This was initially written for submission in the Community Research section of FanGraphs, but they chose not to publish it. That, of course, is their prerogative, so I'm putting it here. The situation itself is dated, but the underlying point remains relevant.
On April 5th, 2012, the Cubs were playing the Nationals on Opening Day. Down 2-1 in the bottom of the 9th with 1 out, pinch runner Joe Mather ran on contact on a ball hit by pinch hitter Jeff Baker directly to Nationals 3B Ryan Zimmerman, who threw Mather out at home. Marlon Byrd struck out looking to finish the game, providing a bleak opening to what many expect to be a long and fruitless Cubs season.
For the 2010-11 seasons, this chart shows the instances of a runner on third with less than two out:
To explain what this chart shows, I'll explain the first line:
--3 shows the instances where there was a runner on 3rd only. Strikeouts, walks and hit-by-pitch are shown because contact was NOT made in this situation, and thus (for the most part) the runner didn't advance. There will be occasions when the runner did advance on wild pitches, balks, stolen base or whatever, but this will be the extreme exception. Therefore, contact was made 3.123 times (4449-715-559-52). Forceouts are NOT included in these numbers, since it's not the fault of the runner (unless he's slow, got a bad jump, whatever) that's he out at home without me going through these plays individually, and with close to 4,500 plays, uh, that's not gonna happen.
The runner on third scored 1,922 times, was out at home 158 times, and either chose not to advance for whatever reason 1,043 times. There are two ways to look at this data, and I leave it to you to make the choice--I would argue that BOTH can work given the situation.
1. What is shown here is Out at Home/(Runner on 3rd Scores + Out at Home), or 7.6%, which assumes that only those situations where the runner actually tried to score should be part of the equation. As such, if the runner ran on contact (and this won't be in every instance, and perhaps not even in the majority--there's no way to know in the data), he was successful in 92.4% of the time.
2. We include those 1,043 instances where the runner stayed on third for whatever reason. I tend against this because running on contact assumes the runner goes at the crack of the bat, but either way, choose for yourself. In this case, the equation would be 158/3123, or 5.1%.
The next three lines work the same way, with one additional caveat--I only looked at the runner on third, so if I did it correctly (and I think I did--all of this required some severe numerical manipulation, and I hope I didn't skip a step in one of the instances), we see similar numbers. if we accept my numbers, running on contact appears to work about 94.1% of the time, or about 19 times as often as it fails. Why this is is fairly obvious when broken down--consider the play that began this. In order for Ryan Zimmerman to throw out Joe Mather, the following FOUR things had to happen:
1. The ball had to be hit RIGHT AT Zimmerman, not two steps to his left or right
2. Zimmerman had to field the ball cleanly, not bobble the catch or double-pump the throw
3. He had to throw a strike to catcher Wilson Ramos
4. Ramos had to apply the tag
If ANY of these steps had more than the slightest variability, Mather scores. It would be fun to test and see how fast a typical player can run to home from third with a decent lead (this site suggests a player can run from home to first in 3.9-4.6 seconds depending on side of plate and obviously speed of player) and compare that with the seconds necessary to make the variations on a ball hit to 3rd. That data is being gathered, and how soon that will be available to regular Joes like me is anyone's guess, but an educated guess would be that in most instances, the player on 3rd can make it home safe prior to the play being made UNLESS near-perfect luck and execution occurs.
It's going to be a long Cubs season, and the heightened attention paid on an Opening Day with a new manager didn't do them any favors, but the odds seem to be in their favor. Four things had to be almost perfect in order for Mather to be out, and the data suggests this to be true.