Gametimes

When does it start?

In my day...

As a kid back in the last 1980s I "learned" from listening to Texas Rangers games on WBAP nearly every night during the season that games started at 7:35pm. Of course east coast and west coast road games would start earlier and later respectively. The Cubs had to play day home games because Wrigley did not have lights yet, and there would be plenty of weekend day games involving other teams. It was not the only start time, but in my mind games normally started at 7:35pm local time.

I do not know how true that actually was back in the day. But I feel like the main variation from 7:35pm for night games would have been 7:05. That 7:35 just seemed natural to me. Now, however common it was back in the day, I think the 7:35pm gametime is pretty much extinct. Here is the schedule for Friday April 14, 2023.  Note that the times are shown in my local Mountain Daylight Time, and none of the home teams that day are on that clock.

Over time I learned/concocted (no idea how much of each) some explanations. Starting at 5 minutes after the hour or half hour gives time to any broadcast to introduce the game. Starting half past 7 gives a little time for the temperature to become a bit less scorchingly hot in Arlington. With 1980s average game lengths you could start at 7:35 yet finish pretty soon after 10pm, an end time requiring 7:05 gametimes these days. At least up to 2023 - we will see what the various rule changes produce in this and upcoming years.

Isn't this about making schedules?

Anyhoo, back in the mid-2000s when I started making schedules for OOTP, the game allowed only an integer for the game time. So it was like "1" for 1pm or "7" for 7pm. The game had no sense of time zones. The easiest thing was simply to use the same value everywhere, and I would go with 7.

Soon the gametime could get a full clock value with hour and minute. I had been away from Texas for a decade, and the Rangers were back to mostly mediocrity through the 2000s. I do not know what their gametimes were then. So that same 7pm, or more precisely 7:05pm, was reasonable. That would be 1905 in the game, and it became my default.

That holds to this day. I am not sure when it started, but now the game can use locations and time zones. So you can have every team with 1905 gametimes, but the Los Angeles Dodgers home games will start 3 hours after the New York Yankees home games.

That is a nice touch of realism. I think with that and how the game will mix up teams within a schedule from season-to-season it makes sense to use a consistent gametime. Maybe for night games Toronto likes 7:07, some western teams go with 6:40, others use 7:10, while 7:05 is still used too. But chasing those exact times feels like a fool's errand.

That does not mean though we simply have to make everything a night game. How about some day games and some afternoon/evening games? I have relatively recently added the capability to specify different gametimes depending on the home team or the day of the week.

Mixing things up, at least for weekend games

I set that control over gametimes through the configuration file I have associated with a schedule being built. Because in general the teams are unknown from my perspective the gametime determination by team is rarely used. In most cases if I do anything it is for certain days of the week, specifically to make some weekend games start earlier than that usual 7pm.

The way the configuration works is to set percentages for the distribution of gametimes and for each relevant game randomly pick one. E.g., I may set that for Saturdays I want 30% at 1305, 20% at 1605, and 50% at 1905. What I do then is effectively pick a number between 1 and 100 - if between 1 and 30 then use 1305, if between 31 and 50 then use 1605, or if between 51 and 100 then use 1905.

Over enough games you expect the counts to work out close to the percentages, but there is nothing forcing that. If there are 10 games from the distribution mentioned just above, it is not that definitely 3 will be 1305, 2 will be 1605, and 5 will be 1905. That certainly could be the case, but it will be whatever randomly gets picked. Just like it is possible ten coin flips could result in ten tails, those ten games could all turn out 1905. Extremely unlikely, but it is possible.

Getaway days

One other gametimes adjustment I can make is specific times for "getaway days". The "getaway" refers to one or both teams playing in a different place the next day. The processing can determine those cases and adjust the gametime to something specified. At this time I only have one universal "getaway" gametime, but that could be expanded to having multiple options like the distribution scheme above.

Taking the cases where either team plays elsewhere is rather aggressive, so I am inclined to limit to only the "both" situation. I forget the details of rules MLB has (though others may recall better), but mostly I am trying to ingest some variety and realism and not aiming for perfection there.

Personally I think of a mid-week game with an otherwise unexpected afternoon gametime as a "getaway" day, but this configuration for a schedule affects Sunday games too when there is a series ending between two teams who play other teams beginning Monday. So I may have set Sunday gametimes using some distribution, but any cases that fit the getaway criteria would get whatever they had overridden by the getaway settings. A 60% day game target may initially get hit for Sunday games, but then it could become more like 70% after because of getaway day adjustments. I think the change would be very unlikely to be that big, but the point is that the initial distribution can get skewed.

What's next?

I would say my "style" for gametimes (in absence of any particularly request) is still being developed. So I may be prone to picking up on what the cool kids are doing. What things am I thinking about myself?

The gametime distributions for weekends... I pulled some numbers out of thin air to push some Saturday games day and some to afternoon/evening, though night is still at least the plurality leader. For Sundays I skew more toward day games. Maybe I could do some research to get some actual stats from MLB to use for setting those distributions.

Getaway days... Again, maybe I could compile some stats on how often there are afternoon games in such situations. Maybe instead of it being simply a "yes, change it" if both teams play elsewhere the next day, it could be mostly that and sometimes a yes if only one team plays elsewhere the next day. I should go ahead and make it so there is the distribution of gametime options instead of just the one as currently, even if that may not be much utilized.

Intersection of gametime distributions for days of the week plus getaway days... Perhaps I should do the getaway days first so that can already be accounted for in doing adjustments so the result will be closer to the prescribed distribution. I do think it is currently a small difference that may not be noticeable beyond the random variation. But it would be good to know better what is happening and have better control over the results.