Friday, September 18, 2009

Yogi Berra As A Clutch Hitter & Clutch Hitting Over Time

Below are two messages I just posted to SABR-L. It looks like Berra's extra good performance in close and late situations is significant. I also found that there is a bigger gap now in how players hit in close and late situations versus overall than there was in Berra's time.

Post 1

On Septemeber 16, 1955, the Yankees beat the Red Sox 3-2. They came from behind with 2 solo HRs in the bottom of the 9th (by Bauer and Berra). That info is, of course, from Retrosheet. The win put the Yankees in a tie for first place with the Indians. It was the first day in Sept that the Yankees had at least a share of the lead. The Yankees had 9 more games and never fell out first place after this. Click here to Retrosheet's box score and play by play. And click here to see Baseball Reference's report on the game.

Berra's HR was an inside the park job. According to Baseball Reference, it was his only inside the park HR in his career (BR lists just 1). There must have been something going on that day, since Robin Yount was born. Click here to see Baseball Reference's breakdown of his HRs.

Anyway, it looks like a clutch hit for Berra. I came across this game looking for something else. So I checked Retrosheet's splits since Berra has a reputation for having been a clutch hitter. It turns out that for the years that Retrosheet has the data posted, 1953-1965, Berra batted .304 in close and late (CL) situations.

Using the none on/runners on data that Retrosheet lists, I found that Berra had an overall average .277 from 1953-62 (Retrosheet says "some situational stats are missing due to lack of play-by-play data for some games"). The combined AB total from the none on/runners data is a little less than his actual total, but it is close). I then found that Berra batted .272 in non-CL situations. So he batted 32 points higher in the "clutch" than he normally did.

That seems like a lot and I wondered if it was significant. To see, I used the formula for a Z-score that Pete Palmer used in his “Clutch Hitting One More Time,” article (from By the Numbers, March 1990). It not only takes into account how much different a player hits in a give situation, but factors in the normal or league average difference. I actually did not calculate the non-CL average in the AL from 1953-62 (it could be done with Retrosheet data). But here are the overall AL batting averages from 1953-62 followed by the CL average (note, this means right here I have not separated out CL and nonCL).


It looks like the overall AVG is just a little more than 1 point better. So the difference between the AL CL AVG and the AL nonCL AVG for these years is probably about 2 points or .002. I then came up with a Z-score for Berra of about 1.90. I think that is borderline significant (there must be people on the list who know more about the binomial distribution who could say if that is right).

But Berra had a .551 SLG in CL situations while it was about .475 overall. So that is about a .075 differential. Over this time, in the AL, the overall SLG was about .005 better than the CL SLG. So that gives Berra a swing of 80 points. My guess is that is significant, although SLG cannot be fit into the binomial model since it is not an either or event. But Berra's CL HR% was 6.6% while it was 4.8% overall. It looks like his NONCL HR% was about 4.5%. So Berra raised his HR frequency by nearly 50% in CL situations.

Now given that SLG was almost the same in overall situations and CL situations during these years in the AL, I assumed no league differential in doing the Z-score for HR%. I got a Z-score of 2.25. That is statistically significant.

Of course, a small % of players will, by random chance, have a Z-score this high. If you have more than 5% of the players be plus or minus 1.96 in Z, we don't know for sure which ones got there by luck and which ones were the true clutch hitters (as Willie Runquist would say). But we can see that there was at least some reason why people saw him as clutch.

Post 2

This is related to my post on Berra. I showed that from 1953-62, the AL batted about .001 less in CL situations than it did overall. Here are the overall/CL AVGs in the AL from 2004-08:


About a 15 point difference per year. Now the NL for the same years:


Almost a 10 point difference per year. From 1991-2000, using both leagues, it was .266/.256. My guess is that over time relief pitching cause the differential to rise.

I also mentioned that from 1953-62, the overall AL SLG was about 5 points higher than the CL SLG. From 1991-2000, using both leagues, it was .416/.388, a difference of 28 points. So, just like AVG, the differential has risen.

When I get some time, I will chart out all the years that Retrosheet has to see how these differences have grown or changed. But I won't mind if someone else does it.

A Yogi Berra interview. He was asked:

"You were known as a great clutch hitter. Somebody once said that the toughest hitter in baseball in the last three innings was Yogi Berra. What made you so good under pressure?

Yogi Berra: I don't know. Maybe I was just lucky. I loved to hit with men on base. Dickey used to holler at me, "You're wasting time at bat with nobody on base," he says."

The Retrosheet page of Yogi Berra's splits shows that he also hit very well with runners on base and with runners in scoring position. His average was about 30 points better than at other times.

No comments: