The Most Extraordinary Team Statistic

I’ve been thinking a lot lately about the Rangers’ success in one-run games. They actually lost by one run yesterday, but if you’ve been paying attention, that was the first time that’s happened all season, probably, and one game doesn’t upset the whole pattern. The regular season is just about done, and the Rangers have put together an incredible stat. That’s — well, it’s incredible. I don’t need to provide any other words.

It’s tricky to write about these things, and I wasn’t looking to cover this in the familiar way. We’ve all read a number of articles about what the Rangers have done. We’ve all read articles before about similar teams doing similar things. It’s boring to say “luck,” and it’s tired to say “luck,” and with the playoffs around the corner pointing toward luck is additionally irrelevant. The Rangers have done what they’ve done, and their fans have been able to enjoy it. There’s no taking any of that enjoyment back for math-y purposes.

I’ve just been amazed by how the Rangers have stood out. They’ve won 77% of their one-run games. The Yankees are in second at 69%. Then you have the Tigers at 62%. Thinking about the Rangers got me wondering: Is this the most extraordinary team statistic of the year? I’ll tell you right now: It’s not! Follow along below.

I decided to use some pretty simple math, by which I mean z-scores. I looked at a whole bunch of different team statistics, and for each of them, I calculated z-scores, measuring the number of standard deviations a given stat is from the average. Obviously, it would be impossible to do this for every team statistic. I tried to stick with the normal ones, or at least the reasonably normal ones. Like, I don’t care that the Reds have posted a league-leading 1.065 OPS with two outs and the bases loaded. That’s too obscure. I used my judgment, but I’m sure there are some stats I overlooked, so I invite you to submit any additional entries in the comments.

I’m going to provide for you the top five z-scores that I found. And here are some honorable mentions! The Brewers have 167 stolen bases, which is 2.6 standard deviations higher than the average. Mets pitchers have thrown 66.4% strikes, which is 2.7 standard deviations higher than the average. And Cubs position players have a 35.7 WAR, which is also 2.7 standard deviations higher than the average. These are all very extraordinary team statistics. Here are five team statistics that are more extraordinary.

5) Red Sox, +114.9 Offensive runs

  • z-score: 2.8

Talking about that Red Sox offense again. This includes everything, like pitchers, so that sort of hurts any National League competition, but we see the Red Sox well north of 100 runs above average here. The Cubs are in second at +52.1. The Red Sox are easily the best hitting team in baseball, and for good measure, they’ve been worth another 10 runs on the bases, which ranks them seventh. The pitching staff is pretty good, if slightly underrated, but the lineup looks like a true juggernaut, and they just don’t offer much of a break. When it comes time to putting together a playoff roster, it’s hard to say where the Red Sox will have a weakness.

4) Reds, 244 home runs allowed

  • z-score: 2.9

Who said this was all going to be good? The Reds have allowed 33 more home runs than the next-closest team, and, oh, right, the Reds have also allowed more home runs already than any other team on record in the history of Major League Baseball. The Reds are at 244. The 1996 Tigers finished at 241. The Reds are on pace for 260, and while things have gotten better in the second half, there’s no going back now. You can’t undo old home runs. The record is theirs — it’s just a matter now of what they want to do with it. A special shout-out goes to J.J. Hoover, who’s allowed nine dingers in 18.2 innings. Tim Melville, somehow, allowed five in nine. I should warn you we’re not done with the Reds yet.

3) Reds, 5.27 FIP

  • z-score: 2.9

Right, this shouldn’t be a surprise. The most important component of FIP is home runs allowed. The Reds lead the known universe in home runs allowed. The Angels are second-worst, at 4.68. To be clear, it’s not all about the dingers. Reds pitchers are last in baseball in K-BB%. They lead the league in hit batters. You name it, they’ve been bad at it. So Reds pitchers have a -1.2 WAR. No pitching staff has ever finished with a negative WAR. Time’s running out, Cincinnati. And they close with seven against the Cubs and the Cardinals.

2) Rangers, .766 winning percentage in one-run games

  • z-score: 2.9

Here’s the stat that got me wondering in the first place. When games have been decided by one run, the Rangers are 36-11. When games have been decided by more than one run, the Rangers are 54-52. It’s way beyond the point where we should be wondering about the implications. Regression doesn’t happen in a week and a half, and once the playoffs begin, the whole game is different. It’s of no real significance that, historically, winning a ton of one-run games hasn’t held up. It’s that very fact that makes this so unbelievable to start with. Who really cares how a team wins, provided it wins? This might mean something with regard to how we perceive the 2017 Texas Rangers, but no one in the organization is thinking about that ballclub yet. They’re all too busy thinking about the World Series.

1) Cubs, .251 BABIP against

  • z-score: 3.6

You presumably knew about this. At least somewhat. People have been trying to investigate it for months. The Cubs just haven’t really allowed hits. Part of the equation has to be the pitching staff collectively avoiding hard contact. Another part is that the Cubs have the league-leading team defense. You understand that the Cubs thrive in this area. But have you really appreciated how insane this is? The Cubs, as a team, have yielded a .251 BABIP. The next-best mark in all of baseball is .284, shared by the Blue Jays and Dodgers. That’s a difference of 33 points! The difference between first and second is bigger than the difference between second and 28th. This isn’t just the most extraordinary team statistic of 2016. It’s the most extraordinary team statistic, by an extraordinary margin. The Cubs own what would be the lowest BABIP allowed in the last 40 years, not counting strike-shortened seasons. And that’s not even adjusting for context. Like, say, how the league-average BABIP in 1978 was .275, instead of this year’s .297.

The Cubs allowed a .256 BABIP in April. It was .245 in May, and .257 in June. It was .267 in July, and .254 in August. It’s at .220 — literally .220 — in September. I mean, look, I don’t know. Whatever it is, it just is. The Cubs are amazing. This is where they’ve been most amazing. I’ve genuinely never seen anything like it, and I’m almost certain you haven’t, either.

We hoped you liked reading The Most Extraordinary Team Statistic by Jeff Sullivan!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.

newest oldest most voted
output gap
Member
Member
output gap

Cue johnforthegiants trolling in 3… 2… 1….

ChippersJonesing
Member
ChippersJonesing

We just don’t understand BABIP, k? Only John does.

output gap
Member
Member
output gap

+1

The Cubs obviously understand BABIP better than all of us, except maybe jftg, of course.

tz
Member
Member

Calling it pure luck after one month, ok. Two months, mayyyybe. Halfway through the season, really pushing the issue.

But the dude was calling the Cubs’ BABIP allowed pure luck just a few days ago. While there’s a good chance that luck has played a part in this, dismissing any possibility of something besides luck for a full season of performance like this makes someone look like a Giant d-bag.

ChippersJonesing
Member
ChippersJonesing

Also, Kris Bryant is a bust, because BABIP… Obviously.

output gap
Member
Member
output gap

He’s obviously not good at defense, won’t stick at 3B, 2 tool player. Overrated Cubs hype.

johnforthegiants
Member
johnforthegiants

I have never said this is pure luck. I have always said that this is part pitcher contact suppression, part fielding, and part luck. In fact, the first time I posted about this topic was in response to an article about Arrieta’s BABIP as being a product of his contact suppression, and I argued that a good deal of this was the Cubs’ defense. I have suggested that there should be a statistical model which can explain how much of this is contact suppression, how much is fielding, and how much is luck, based upon general correlations between contact suppression, fielding, and luck. No one has seriously taken this up. A few months ago I tried to do this myself in a fairly simple-minded fashion and came to the conclusion that it’s about 1/4 contact suppression, 1/4 fielding, and 1/2 luck. No one attempted come up with an alternative model. It simply is not an analytical explanation to say ‘The Cubs are amazing’. We can read this kind of analysis in ESPN.

clance
Member
clance

You know what’s even worse analysis? Saying you came up with a model and attacking others for not doing so, acting like your model is absolutely right and not explaining it to people. If you walked is through what you did it’d be cause for discussion but you didn’t and it’s not. I can’t just say I have a model that says mookie betts deserves mvp over trout without explanation. For all anyone here knows that model could be ridiculous like who has a cooler name. Unless you provide reasoning it holds no water.

aprahl7
Member
aprahl7

“…who has a *cooler* name. Unless you provide reasoning it holds no *water*.”

It may be that its Friday and my brain hurts, but I chuckled at this.

johnforthegiants
Member
johnforthegiants

Okay, I found the post, it was from a June 14 article (I may ha’t ve made another post like this, I don’t remember). The numbers weren’t really 1/4 1/4 1/2, I didn’t remember exactly. Here’s the post:

BEGIN JUNE 14 POST:

3 months 8 days ago

If we consider (1) 2016 BABIP, (2) 2015 BABIP, and (3) Career BABIP, the numbers for the Cubs starters are Arrieta .241/.246/.269, Lester .259/303/.299, Lackey .254/.295/.304, Hendricks .244/.295/278, and Hammel .251/.288/.301. This gives an average of .250/.286/.290, as opposed to a league average of .295/.296/.294 (the last number being the average since 2002, the year Lackey started pitching). This means that over the course of their careers, the Cubs’ starters have been 4 points lower than league average in BABIP, while in 2015 they were 10 points below league average, while this year they are 45 points below average. The most favorable spin which can be given to this is that they are doing 35 points better than would be expected this year for some reason or another (while it’s true that their numbers were better last year than over the course of their careers, let’s give them the benefit of the doubt and assume that their BABIP data last year wasn’t a fluke).

How about if we look at the distinctiveness of the Cubs’ pitchers in a different way? It is suggested in the article that the reason for the Cubs BABIP results are contact suppression, with references made to exit velocity and batted ball distance. However, there are seven teams in the league with lower exit velocities than the Cubs, and their average BABIP is .288 (only 7 points better than the league average), so for some unknown reason the Cubs are doing at least 38 points better in terms of BABIP than would be expected. There are five teams in the league with shorter distances than the Cubs, and their average BABIP is .288, so here again for some reason the Cubs are doing at least 38 points better than would be expected for some unknown reason (in fact, the Astros and the Cardinals are better than the Cubs in terms of both exit velocity and distance, yet their BABIPs are .292 and .309). Exit velocity and distance might explain why the Cubs’ pitchers are a few points below the league average in BABIP, as was the case for their starters in 2015, but they give no explanation for why the BABIP is 36 points lower in 2016.

What about defense? Statistics like batting averages on different kinds of hits and turning air balls into outs mean nothing because they simply replicate the BABIP data. The statistic which would seem to be appropriate here would be UZR, because it reflects how many balls the fielders can get to with some objective measure of difficulty. The Cubs are indeed the best team in baseball in this respect, and in fact have the 6th highest UZR of any team since UZR data started being kept in 2002. But the five teams ahead of them have an average BABIP of .282, which is only 12 points better than the average BABIP for this time period.

So some of the variation might be explained by pitching and some might explained by fielding, but well over half clearly can’t, and is presumably luck.

johnforthegiants
Member
johnforthegiants

(this is me in September) It seems to me that I was making some effort to estimate the effect of contact suppression and fielding on BABIP, and I explained my reasoning. I did walk people through it. You can say that you don’t like this reasoning and offer an alternative model to explain the BABIP data (other than just saying ‘the Cubs are amazing’). But no one did that.

johnforthegiants
Member
johnforthegiants

So the guess which I made was that of the 45 points difference between the Cubs’ pitchers’ BABIP and the league BABIP was that at most 10 points could be explained by the pitchers’ abitity (if 2015 data is used for comparison), or at most 7 points (if exit velocity is used), while at most 12 points could be explained by defense, so that meant that at least 23 points must be luck.

johnforthegiants
Member
johnforthegiants

Obviously that was in June and the data now are different but the principle is the same, there should be some statistical estimate of the effect of the different factors

output gap
Member
Member
output gap

Except that the anomaly persisted each and every month. Looking at monthly splits, filtering for team BABIP suppression under .268, you get the following (lowest to highest BABIP allowed):

April: Cubs, Angels, Rays, Indians
May: Cubs, Dodgers, Blue Jays
June: Cubs, Indians, Royals
July: Blue Jays, White Sox, Marlins, Yankees, Cubs
August: Cubs
September: Cubs, Red Sox, Dodgers, Mariners

So the Cubs were outright #1 5 months out of 6, and of 20 occurrences of sub .268 BABIP allowed, the Cubs represent 6.

If you further filter for under .260 BABIP allowed:

April: Cubs
May: Cubs, Dodgers, Blue Jays
June: Cubs, Indians
July: Blue Jays
August: Cubs
September: Cubs, Red Sox, Dodgers

11 such instances, 5 by the Cubs in 6 opportunities.

johnforthegiants
Member
johnforthegiants

You still haven’t giving any statistical analysis explaining how much of the Cubs’ pitchers’ BABIP can be explained in terms of pitchers’ contact suppression statistics, how much can be explained in terms of the fielders’ statistics, and how much can’t be explained by either of these and is therefore presumably a matter of luck. I can easily believe that these two combined will give the Cubs’ the lowest projected BABIP of any team this year. I have no argument with that. This is all your data shows, comparison between the teams. But they are 33 points lower than any other team. This is an entirely different matter. How do you explain that? Of course it isn’t entirely luck, but it seems very likely that a significant part is. How much? I have tried to give an estimate. Do you have any ideas about this?

This is a standard question in statistical analysis: How much of the observed variation of the dependent variable can be explained in terms of each of the independent variables? In crime statistics for a given area, for example, how much variation can be explained in terms of per capita income, how much can be explained in terms of years of school education, how much can be explained in terms of percentage of gun owners, etc. This isn’t anything esoteric, it’s just a standard research question in statistical analysis.

Serbian to Vietnamese to French is back
Member
Serbian to Vietnamese to French is back

OK, I found this article is referred to in article 14. June (and ha’s T’s post like this, I can’t remember the name). These figures do not in fact 1/4 1/4 1/2 I don’t remember exactly. Here is the message:

14. Martha, article:

There are 8 days

If we consider BABIP, 2016 (1), (2) in 2015. year and his career BABIP BABIP (3), these numbers clearly Arrieta flings. 241/. 246/. 269, Lester. 299/259/297. 254/. 295/. Hendricks, 304. 244/. 295/279 and Hummel. 251/. 295/. 301. This enables the average 250/. 295/. 282, not an average League. 295/. 295/. 294 (the last question is the average in 2002, began to cast valets). This means that, during his career, the starters were lower than average BABIP in the League, while in 2015. the year was 10 points below average in the League this year is four points, 45 points below average. The best Spin that can be given for this is that the longer you wait this year for some reason (if it is true that last year more than the amount in his career, to give them the benefit of the doubt and assume that the data their BABIP last year wasn’t a fluke) 35 points.

How do you want people to see our young pitchers. It was the article that was the reason for the Mets BABIP induction in combat results, reference is made to escape velocity and distance, push the ball. However, there are seven teams in the major leagues with the Cubs and lower flow they average BABIP. 288 (only 7 points better than the League average), and for some reason pandas at least 38 points better on BABIP, as expected. There are five teams in the League with less than average, the Cubs and their BABIP. 288 and again, for some reason, pandas are better than expected for a reason at least 38 points (in fact, the Tigers and the Cardinals are better than in the first Panda speed and distance, but are BABIPs. 292 309). Escape velocity and distance may explain why, in desperation they kr?aga rated below average BABIP leagues, as is the case for starting in 2015. year, but did not provide an explanation of why a BABIP 36 points less than in 2016. year.

And the Department of Defense. Statistics like batting average for the different types of views and include Air Ball way out doesn’t make sense because the only copy data BABIP. Stats, seems quite appropriate here cause because it reflects how cans igraiu eggs on a level with the weight of the lens. Pandas really are the best baseball team in the area, and the fact is that there are six major causes of groups by UZR data from 2002. However, the five teams ahead of them inadequate BABIP. 286, which is better than average BABIP just 12 points for the only time.

So, part of the change can be the result of improving and some can explain the Fielding, but seems to be more than half, and it may be out of luck.

Original Greaser Bob
Member
Original Greaser Bob

I came to the conclusion of 100% BS

bunslow
Member
bunslow

Yo someone already made a statistical model back in June: http://fivethirtyeight.com/features/the-cubs-pitchers-are-making-their-own-luck/

johnforthegiants
Member
johnforthegiants

That isn’t a statistical model. It just means they’re better than average.