had results like this?

fav, samples=10094 roi= -0.0409333977708
dog, samples=10094 roi= -0.0297076690299
home, samples=10094 roi= -0.00206935809566
away, samples=10094 roi= -0.0360814461771
away dog, samples=6991 roi= -0.0282185616397
home dog, samples=2935 roi= 0.022285094799
away fav, samples=2935 roi= -0.0575084484313
home fav, samples=6991 roi= -0.0109275383997

I do have several lingering questions spurred by the last post, however-

1. Doesn’t the scalability make the market much more efficient? (I don’t think its hard to get around 5000 limits)

2. What evidence is there that thousands of people are profiting consistently? Maybe you speak from experience?

Pat, thanks for posting those results. I haven’t done any research like that myself. If I was a betting man, I would be selective on what games I would bet on. A rule of thumb whether to make a bet, or not, is if you have a ROI of at least 10% or greater on a given game, series, etc. In theory, the higher the ROI percent you get the higher the bet you should place.

Pat, isn’t that just due to natural variance?

Despite the sizable number of ballgames in the four-year period, it seems to me that someone who is omnicient - who knows the exact chance of each team winning each game - would be +/-5% overall after four years on any subset of the population of teams.  And indeed that’s what’s happened.

Is there any reason to really think that Vegas consistently underestimates the chances a favorite will win a game?  The only reason I can think of is if they know that more money tends to be bet on the underdogs and they can squeeze a few more nickels out of folks that way.

#22 I agree there is going to be some lingering variance, even for such a big sample, but I think its smaller than +/-5% from the mean of -.02. This is a link to the results of a monte carlo simulation that show the relative frequency of roi’s over the 4 year period. That is, looking at all random strategies that include betting on every game. The distribution is similar if you look at monte carlo sims where you only bet on half the games (again, chosen randomly), or a third…

http://1.bp.blogspot.com/_RuMK3cdFg1E/TVMCO49YRBI/AAAAAAAAAEo/Y1qoTr9RmD0/s1600/sbr_close.png

Artu/21, There is actually a mathematical formula (called the Kelly criterion) for the “optimal” fraction of your bankroll to bet on an outcome assuming you know the odds of winning and the payout.  By “optimal” I mean that it maximizes long-term expected growth of your bankroll.  In reality it turns out that this is often more aggressive than desirable for humans because a) the odds of winning are an inexact estimate and b) people don’t like the downswings it can produce.  But you are of course correct that when you expect a bigger ROI you should bet more.  For practical purposes due to uncertainty in forecasts, setting a threshold below which the bet is 0 is probably a reasonable alternative to Kelly.

Mickeyg13, Ok, I get what you are saying, after playing around with the formula a bit. I still don’t understand what you mean in your last sentence(were you trying to say this below)?

“If Kelly Criterion application reveals a result f that is very close to zero, it suggests that the odds offered are not sufficient to give any significant amount of expected value to the bet, and the bet should probably be avoided”

Also, some people prefer to use the Half-Kelly or Quarter-Kelly for an optimal approach to reduce the volatility. Does anyone have any idea which fraction of the Kelly formula to use for an optimal approach(if there is one)?

#25: It depends on things like how much volatility you can handle. The full kelly is the optimal solution in the most tangible sense, your bankroll will surpass that produced by any other strategy given enough time. People modify the criterion because they dont want as much volatility, or they think that their edge estimates arent perfect, which they never are. In order to find a kelly fraction that is “optimal”, you would need to specify your optimality criteria more concretely. The basic idea is that the optimality of the kelly criterion depends on your edges being the true edges, which you can never be totally sure about.

Bookies definitely don’t set lines to what they believe is the “fair” line, depending on the game and the particular book’s clientele, they will move the line one way or the other. As an example, a few years ago, Bodog offered two sets of lines depending on whether you had proven to be a “sharp” or “square” bettor. Those who bet on what they considered to be the “sharp” side of their lines, with consistently sizable sums, would eventually stop getting the square lines because they were too easy to beat. But these square lines maximized profitability for the book against average customers (as well as drawing in more big money who might be a smart sports handicapper but was unfamiliar with the subtleties of online sports books vs. Vegas houses).

There were ways to tell where the implied true line was based on the lines across several books, and some of the slower to move books therefore offered arbitrage opportunities. Pinnacle was generally seen as the sharpest book, and was a good gauge of the true fair market line, although they didn’t operate in the US (not sure if that’s changed now).

As a general rule, there are way better opportunities to make money for the few people truly good enough at sports betting. The market just doesn’t offer all that much volume to sharp bettors, which is probably a good sign they know exactly what they are doing.

#1-while the lines for W/L totals and other season-long futures bets may seem weak, that’s because there isn’t really much upside to those bets due to the 6 mos. or so free cash flow you’re offering the book for holding your capital, and the tiny limits usually set on these type of bets. I bet the breakage on these Vegas bets is quite high as well, gotta find the slip and go back to town almost a year later...nice little draw to bring people back to the sports book to redeem their slips and put down some money on the next season.