THE BOOK--Playing The Percentages In Baseball

"You can certainly make the case that there’s been a huge shift sometime around 2005. “
---
I take it you mean a shift from league parity to AL superiority.  In fact it appears that the AL didn’t catch up to the NL in wins until during the 2006 season.  Through 2005 it was NL .502, AL .498.

I am not sure how you can determine any differences between the two leagues by using WARP or any metric for that matter.  Since each league has different parks and pitchers cancel out hitters and vice versa, the only way that I know of to determine the relative strengths of the leagues is to do one of two things:  One, look at the performances in interleague games by comparing how, for example, AL offenses perform against NL pitchers in NL parks compared to NL offenses against the same teams in the same parks.  Vice versa for the pitchers of course.  Or, alternatively, or in conjunction, look at how batters do, relative to the rest of the batters in the league, when they switch leagues.  For example, if an average batter in the NL goes to the AL and he becomes a below average batter and an average batter in the AL goes to the NL and he becomes an above-average batter, then the AL has better overall offense (usual sample size caveats applying of course).

There is no doubt that the strength of both leagues (pitching and offense) fluctuates from year to year (which is actually one reason why run scoring fluctuates from year to year) and therefore, so does the balance of power, although when one league is ahead it tends to stay ahead of course, at least for a while.  And certainly if one league either spends more money, is smarter, or has some other advantage, it will also tend to stay ahead.

I try and determine the relative strength of the leagues each year.  There is little doubt the AL has been ahead for the last few years (I forgot when it started, but 2006 seems reasonable).  There is evidence that most or all of the current difference is in pitching and that the NL has caught up in hitting since last year.

In any case, as I said, there is no way to do it unless you do one of the two methods I describe above.  Even if both leagues played in the exact same parks with the exact same rules, if one league scored exactly the same number of runs as the other, the two leagues could be equal or one league could have better pitching and hitting, or worse hitting and great pitching.

Again, how can you use WARP or any other metric for determining league balance?  If one league had equally worse pitching and hitting than the other league, the total WARP or any other metric would be the same for both leagues.

Sh!t.  Once again, I’ve fallen prey to looking at the “pythag win%” column on B-R.com, which is the last column in the table.  Drives me nuts.  Here’s the corrected numebrs:

(8) 1997
1 1998
(6) 1999
7 2000
4 2001
(2) 2002
(7) 2003
1 2004

6 2005
18 2006
7 2007
15 2008
8 2009

2005-2010 is an average of 10.8 wins.  The historical average is 3.4 wins.  Halfway is 7.1 wins.

If you do a weighted average, with 10% less weight to each previous season, the average is 5.4 wins.  With 20% less weight, it’s 7.3 wins.

MGL: he didn’t use WARP to figure it out.  The implication is that Clay has set WARP based on a certain gap in talent between NL and AL, and we can see the results of that by looking at WARP.

Rally does the same thing.  For example, after adjusting for league quality, Rally’s AL WAR from 2005-08 is a total of 1863 wins on 14x4 teams, for an average of 33.3 wins per AL team.  Over the same time period, the NL WAR is 1590 on 16x4 teams, or 24.8 wins per NL team.

That’s a gap of 8.2 wins.  That’s right in line with our expectations.

Basically, BPRo is overvaluing NL players and undervaluing AL players.

OK, got it. So how do Rally and Clay determine the differences?  As I said, it is very tricky.  Obviously using W/L records from IL games is a coarse way of doing it and, using those records, the uncertainty of the magnitude of the differences is huge, especially if one wants to say what those differences are in any one year.

And Tango, if you want to estimate the true difference between the leagues, isn’t looking at pythag w/l better than actual w/l?  And the 3rd order wins (based on offensive events) even better?

I can’t remember the league adjustment Justin Inaz uses in his power rankings on Beyond the Boxscore but it’s substantial and was a big time talking point when he started going them last year.

Pythag is only good if you ensure that you handle the bottom of the 9th issue.  For example, a great team will not bat in teh bottom of the 9th, and therefore, have fewer outs per game.  We wants runs per 27 outs, scored and allowed.

***

I believe Rally said that he compares players who switched leagues in the same season or adjacent seasons, and compared performance that way.  That’s one of the criteria you used I believe in your THT series.

Anyway, the status quo has to be somewhere around 7 or 8 win gap, and if Clay wants to say it’s under 3, he should prove his case.

MGL, can you compile the implied Vegas odds in the 252 interleague games in 2009?  I’m going to guess it was .550 AL, .450 NL.

In 2009, for all IL games in a NL park, the home team was a .515 favorite.  In the AL parks, it was .566.  So the AL teams overall were .5255.

In 08, it was .518 and .570, or .526 for the AL, and in 07, it was .516 and .568, or .526 again. 

So they are consistent if nothing else.  They also only have a slightly higher HFA for the IL games, around .544 versus around .541 for non-IL games.

Wow.  I think I’m going to make 252 bets this year.  Would I be wrong to do that?  And how can I do this from NJ?

From 2005-09:

AL at home: .619
AL ON ROAD: .514

All-time:
AL at home: .573
AL ON ROAD: .471

It seems to me that Vegas is using the historical basis (1997-2009) in setting its odds.  (Or rather, the bettor is not appreciating the recent disparity.)

Also note: The standard spread in Home minus Away should be 80 points.  Historically for inter-league, it’s 100 points.  So, there’s a huge home park advantage for the leagues.

Vegas/bettors have not caught up.  Should I shut up?

Vegas is actually pretty sharp (these days) about not putting too much weight on small sample sizes.  That is basically what they are doing here.  The are very good at regressing trends.  And don’t be surprised if this year they have the average AL team as a .540 fave. And that may be right.  And, contrary to popular belief, these days, Vegas lines by and large do NOT reflect the biases of the public.  There is WAY too much smart money in sports betting these days. If Vegas lines reflected Joe public, the bookmakers would get killed.

MGL: Just to make sure I understand how all this works. Bookies are essentially bet brokers who are in it to collect their brokerage fees (the juice) and exposing themselves to a little downside risk as possible by having the bets balance on each side.

Wouldn’t being too perfect with odds expose them to risk from the Joe Public betters? It seems like if 50% of the money was “smart” and the other 50% was “joe sixpack” the bookies would want to be halfway between the two. That way, the smarties are taking joe’s money and they collect the juice.

Maybe I don’t understand the business.

Matt,

Professional bookies (casinos, off-shore sports books) do not give a hoot about risk.  They take tens of thousands of wagers on thousands of games throughout the year. There is no risk (other than the risk of putting out bad lines and getting beat by smart bettors).

If the bookies put out the perfect line from God, they would simply make their juice and that is that.

If they put out the same lines that the smart bettors would put out, they also simply make their juice and that is that, because the smart bettors can’t beat their own line and the public can’t beat any line, be definition.  That is essentially what they do, since they don’t know the “perfect” line. They try and put out a perfect line, and then the smart bettors move that line to their own line and then the bookies are happy, although they would make a little more money if they put out the perfect line or the smart money line in the first place, because the smart bettors are making money from the book while they are moving the line.

The only other thing that happens is that the sports book will shade the line a little in the direction of what the public likes to bet.  A little.  Not enough to allow the smart bettors to make money (remember the smart bettors not only have to see a “bad” line but they also have to overcome the juice).  They do NOT do this to get balanced action from the public in order to minimize their risk.  Remember I said that the professional bookmaker (with smart lines) has very little risk in the first place.  They do that to make more than the juice from the public!

For example, let’s say that the correct line for the Yankee game is -150 (Yankees will win 60% of the time) today.  With juice, the sports book will have a -155 line (with a 10 cent line, which most serious sports books have on baseball).  The public prefers to bet on the favorite and they also like the Yankees (lots of NY bettors). So the books will put out a line of Yankees and since they will get more action on the favorite in general and the Yankees in particular, they will make more money than if they had put out a “fair” line of -155.  And the -160 is not enough to entice the smart bettors to bet on the other team at +150 (since they have no edge, as the other team will win 40% of the time).

MGL - What you say makes sense and it seems like the way it should work, but then how do you explain the lines on today’s (4/30) LAN - PIT game.  The early line (4 PM EDT thurs) from Pinnacle was -250 LAN, 233 PIT.  At around 11AM friday it was -285 LAN, 264 PIT.  At around 5PM it was back to -252 LAN, 235 PIT.  That doesn’t seem like a book that has set a confident line that has adjusted to the early smart money.  Its acting more like a book that is balancing its action.  Especially since LA, like NY, is going to have a lot of home town bettors who want to play the favorite.

Hello, MGL…

I notice you used data on public betting trends from Wagerline.

It seems that two teams EVENLY MATCHED would be bet at -103 in Major League Baseball AS OPPOSED TO -110.

The NBA bets go off at -110 with the pointspread.

NHL Hockey seems to go off at -105 for evenly matched teams.

Where can I find a -103 line?

#16, I don’t know what you mean.  I was quoting the “no juice” lines or the midpoint of the actual lines from an actual sports book. 

Some sports books have a “20 cent” line for baseball, like they do in football and basketball, such that two evenly matched teams would both be -110.  Some books have a 10 cent line, where those teams would be -105 each and a few books have a “reduced juice” line of less than that.  I don’t know who has what.  I live near Las Vegas and most sports books have either a 20 or 10 cent line. Of course once one team is a higher favorite, the juice in cents becomes higher, since the sports book makes their vig as a function of the line and the juice in cents.  Most “10 cent line books” for example, have a 20 cent line when the line on one team is more than -200.  If you have any more sports betting questions please go to a sports betting forum.  There are plenty of them on the web I think.  Thanks!

MGL - Any thoughts on my post #15?

Peter, books move their lines all day long based on the action they get, for several reasons.  Smart books like Pinny move their lines based on the action and the caliber of the bettor. If a known smart bettor makes a large bet, they move their line a lot.  If a known sucker makes a bet they don’t move their line at all.  The two reasons that large books move their lines at all are, one, because it every bet suggests that their line was wrong in the first place (again, unless it is a known sucker), and two, a bet on one side suggests future bets on the same side, so they simply want to make the line worse for the future bettors, even if they were confident that the original line was correct. For example, if everyone is betting on the Dodgers, they do NOT move their line in order to encourage bettors to bet on the other side in order to get balanced action. That would be really stupid!  If I put out a line of -200 and I am sure that the fave team will win 2/3 of the time, why would I want to move the line to -220 and encourage people to now bet on the right side?  Sports books are not stupid anymore.  I might move my line to -210 because I anticipate more action on the Dodgers and now I am giving Dodger bettors a crappy line.  Of course I will now get some more Pittsburgh action than I would had I left the game at -210, and I might even get some smart money on Pit (but only if they think that +195 is a really good line on PIT), but that is well worth the fact that I will be gouging all my Dodger action.  If you knew that -200 were the correct line and you found out that 70% of all your bettors were betting LA at that line and you knew that you could keep the line at -200 and still get 70% action on the Dodgers all day long, or you could move it to -220 and get 60% action on the Dodgers, what would you do?  Which is more profitable?

If I put out a line of -200 and I am sure that the fave team will win 2/3 of the time, why would I want to move the line to -220 and encourage people to now bet on the right side?

MGL - But if you are a book and you are sure that the correct line is -200 and that is the line you initially post won’t you get no action from the smart bettors at all?  There is no incentive for a smart bettor to bet what he has calculated is an actual break even bet.  Wouldn’t the book want to set the initial line at -190 or so to try and capture the smart bets and then raise it to -230, or whatever it takes to entice the bettors who bet the big dogs, until the action evens out and then lower it to -205 to try and catch the home town fanboy bettors at a little profit?  Doesn’t that maximize the action on both sides and hence maximize the profits for the book?  It also seems to model what I am seeing with the daily time changes in the line.

"Wouldn’t the book want to set the initial line at -190 or so to try and capture the smart bets and then raise it to -230, or whatever it takes to entice the bettors who bet the big dogs, until the action evens out and then lower it to -205 to try and catch the home town fanboy bettors at a little profit? “

No, no and no.  I don’t know what you are talking about “capturing the smart bettors, etc.”

If a book puts out a fair line, they make the juice period, no matter who bets, and yes, it is true that you won’t get any smart bettors. But…

One, a book can make more money than just by putting out a fair line by setting the line worse in the direction that the bettors (not the smart ones) will bet, which is generally the fave.  So all books at the end of the year, will shade the favorite.  That is a given.

Two, they do not know the fair line and they rely on the smart bettors and/or the wisdom of the crowds to correct their initial lines.  They open up the day before for small limits so they don’t get hurt by the smart bettors.  Then the lines get “corrected” and they raise their limits at around 8:00 AM.  During the day there is still movement, sometimes small and sometimes large.  I already explained why there is movement.  Large wagers are considered smart bets unless by a known sucker, so they will move the line to correct it.  Wagers by suckers or unknowns will also move the lines simply to make more money on bets on the same side because when one bettor makes a bet on the Dodgers, the next bet from a random better is more likely to be on the Dodgers than the other team. There are other reasons to move lines which have nothing to do with balanced action.  If your line is -200 and everyone else has moved it to -220, them you will only get action on the favorite (dog bettors will bet at the other places because their dog line is better), and why should you give all the action to the favorite betters at -200 when you can give it to them at -205 or -210 and make more money.  Plus if your line is wrong (at -200) and everyone else’s line is correct (at -220), then you will either make no money booking the -200 or you will lose money if it is really wrong (and every smart bettor will bet at your place and your place only).

Now let me specifically address your scenario since I threw a lot of stuff out there.

Wouldn’t the book want to set the initial line at -190 or so to try and capture the smart bets...?

Of course not.  If the true (no juice) line is -200 and you put out -190 the smart bettors will bet the -190 and you will lose $3.33 for every $100 the smart bettors bet.  Why would you want to do that?

“...and then raise it to -230, or whatever it takes to entice the bettors who bet the big dogs..”

What? Again, if you “entice” bettors to bet the dog at +210 when the true line is supposed to be +200, you lose money again.  They will win $220 1/3 of the time for every $100 they bet and lose $100 2/3 of the time for a profit of $6.67 per $100 bet.

“...until the action evens out...?

If your action is balanced but the lines on both sides are different, you are not necessarily making any money. If you have to pay the dog bettors 220 when they win and the favorite bettors only lose 190 when they lose, you figure out what happens for every 100 games where the fave team wins 2/3 of the time. You lose money!  That is called being “middled” as a bookie!

Need I go on?

Why I love this blog…

"Again, if you “entice” bettors to bet the dog at +210 when the true line is supposed to be +200, you lose money again.  They will win $220 1/3 of the time for every $100 they bet and lose $100 2/3 of the time for a profit of $6.67 per $100 bet.”

That should be $210 (not $220) 1/3 of the time of course, for a net profit of $3.33.

If Peter opens up a sports book soon, break those piggy banks!

Disclaimer:  All information on this blog for entertainment purposes only.

Most offshore books are nowadays in a race merely to release the line, and they let the big money bettors, typically the sharps, move the line by pure force of money

There is a more than adequate stabilizing device that is in play between the betting public and the sharp bettors.  The percentage of bets by sharps is lower in proportion than the remainder of the overall consensus, but the total money bet by the remainder, basically the public, balances the market, providing a money making proposition by the bookies long term.  The emphasis is not on one particular game but on the market, but over the course of the entire season.  Like the aforementioned LAD game, in which the dodgers happened to lose BTW, the dodgers lose as -220 favorites means obv you lose -$220 per $100 base unit, and conversely win around $180, or the reciprocal at whatever designated book, per $100 base unit if the Pirates win.  The public was obviously on the dodgers, though some sharps may haev been on the pirates.  Now if the public wins the big moneylines roughly 70% of the time, the high juice stabilizes the bookies risk at length, if any risks exists at all due to the juice.

Mike/6,

I’m using essentially an 8 win difference in the BtB power rankings.  I boost AL teams by ~4 wins per year and penalize NL teams by ~4 wins per year.  As Tango pointed out above, it’s actually conservative relative to recent history (~12 win difference), but it makes a big enough difference that I have a hard time boosting it any higher.  I get enough hell about it as it is.
-j