FanPost

To Roof or Not To Roof

Bruce Kluckhohn-USA TODAY Sports

Happy opening day everyone! Ok, that’s all we’ve got for now, come back tomorrow for actual baseball.

In honor of shit baseball weather in the Midwest this week, and in the hopes of provoking even more arguments in Moose Tracks about the necessity of stadium roofs, I thought I’d share some fun data I pulled from BBRef on game postponements. If numbers aren’t your thing but you still want to get in on the argument the next time someone gripes about weather-cancelled games, here are the main takeaways:

  • Midwestern teams have an annoying tendency to cancel early season games and make them up later in the year
  • Baseball simply should not be played outdoors, ever, in the Northeastern corridor
  • Hurricanes are very bad for business, regardless of the number of roofs you have installed on your ballpark
  • In 2018 Toronto had to cancel a game because a chunk of ice fell off the CN tower and through the roof of the Rogers center, which isn’t really relevant to any of this but which I’d completely forgotten and enjoyed having an excuse to remember
  • The Mariners have, all things considered, made out pretty well when it comes to postponed games

Ok, let’s dive in. First off, quick description of methods: I pulled the data I’m working off of here from Baseball Reference using their historical season schedules. While BBRef doesn’t list games as postponed on their historical schedules, they do note if a completed game was a make-up, and, if so, the date and cause of postponement for the originally scheduled game. BBRef is a little spotty with their records – I had to cross-check against MLB and ESPN schedules in a couple of cases to confirm that a game had been postponed rather than scheduled intentionally as a double header.

Ultimately, I was able to pull together a fairly complete dataset of 150 total ballgames postponed between 2016 and 2019, including the originally scheduled date, reason for delay, and, where it was announced, the fan attendance at the makeup game. I limited the sample to games played prior to the 2020 season because, well, we’ve had some ‘atypical’ causes for game postponement over the past couple of years. I cut the sample off at 2016 because I’d finished my drink and that seemed like as good of a reason as any to stop copy/pasting CSV for fun, like an idiot.

Team results:

Seattle was subject to four postponed games across 2016 thru 2019 seasons, tied for third fewest postponed games with three other clubs. None of those postponements came under our own park's roof, which might seem obvious, but Gulf Coast hurricanes resulted in home postponements in the enclosed domes of both Tampa Bay and Houston over that time span! The only clubs that experienced fewer postponements than Seattle over the selected seasons were the Giants and the Blue Jays (with three postponed games each) and Arizona, who did not have to push a single game. Again, I have to point out that one of the Jays cancelled games came at home – with a moving roof design similar to T-Mobile – due to a freak event that left too much snow and ice on the field to clear before first pitch while the roof was closed.

Postponements

Attendance

Team

Home

Away

Total

Gained/Lost

ARI

0

0

0

ATL

5

6

11

-22,833

BAL

11

9

20

-11,330

BOS

7

5

12

-8,603

CHC

7

7

14

-7,689

CHW

10

6

16

33,919

CIN

4

5

9

14,769

CLE

6

6

12

10,581

COL

5

4

9

-34,149

DET

9

8

17

6,108

HOU

4

6

10

-75,388

KCR

8

6

14

2,488

LAA

2

2

4

-6,329

LAD

0

5

5

MIA

1

5

6

-926

MIL

0

4

4

MIN

12

4

16

-19,518

NYM

5

8

13

-7,046

NYY

12

10

22

33,480

OAK

1

3

4

PHI

5

5

10

2,748

PIT

4

7

11

6,761

SDP

1

4

5

12,739

SEA

0

4

4

SFG

1

2

3

2,844

STL

9

3

12

4,025

TBR

2

7

9

-7,858

TEX

3

5

8

-11,006

TOR

1

2

3

WSN

15

2

17

-16,811

On the other end of the spectrum, the Yankees had to push 12 games at their home park and were rewarded by an additional 10 games pushed while they were on road trips - that's 22 total Yankees games postponed. The Orioles came in second with 20 total games postponed, 11 at home. The Nationals get top prize for screwing over visiting teams, with 15 postponements at Nationals Park (most of any ballpark in the sample) while only suffering two on their own road trips.

With the caveat that attendance numbers are made up and the points don’t matter, I pulled the average attendance for each team in each season and compared the announced attendance at the makeup games. This gave a sense of how many fans were gained/lost by each home team over the sample timespan as a result of postponed games. The most obvious loser was Houston, but this is a pretty unique case. You might remember that an entire HOU-TEX series in 2017 was, due to Hurricane Harvey, moved out of the state and played at a neutral site in Tampa Bay. Houston never made up those games at Minute Maid, so their numbers take a huge and hopefully nonrepresentative hit. I’m a personal fan of the suggestion that roofless teams make up lost games in their opponents’ cities, but I think this makes it pretty clear that it’s not going to happen – the revenue redistribution just wouldn’t fly with enough teams to make this a viable policy.

Aside from Houston’s huge attendance loss in 2017, I have the sense that on a team-by-team basis, attendance differential is probably more noise than signal. Teams that are able to reschedule games during better weather, against more popular opponents, on weekends, etc are more likely to gain attendance relative to their peers. I’m personally indignant that the Yankees and White Sox both had to cancel double digit home games and somehow gained 30,000 fans in their make-up games, feels like being rewarded for bad behavior, but I wouldn’t say with confidence that it’s a feature of those teams specifically.

Division results:

Aggregating attendance to the level of the division smooths over some of the issues with rare event variability, especially if we toss out Houston’s no good very bad year.

Attendance

Division

Gained/Lost

ALE

5,689

ALC

33,578

ALW*

-17,335

NLE

-44,868

NLC

17,866

NLW

-18,566

*Houston 2017 excluded

Hey, wait a minute. Why are the centrals gaining attendance in make-up games when everyone else is, in aggregate, losing attendance?

Postponements

Division

Home

Away

Total

ALE

33

33

66

ALC

45

30

75

ALW

10

20

30

NLE

31

26

57

NLC

24

26

50

NLW

7

15

22

Well, it’s definitely not because they’re cancelling fewer games. Incidentally, all baseball should be played in the NL West.

Home Postponements

Division

Spring

Early Summer

Late Summer

Fall

ALE

11

11

6

5

ALC

26

11

5

3

ALW

4

1

4

1

NLE

5

10

10

6

NLC

13

9

2

0

NLW

3

3

1

0

Oh. Ohhhhhhh. Yep, this looks an awful lot like early season central division games being made up when people actually want to be outside. Later in the season, the east loses games to rain and more extreme weather events, but in the spring we get my favorite non-hole-in-the-roof-of-the-Rogers-center reason for game postponements: "COLD".

Conclusion:

Given our current circumstance, it’s no surprise that the Twins are a major culprit for early season delays. Not many hurricanes in Minneapolis, but April snow? Yes. Often. If you’re frustrated, though, I hope you can at least take some solace in the knowledge that, unlike many other teams in the central divisions, the Twins haven’t seen above-average attendance at makeup games. Maybe it’s snowing at the makeup games too?

Should the Twins have built a roof? Probably. So should the Yankees, the Orioles, and definitely the Nationals. Barring mandatory roof construction, though, it really does seem like the league could organize schedules to maximize games played in cities/parks least likely to be weather-impacted throughout the season. At the same time, I’m not sure Mariners fans should want to sacrifice Seattle summer games at T-Mobile just to minimize chances of postponements – especially given that the M’s disproportionately avoided cancelled and rescheduled games compared to the rest of the league in the seasons leading up to COVID.

Hope LL gets some entertainment out of this on an otherwise frustrating day to be an M’s fan - it’s been a good distraction on a morning when I should be ducking work by gearing up to watch a ballgame.

You can take a look at the raw data and some of my aggregation work here, if you're interested.