March 17, 2010

Close Games

Tom Tango doesn’t like the final score as a measure of closeness of games.

The “one-run game” is an AFTER-THE-FACT categorization of the game, but we use the category to describe it in real-time or before-the-fact scenarios. A much better way to categorize a game as close is to use win expectancy and leverage index. The closer the win expectancy is to .500 for one team, and the higher the LI, then the closer the game. There. That’s what we want.

Yes, but that’s complicated since you need to keep a win expectancy table around. What I’d like to suggest is the average of the score difference at the end of each half inning. The closest games would be ones in which the home team wins 1-0 in the bottom of the ninth or later. The biggest blowouts would be where the visiting team scores a ton of runs in the first inning and keeps piling on. The nice thing about this, you can do the calculation in your head from the box score.

The Day by Day database contains events going back to 1974 than allowed me to calculate closeness scores for each of the 78343 games. The closest game was a 0-0 tie between Pittsburgh and St. Louis on 9/13/1989. The game was called in the sixth inning. The closest games that played to completion were four 17 innings games in which the home team won 1-0 in the bottom of the 17th. The most recent occurred on 9/29/1993, when Jeff Kent hit a game winning double to put the Mets over the Cardinals 1-0. Each came in with a closeness score of 0.0294 (1/34).

The biggest blowout happened in a 22-0 shellacking of the Cubs by the Pirates on Sept. 16, 1975. The closeness score was 16.1111.

What I like about this measure is that the closest games have the lowest scores (unlike LI, in which the closest games will have the highest scores). The median score is 1.8235, games that the home team wins without batting in the ninth, with a score diff total of 31. In other words, in the average game, you should expect the score difference at the end of any half inning to be around two.

As always, I welcome comments.

9 thoughts on “Close Games

  1. rbj

    Slightly more than half the score of the biggest blowout, even though the run differential was larger, 27 to 22. Makes sense as Baltimore actually led through the first 3 innings, and it only 5 – 3 through the fifth, and a not to unusually 14 – 3 (a laugher, but one that we will see a few times this year) through seven.

    Thanks.

    ReplyReply
  2. ptodd

    I guess numbers can be used to argue anything that seems obvious is false.

    Baseball is a 9+ inning game. At the end of the game, whoever scores the most runs wins. If you lose by 1, 2 or 3 runs, that was a close game. If you lose by 10 runs, that was a blow out, even if the runs all came in the 9th.

    It’s like a horse race, if the horse wins by a nose, it’s close, it matters not how far his nose was ahead or behind earlier in the race. Close is close, we don’t need a new stat other than the final score to know it.

    Now it may be that at times during the course of the game, that the score was not very close, as one team or another had a large lead. So I would agree, at the risk of contradicting myself, that a game where a team had a larger than 3 run lead, and ended up by winning by 3 runs, that this was not really a close game. So call it 2 runs at the finish to eliminate this scenario. 2 runs is close, just 1 HR with 1 ROB from a tie. I mean, we all agree the 3 run save is apiece of cake that even an average releiver can handle, so it should not have been called a close game in the first place.

    Eliminate the 3 run game as a close game and all is well in the real world where all that counts is the final score.

    Now if he wants to integrate the closeness of the game over 9 innings, and has the time and/or tools to do so, god bless him.

    ReplyReply
  3. Dave P.

    A couple of questions, since I’m interested in this also.

    1) If a game ends after 8 1/2 with the home team winning, do you include the 18th “lead” in your average, or just go with the first 17?
    2) I’m assuming extra inning games are included as well.

    I ask because I’m not seeing the ‘spikes’ at the integer values in my work so far, but I’m currently only looking at subsets of games so I could be missing something.

    I like the concept presented here from the standpoint of a fan watching the game – a 1 run game that’s tight throughout has an entirely different ‘feel’ to it than a blowout that turns into a 1 run game with a late rally.

    ReplyReply
  4. David Pinto Post author

    @Dave P.: I divide by 17, not 18, when the home team does not bat in the bottom of the ninth. EI games are included. The spikes are an artifact of the rounding.

    ReplyReply
  5. Ed

    Since all games go nine innnings, if a 10-3 final score flashes on the scoreboard or ESPN, with no details about the game, I am going to think the game wasn’t close. The final score is usually all I will have to go by. I have no idea if the winner scored all ten runs in the ninth inning.

    If the rule was instead that the first team to score three more runs than their opponents win, regardless of what inning it is, so you might get a 2 inning game followed by a 25 inning game the next day, then this problem wouldn’t exist. The more innings the game lasts, the closer the game. It exists only because all games will go at least nine innings. But we know that all games that go into extra innings are close by definition.

    No nine innning game will be as close as an extra inning game. But when I see the score of a game in progress, I do a sort of mental calculation comparing the score with the number of innings left in play to see how close the game really is.

    A better calculation would be to add the margin by which the eventual winning team led, for each inning they led. In the 10-3 game where the winner scored all ten runs in the ninth, the closeness estimate would be 7, they only led in the ninth, by 7 runs. If you have a 1-0 game where the winning team scored in the third inning, and no other runs were scored the whole game, the closeness factor would also be 7.

    ReplyReply
  6. Eric R

    “No nine innning game will be as close as an extra inning game. ”

    By my understanding of this process, that isn’t the case.

    A 2-1 games where all of the runs were scored in the 9th would get a closeness score of 0.056

    A 2-1 game where one run scored in the 1st, the game was tied in the 9th and the winning run scored in the 10th would have a closeness score of 0.9

    ReplyReply

Leave a Reply

Your email address will not be published. Required fields are marked *