Last week my column on Baseball Prospectus included discussions of both postponements from an historical perspective spurred by the rainouts during this most recent post season coupled with one solution: bring back the regularly scheduled doubleheader.
The data for postponements was provided by fellow SABR member Paul Rivard who kindly shared the information he has collected and gave me permission to publish it. As I shared in the column:
"To summarize, Paul has access to both original and as-played schedules in a spreadsheet. He then uses the online service ProQuest to sift through the New York Times baseball section for each of the 170 or so days of the season and identify which games were actually postponed. Easier said than done, I'm sure. By indicating the postponements in his Excel files he can then run basic queries to calculate the percentage of games that were postponed. As a SABR member Paul has, for the time being, access to ProQuest through his membership. Unfortunately, that organizational access expires at the end of the year, so he's busily working his way through as many seasons as possible.
As Paul mentioned to me in an e-mail, this data is preliminary, and there are bound to be small revisions in the numbers once he has more time to comb through the data in detail. For example, complete results will also include make-up games that were once again postponed. Of course, these shouldn't significantly change the numbers, but keep in mind that this is a bit of a work in progress. In addition, this data will also be combined with that from other SABR members and should eventually be made available on Retrosheet as part of the Game Schedules effort."
Since the publication of the article which included data from 78 seasons including 1926-1955 and then sporadic years up to 2004, Paul has sent data for both 1925 and 1956. He's also done some more digging into 1949 and found that although there were 1232 games originally scheduled, there were a total of 1316 games played or intended to be played during that season. For that season 76 were postponed which results in a postponement rate of 5.78% (76/1316). The preliminary figure, which is what I shared in the column and will show in the graph below, results in a figure of 5.76% (71/1232). So the difference from a rate perspective is likely not significant meaning that the numbers you see in the graph below are illustrative of the overall trends.
For kicks, here is how 1949 worked out (from an email from Paul):
Putting that together equals 1316 games (1232+8+71+1+4) which were played or were intended to be played. The total postponements were 76 (71+1+4).
As for the overall trend the graph below shows the data points Paul has provided along with a best fit exponential trendline.
As you can see, in the 1920s postponements were around 10-13% and that number has steadily declined to around 2% for 2004. There are a number of reason why this is the case and SABR members had some discussion of the topic when Paul originally shared this data several weeks ago. The most popular, though, included better grounds keeping (fields that drain better), the introduction of lights (allowing scheduled day games to be played, especially after 1950), geography (a greater percentage of the game being played on the west coast where it rains less), and stadium construction (domed and retractable roof stadiums along with artificial turf). All of these I go into in a little more detail in the column but of course the happy result is that 98% of the time you pack your family in the car, pay for parking, and shell out oodles of money for hot dogs and Dippin' Dots, you'll see baseball played.
Paul also shared that indeed the ultimate goal is for this data to find its way into the Retrosheet game logs to provide a full record of how each major league season unfolded. And what more could any baseball fan really want?