Hitting home runs when it counts

It was almost a decade ago on the north side of Chicago when Cubs fans began to turn on their fading superstar Sammy Sosa.

In 2003 Sammy’s WAR had dropped to a pedestrian 2.7, his lowest point since the historic home run race of 1998, according to FanGraphs. With all the bitterness of the club’s catastrophic collapse in the 2003 postseason still lingering, stories portraying Sosa as a clubhouse cancer began filtering out into the local papers. In 2004 Sosa’s value slipped yet again, and things went from bad to worse. By January of 2005 the Cubs front office had had enough and finally traded Sosa to the Baltimore Orioles.

It was about this time that I started hearing a very common complaint about Sosa and the untimeliness of his home runs. Sosa was very unpopular by the time he had been traded, but the sentiment that his long balls were conveniently un-clutch seemed to grow in the ensuing years.

I’ve often wondered how true this was. Individual moments in baseball can leave powerful and lasting impressions about the game we observe on a daily basis. If we see Sosa homer in a blowout on Tuesday, then strike out in a tie game in the ninth on Wednesday, that impression based on a two-day sample can last a lifetime.

Enter LI and WPA

Ten years ago, however, I didn’t know about Win Probability Added and Leverage Index. And today we have access to all sorts of wonderful information to gauge “timeliness” at FanGraphs. So I wondered if we could figure the average Leverage Index and Win Probability Added when Slammin’ Sammy launched one onto Waveland Avenue.

The first thing I looked at was the league average Leverage Index and Win Probability when home runs are hit for each year, so we have some idea of what “normal” might be. From 1974 to 2013 the WPA on home runs has remained fairly stable at .13 to .14, dipping to .12 only during the peak years of the steroid era. League Leverage Index for home runs over Sosa’s career bounced around a bit, but averages out to around .96 during that time frame.

We can then compare Sammy’s average WPA and LI on his home runs to those of the league for his entire career:

Sammy Sosa WPA and LI on home runs

Year HR WPA HR LI HR WPA HR above average LI HR above average
1989 4 0.07 0.46 -0.07 -0.52
1990 15 0.14 0.98 0.01 0.02
1991 10 0.20 1.45 0.06 0.46
1992 8 0.14 1.00 0.00 -0.02
1993 33 0.12 0.80 -0.01 -0.17
1994 25 0.11 0.84 -0.01 -0.10
1995 36 0.11 0.76 -0.02 -0.23
1996 40 0.14 1.08 0.02 0.12
1997 36 0.14 0.91 0.01 -0.04
1998 66 0.12 0.93 -0.01 -0.03
1999 63 0.11 0.84 -0.01 -0.11
2000 50 0.14 1.07 0.02 0.12
2001 64 0.12 0.89 -0.01 -0.05
2002 49 0.11 0.85 -0.02 -0.10
2003 40 0.14 0.98 0.02 0.05
2004 35 0.14 0.98 0.01 0.01
2005 14 0.11 0.74 -0.02 -0.23
2007 21 0.12 1.02 -0.01 0.05

For his career, Sosa’s home runs had an average WPA of .13, which matches the league average almost perfectly. Yet the average LI on his homers was significantly below average at just .92 over the course of his 18 seasons. The weighted average of Sosa’s LI on homers is -0.03 below league average. So as it turns out, Cubs fans may have a legitimate grievance with their controversial right fielder.

I should also note that his low LI on home runs is not simply for lack of opportunity. Sosa’s average LI on plate appearances is a healthy 1.03 for his career.

How does this low LI on dingers compare to other sluggers of the WPA era? If we look at all batters with at least 500 home runs since 1974 (when LI is first available at FanGraphs) we see that Sosa’s .92 is the lowest of them all. Among players with 400 homers, his low showing in average home run LI is bested (worsted?) only by Andruw Jones.

WPA and LI on home runs for players (minimum 500 homers)

# Name HR WPA HR LI HR
1 Eddie Murray 504 0.16 1.11
2 Mark McGwire 583 0.14 1.05
3 Mike Schmidt 529 0.14 1.02
4 Gary Sheffield 509 0.14 0.99
5 Manny Ramirez 555 0.13 0.99
6 Rafael Palmeiro 569 0.13 0.99
7 Ken Griffey Jr. 630 0.13 0.98
8 Jim Thome 612 0.12 0.96
9 Alex Rodriguez 647 0.12 0.96
10 Frank Thomas 521 0.13 0.95
11 Barry Bonds 762 0.14 0.95
12 Sammy Sosa 609 0.13 0.92

So how does this happen? What makes Sosa’s average LI on home runs so much lower than a true example of a bona fide clutch slugger like Eddie Murray? Let’s see if we can break this down a bit.

Taking some of the components of Leverage Index to find out how Sosa and Murray’s home run rates compare in each, we start to gain some insight into the differences between the two. We see that Murray on average hit home runs later in the game, averaged more RBI per home run, and did so more often when the run differential was tighter:

Name Score Inn RBI
Murray -0.01 4.81 1.71
Sosa -0.25 4.75 1.63

I’ll admit, the differences between the averages may appear small, but when we look a bit more closely, we see what a difference it makes. Take the average run differential, for instance. Sammy hit more home runs when his team was down, but he also hit many more when his team was out by at least four runs or more:

Percent of home runs by score

Name 1-run / tie Down 2-3 runs Up 2-3 runs Down 4+ runs Up 4+ runs TOTAL
Murray 59.8% 12.7% 12.3% 7.4% 7.8% 503
Sosa 50.2% 14.3% 12.5% 13.1% 9.9% 609

Sammy’s poor performance when it “counted” is not limited to home runs. In fact, his overall performance in critical game scenarios is one of the worst, if not the worst of all time. If we look at FanGraphs career “Clutch” leader boards, we see that Sosa rates as the most un-clutch hitter of all qualified hitters (since 1974). Eddie Murray, incidentally, shows up as the 23rd most clutch hitter with the same set of criteria.

Other losers

Still, Sosa’s knack for hitting long balls when it appeared to benefit only the back of his baseball card is not unique. Other players of the last few decades have shown that same ability, and some have even outdone Sosa. Here are the worst average LI for home runs for all players with at least 100 career homers:

# Name PA HR WPA HR LI HR
1 Don Money 4288 123 0.10 0.78
2 Royce Clayton 8226 110 0.10 0.79
3 Steve Buechele 4800 137 0.11 0.79
4 Kelly Gruber 3463 117 0.12 0.80
5 Randy Velarde 4851 100 0.11 0.81
6 J.J. Hardy 4251 146 0.11 0.82
7 Grady Sizemore 4059 137 0.11 0.82
8 Willie Upshaw 4764 123 0.12 0.83
9 Richard Hidalgo 3948 171 0.12 0.83

Trivia

Since 1974, there have been 15 players who have hit just one career home run with a Leverage Index of 0.00. In other words, the only home run of their career was utterly meaningless (assuming they had not hit another home run pre-1974). Knuckleballer Charlie Hough roamed the big leagues for 25 long seasons and had what is potentially the most meaningless home run of them all.

On April 24, 1977, Hough entered the game in relief in the eighth inning of a blowout. With his team up 15-6 with bases empty in the top of the ninth, Hough hit for himself and put his first and only major league baseball into the outfield seats in Atlanta-Fulton County Stadium. The event added exactly 0.00 WPA to his team’s chances of winning the ball game.

Conversely, Edgar Caceres, in his lone season in the major leagues, as a 31-year-old utility infielder for Kansas City, decided that the only home run of his career should really make a difference.

On Aug. 4, 1995 with the Royals down 4-3 in the seventh inning, Caceres entered the game as a pinch runner. He would come to bat in the eighth inning down a run with runners on second and third. The Leverage Index at that point was an astonishing 4.7. Caceres went yard, winning the game for the Royals, earning himself a whopping .50 WPA for his lone career home run.

Lee Mazzilli and his walk off two-run blast down one run in the bottom of the ninth on Sept. 20, 1976, rates as the highest WPA for a home run at .93 WPA.

Biff Pocoroba earns the honor of hittng the homer with the highest Leverage Index. On May 17, 1977, Pocoroba stood at the plate with two outs and the bases loaded in the bottom of the ninth inning. With his team down a run, the leverage index was at a stratospheric 10.98. Pocoroba homered, winning the game for the 9-6, proving that not every Biff is as much a loser as Biff from Back to the Future.

Tony Gwynn ties world-renowned clutch superstar Bobby Murcer for the highest average WPA on home runs for all players since 1974 with at least 100 career long balls. Both players averaged .17 WPA. Former NLCS MVP Cody Ross ranks up as sixth on that list, while Eddie Murray showed up ninth.

Yet, 16-year journeyman outfielder Bill Robinson still beats them all—Murray, Murcer, Ross, Gwynn, et al. Robinson, who hit just 117 home runs from 1966-1983, stands alone as the most clutch home run hitter with an astonishing 1.24 career average LI on his home runs.

References & Resources
Thanks to FanGraphs, Baseball Heat Maps, and Retrosheet.


9 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Aaron (UK)
10 years ago

Are Bonds & Sosa not getting IBB’d (or at least, pitched around) at the high LI junctures?

Or having specialist relievers brought in?  The changing use of bullpens means a comparison from 1974 through to the present day.

That said the difference between Sosa & McGwire is notable.

Tom Dockery
10 years ago

What mark does Bo Diaz get for his walkoff,three runs down,grand slam for the Phillies on April 13,1983?

Michael
10 years ago

Bo Diaz’ slam in that 1983 game gets a Leverage Index of 3.65. 

Neil Allen came into the game with the bases loaded with the Mets needing just one out, and instead gives up the slam.

Two months later, Allen is traded from the Mets to the Cardinals for Keith Hernandez.

Marc Schneider
10 years ago

“proving that not every Biff is as much a loser as Biff from Back to the Future.”

Or, in a slightly more elevated vein, Biff Loman from “Death of a Salesman.”

James Gentile
10 years ago

Thanks, all.

bstar, the first table is the LI for players with at least 500 home runs, while the last table is worst LI for players with at least 100 home runs. I recognize now that might be confusing.

The best LI guys (with minimum 100 HR) were actually also lower career HR guys like, Bill Robinson, Gwynn, Murcer and Cody Ross.

Aaron, I’m certain the changes in bullpen usage are an issue, but as you mentioned, the McGwire juxtaposition doesn’t really help Sosa out.

And Marc, I think you may fail to fully appreciate the profound themes about life and the triumph of the human spirit in Back to the Future.

bstar
10 years ago

Cool stuff, James. I’m bookmarking this one.

Help me understand why the best guys are all high career-HR guys and the worst, not so much. It’s probably obvious but the light bulb isn’t coming on for me.

studes
10 years ago

James already answered the question of when events are likely to occur in his previous article:

http://www.hardballtimes.com/main/article/average-leverage-index-of-events/

blowry12
10 years ago

Enjoyed the article.

I also enjoyed BN and Aaron’s comments. The correlation between HR’s and WPA (for the 12 players listed at least) is -.33 and between HR’s and LI is -.5.

As BN pointed out perhaps players like Arod, Bonds and Sosa were intentionally walked or teams pitched around these players in leveraged situations. Perhaps these players were also negatively affected by the players hitting behind them (I can’t remember Bonds ever having a good hitter behind him).

To Aaron’s point, I believe Sosa and Arod are in the top 5 in multi hr games. Perhaps you could factor out the 2nd and 3rd hr’s of players who have had multi hr games and compare each players first hr in each game this way a player would not be punished for hitting three hr’s in a game.

B N
10 years ago

Wouldn’t you expect the overall “clutch” factor of home runs to be lower than 1 (where 1 is established by a uniformly random distribution of when home runs occur)?  After all, you’re going to hit more runs against bad pitching.  You are also more likely to score more runs (of any kind) against bad pitching.  Therefore, on average, your home runs will be worth less than you would expect if they occurred in a perfectly random fashion (since they are likely to occur in games where you’ve already scored more runs).