Last night we saw what was certainly one of the most exciting 9th innings of World Series history. It pretty much turned into a party. The three most exciting plays of the inning were undoubtedly Kinsler's bloop single, his subsequent steal of second, and then of course Andrus's line drive and hyper-alert base-running. These 3 events can lay claim to the majority of the WPA for Texas, and really got the party started. Looking at the graph above we can see the St. Louis Win Expectancy nosedive from a most excellent 84.4% to a totally bogus 39.2% after the Kinsler/Andrus at-bats. Yet still, while the Rangers' Win Expectancy was over 60% at this point, they were still losing the game.
It still remained to be seen, of course, if Hamilton and Young could execute the task of driving in the tying and winning runs from 3rd base with less than two outs. And the WinEx for the most part assumed Hamilton and Young would be successful. The difficulty of this task became a sort of point of contention at the local watering hole I was attending, so I took to the retrosheet event files and created a few spreadsheet parties to help shed some light.
Since 1950, in that specific base/out state of having runners on 2nd and 3rd, no outs, the runner at 3rd has scored ultimately about 82% of the time. The runner at second has scored 58% of the time. So Texas was in pretty good shape. So much so that the WinEx went all nuts, jumping to conclusions.
For the individual plate appearance, since 1950 the average batter in a runner on 3rd, less than 2 outs situation (or "R3L2O") has driven in the runner 50.6% of the time. This is a scenario that is regularly simulated in batting practice and is considered one of the important fundamentals of the game. Hamilton and Young, both being adequate bat-handlers over the course of their careers have both succeeded above that average as you may have expected, Hamilton at 58% of the time, and Young at 53%.
Larussa's decision to bring in a lefty to face Hamilton could hardly be criticized, but I had the sneaking suspicion it was a hopeless effort. Retrosheet confirms this: since 1960 Lefty vs Lefty matchups have been under 1% less effective than the average match-up in R3L20 situations. In fact, in 31 of the seasons I queried the platoon advantage was reversed, where the left-handed hitter was more successful than average at driving in the runner against left-handed pitching during R3L2O.
All in all, the 2 sac flies produced a WPA of +.209, while the Kinsler/Andrus heroics led to a WPA of .452. Not that what Hamilton and Young did was easy, but asking them to do something that the average player does half the time isn't that spectacular. If we want to look at players that were more spectacular in R3L2O situations throughout history, we have to again refer to retrosheet.
Since 1950 the best players at driving in the runner at 3rd with less than 2 outs (minimum 250 opportunities):
[Batted In %, or BI%, here is the rate at which the runner on 3rd was driven in by the batter per PA]
Now, some of these names are pretty obvious, as they were also simply great hitters in general. And I suppose they should be applauded for being great in general and great in the clutch, but not here. I mean, for some of these guys, the party never stops. And you never know what can happen when you invite those guys, next thing you know, someone broke your parents stereo. Then what are you going to do? Let's instead look at this same group of clutch performers and eliminate all players with a lifetime OPS above .760. In other words, let's find the guys who really know how to get down and party:
B.J. Surhoff? Now I can get down with that guy!
But as much as I want to get down with B.J. Surhoff, we have to restrain ourselves for a moment. Why? Because Rey Sanchez has shown up to the party. Everybody hide the beer. We obviously have a problem. And we have the same problem when we try to look at players who performed substantially worse than average in R3L2O situations. And in this case, in order to weed out those players that were simply bad all the time, let's limit our query to players with a Career OPS above .760 (through 2010):
I cut this query short as we almost immediately run into a problem. Rob Deer, though he tops the list here, just barely breached the .760 OPS threshold. So whether or not he belongs in this group of "under-performers" is up for debate. But he is also obviously very inept at driving in the runner at 3rd failing 65% of the time, and I have no problem with him being at the top of this list. He is obviously handicapped by his severe inability to put the ball in play, a vice which plagued him his entire career. Of all players since 1950 who qualified with at least 250 R3L20 opportunities, Deer had the highest Career K% by far.
What I did have a problem with here is Harmon Killebrew showing up second. Could one of his era's greatest hitters performed consistently below average at driving in the runner from 3rd? What gives? Naturally, we have to remember that because Killebrew was one of his era's greatest hitters he was also therefore walked in 12% of those situations and was intentionally walked an additional 9%.
So let's therefore revise the data to exclude all BB and IBB plate appearances. We then have a new historical average success rate of 55.6%, and a new list of under-performers:
Deer still tops the list as most inadequate batter in R3L2O situations. There is no getting around it, he was not the guy you wanted to have up in that situation. But K% is a large part of the story here. Getting the ball in play is half the battle, really. Adam Dunn is also done in by a large K rate (and this is excluding 2011 data mind you).
Killebrew benefits from the removal of walks from the data set as expected, but still falls 7.2% below average. At this point I've become paranoid, and it's not for reasons you may think. I know, for example we are playing with the fire of small samples, I know this. Largely this was a fun exercise that is now clearly gone out of control. But for a moment here I was paranoid I was not properly giving Killebrew the proper context. I mean, he's a masher, he's expected to mash. So what if there is a runner on third? The Twins could be down 5 runs for all we know. So I again thinned the data to only include tied and 1-run ballgames at the time of the PA. The results are not too different from the larger sample, shown here as "BI% [1-Run]":
Killebrew does see the best improvement thankfully, but it's still only 2.25%. Historical average interestingly drops slightly with the added qualifications to 55.1%.
Now let's return to our over-performers group one last time and remove BB and IBB, because if we truly want to have a glorious party, we should not invite someone for not being able to walk. And we definitely don't want Rey Sanchez at this party.
Mel Hall! Now this is a party! Hey look BJ Surhoff made it after all! Carney Lansford, I haven't seen you in forever! WHAT UP, Delino! Hey look everyone, Rey Sanchez is outside and no one will let him in!! Up high, Jorge Orta!
Now then, it is time to unveil the Best of the Best: Here are the Top 25 Players at driving in the runner at 3rd with less than 2 outs, excluding all BB and IBB PA's, in tied or 1-run games:
Sample size obviously becomes much more of an issue here but we're at least dealing with 100 PA's for the most part, save for Teddy Baseball.
Notice that Mel Hall beats out everyone yet again even without the OPS filter. That dude is a party animal. To be honest, this little project began as a sort of haphazard, fun, careless excercise. But as it evolved I started secretly hoping there was some meaning to the results. For instance, if the world depended on it, and you had 1 out, runner on 3rd, ninth inning and you only have two guys on your bench:
1. Mel Hall, and his 75% success rate, and .755 lifetime OPS
2. Harmon Killebrew and his 50.6% success rate and a .885 lifetime OPS
Keep in mind that after this at-bat, you have me on-deck. And also keep in mind that both my arms were eaten off by velociraptors prior to this at-bat. You still choose Killebrew, right? I'm trying not to fall in love with data here, which I'm prone to do. How confident can we be that Hall knew what he was doing in these situations, and that it's not simply noise?
Anyway, I'm not sure how much you want to invest emotionally in Mel Hall "getting the job done" 20% more often than average in a sample size of 121 PA's, but I'll tell you what, it's enough for me to salute him right here, right now.
Mel Hall, we had the best time at your party.