Back around the AS break, I took at look at the Nats' Run Expectancy, which is table of how many runs you expect to score (on average) starting from a given number of outs and men on base. At the time, I noted that the Nats were substantially worse at scoring runs with two outs or bases-loaded than MLB-average. Of course, the only stats I had for comparison combined NL and AL. With the abomination that is the DH, I wondered if that skewed things a little, so I finally went back and figured out league-specific run expectancies.
Okay, here's what I calculate for run expectancy for all of MLB (combined AL/NL):
| On Base | no out | 1 out | 2 out |
| --- | 0.50 | 0.28 | 0.10 |
| 1-- | 0.86 | 0.53 | 0.22 |
| -2- | 1.08 | 0.67 | 0.30 |
| --3 | 1.28 | 0.93 | 0.39 |
| 12- | 1.45 | 0.90 | 0.43 |
| 1-3 | 1.73 | 1.14 | 0.55 |
| -23 | 1.98 | 1.39 | 0.55 |
| 123 | 2.21 | 1.53 | 0.73 |
For example, this means that (on average), if a team has a man on second with one out, it will score 0.67 runs before the inning ends. Similarly, if a team has a man on first with no outs, it'll score 0.86 runs on average before the inning ends (this is why stat-heads don't like sac bunts... you're "losing" about 0.2 runs by trading an out for a base).
Charts for each league and for the Nats after the jump!
This is slightly different than the run expectancy I found on Baseball Prospectus back in July. However, most of the numbers are within 0.03 runs of theirs, and none differ by more than 0.06 runs. I'm not sure whether we're looking at exactly the same set of games (theoretically, my data set is all plate appearances in games played in 2009 before the All Star Break), but it's close enough to make me confident that I don't have anything too screwy in my data or calculations.
Here's what you get for AL run expectancy:
| On Base | no out | 1 out | 2 out |
| --- | 0.53 | 0.30 | 0.11 |
| 1-- | 0.91 | 0.56 | 0.23 |
| -2- | 1.06 | 0.68 | 0.31 |
| --3 | 1.36 | 0.89 | 0.37 |
| 12- | 1.52 | 0.92 | 0.45 |
| 1-3 | 1.87 | 1.20 | 0.54 |
| -23 | 2.14 | 1.46 | 0.61 |
| 123 | 2.25 | 1.59 | 0.76 |
And here's the NL run expectancy:
| On Base | no out | 1 out | 2 out |
| --- | 0.48 | 0.26 | 0.09 |
| 1-- | 0.82 | 0.51 | 0.21 |
| -2- | 1.10 | 0.66 | 0.29 |
| --3 | 1.22 | 0.97 | 0.41 |
| 12- | 1.40 | 0.89 | 0.42 |
| 1-3 | 1.60 | 1.08 | 0.55 |
| -23 | 1.81 | 1.32 | 0.50 |
| 123 | 2.18 | 1.47 | 0.70 |
This seems about like we'd expect, right? With 9 hitters in the lineup instead of 8, the AL scores more runs generally. (Note that I haven't tried to factor out inter-league play, this is purely by team... Still, the vast majority of play is within leagues). There are only two cases where NL RE is higher (1-3 with 2 out and --3 with 1 out), with one of the cases being only 0.01 runs more.
Note that you start every inning with no one on and no outs, so you'd expect the average NL team to score 9 x 0.48 = 4.32 runs per game, while the average AL team scores 4.77 runs game (this will be slightly off because of extra-inning games, of course... D'oh!).
Recall the Nats' RE (again, this is only counting games before the AS break):
| On Base | no out | 1 out | 2 out |
| --- | 0.47 | 0.25 | 0.08 |
| 1-- | 0.77 | 0.42 | 0.14 |
| -2- | 1.10 | 0.67 | 0.30 |
| --3 | 0.33 | 0.81 | 0.40 |
| 12- | 1.69 | 0.74 | 0.37 |
| 1-3 | 2.00 | 0.97 | 0.20 |
| -23 | 1.90 | 1.44 | 0.17 |
| 123 | 1.89 | 1.29 | 0.47 |
The Nats don't look quite as bad in comparison to the rest of the NL, but the original point stands: load the bases or put a man on 3rd and teh suk ensues! Or at least, it did while Manny was still in charge. I'm sure things look a bit better on the bases-loaded line with Willingham's two grand slams averaged in, and my impression is that situational hitting has been better/luckier generally than in the first half. We'll see how things look at the end of the season.
Again, remember the usual caveats: this is interesting, but not robust enough to be determinative. Some of the entries in the table are based on a fairly small number of PA (123 no out, 1-3 no out, -23 no out, and --3 no out are all based on 11 or fewer PA). I'm planning to revisit this at the end of the season (downloading all data is a pain, so I've decided I'm only doing it twice a year), but I've been thinking about different things to do with the data that I have over the next couple of weeks.