Is the last out of the game the toughest?

Taking sportscasters to task for the dumb cliches that they spew is one of my favorite hobbies here at StatSpeak. One thing that I’ve heard more than once is the idea that “The 27th out of the game is the hardest one to nail down.” Of course, this one is usually said in situations where a team is trying desperately to close out a win in the ninth inning, while the other team is not being cooperative and insists on pecking away at that lead. What sportscasters really mean to say is, “Gee, I’m really nervous that we’re not going to get to that last out before the other team overtakes our lead. So, I’ll make myself feel better by saying that this is the toughest out in a game. If we end up losing, at least we lost trying to do the hardest thing that there is to do, and plenty of people fail at the really hard things. Ah, I feel momentarily better.” My training is as a psychologist.At the office, Imight call that self-handicapping or a defense mechanism. Either way, it’s a sure sign that sportscasters are both human and anxious.
Is the 27th out of the game really the hardest out to record? That’s actually a little tougher of a question than it appears. How hard an out is to record will depend on who the pitcher is on the mound and who the batter is at the plate. If Joe Nathan is facing some AA kid who’s getting a September call-up that he’ll talk about for the rest of his life and who’s so nervous that he can’t stand straight, that’s going to be an easy out. If it happens to be the ninth inning with two outs, then the 27th out will be rather easy. For these analyses, I didn’t quite go that deep.
I looked at the level of difficulty of the out by looking at how many batters it took for the pitching team to move from the previous out to the one in question. (So, if fifteen hitters came to bat between the recording of the 1st out in the 7th inning and the 2nd out in the 7th inning, then that out was very difficult to come by.)This makesit possible for it to take zero hitters for a particular out to be recorded. A double play to end the fourth inning means that it didn’t take any additional batters to get from the 11th to the 12th out. It was all done in one swift motion. This also means that outs which were recorded on caught-stealings are not counted either since they don’t require an extra batter, although a batter would have had to get on base to be caught, so that about evens out. Not precisely, but baseball is a game of blunt instruments. They’re called bats. Now, of course, this ignores the fact that better pitchers are generally used in certain innings or that pinch hitters are more likely to be used in the later innings. Deal with it.
I took the Retrosheet PBP file for 2006, and found all team-games (two teams play a game) in which at least 27 outs were recorded. (A full nine inning game has 54 outs, 27 per team.) So, assuming that the game wasn’t called in the sixth inning because of rain, the visiting team was always represented. The home team, assuming that they a) lost or b) went to extra innings was also represented. This introduces another small bias, because home teams that won are going to be better than average at scoring runs (they at least scored more runs the other team that day!), and probably a little better at avoiding outs, and it’s possible that there could be some systematic variation in where they spaced those avoidances of outs. But again, we move along. I told you this was a harder question than it looked. There were 3622 team-games accounted for in this analysis. In all 3622 team-games, outs number 1-27 were recorded at some point.
So which out in the game did teams spend the most number of plate appearances trying to record each of those 3622 outs? The first out of the first inning, with 5430 plate appearances needed to record those 3622 first outs.
Strangely enough, in second place was out #17 (second out of the sixth inning) with 5420 PA’s needed to record 3622 outs. The pitcher is more likely to be either a starter who is getting tired or a fourth-tier reliever. Rounding out the top five are outs #23, #14, and #2. Where was out #27? It was the 17th toughest out to come by. The easiest out to get was out number 7 (first out of the third inning).
As always, remember that just because a lot of people say the same stupid thing, it doesn’t make it true.


9 Responses to Is the last out of the game the toughest?

  1. Matt Souders says:

    This isn’t that surprising. The first out of the third inning is usually the 9th place hitter (the average team gets one man on base per inning +/- 1, so the first two innings often involve 8 batters.
    The first out of the game is harder because the lead-off man always bats here and they’re usually the guys with the highest OBP.
    I think what sportscasters are really thinking when they say that the 27th out is hardest to get is “when the batting team is sending quality hitters up there and the leverage index is high, the 27th out is the most stressful to get” which might be true.
    What is needed is an analysis of whether there is a difference in out difficulty AFTER you factor out the skill of the batters and pitchers using the odds ratio method.

  2. Pizza Cutter says:

    Matt, if I tell you that you’re right, will you come back to StatSpeak?

  3. tangotiger says:

    You do have the selective sampling issue, as you noted. My approach would be to look at the performance of the players with 26 outs. This way, a game could end with 26 outs, and all the PA would count. And, take the career (or last 2-3 years) stats to normalize the players. I’d be surprised if you find anything of note.

  4. Pizza Cutter says:

    Tango, I’ll bet dollars to donuts that you’re right on that one. There are probably minor effects here and there (on the same magnitude as clutch hitting), but I bet they’re not stable. Like I said in the post, this one wasn’t about methodological soundness, but rather about a quick test of the cliche.

  5. Mason says:

    Did you try the results in other years? How robust were the results? I expect that some of that ordering isn’t going to be the same from year to year, and I’m guessing that several of the ones that happened to be near the top in 2006 weren’t that way in other years. (Also, you gave a rank-ordering rather than a value, so you’re not showing how close the various outs were, and that would suggest which ones might be robust in other years.) Another interesting thing one could do with year-to-year stats is use this as a way to investigate how the game has changed over the years (e.g., the way relievers are now used versus the way they used to be used).
    Anyway, the article was interesting!
    (Also, excuse the academic questions. I am …um… an academic.)

  6. Pizza Cutter says:

    Mason, no worries about being an academic. I’m a psychology grad student who has decided to use all of that grad student statistical training intended to be used to rid the world of depression and anxiety to the purposes of analyzing baseball. So many of the folks who dip their toes into this type of work are academics leading secret double lives.
    To answer the question of magnitude, the 3622 first outs took something like 5400 PA to get. The last of the outs took around 4800. (I don’t have those exact numbers in front of me right now).

  7. Matt Souders says:

    The only reason I stopped posting to Statistically Speaking was the lack of available free time…it certainly wasn’t a lack of interest. I need to figure out a way to balance things so I can contribute more often. 😦

  8. floridabart says:

    I think you are leaving out all the cases that the announcers were referring to: where the last run was scored in the bottom of the ninth. You should probably add these at bats to your ninth inning totals (ie- include games where there were26,25,24 outs) to see what happens to these outs in your rankings. Other than that, I like your approach…

  9. Matt Souders says:

    there is definitely a selection bias going on in his study, yes. He ignores all of the ninth innings where the last out was NEVER recorded (since, how do you quantify that numerically, infinity?). What he really should do is check what the OBP is for batters in each inning…that won’t have a selection bias when he includes all innings…then at least he could see if the 9th inning is tougher as far as how the batters do.
    Of course we all know that the 27th out is not going to be that difficult to get compared to the first out, in which you face the team’s highest OBP guy (the lead-off hitter).
    Or we could translate announcer-speak into something intelligent and suppose they’re saying “outs are harder to get in high leverage situations like this one)…which we can test. Does OBP go up in high leverage situations relative to the OBP of the player in question and the pitcher he’s facing? You’ll probably find that it doesn’t, but it’s worth a look.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: