OBP from the leadoff spot / RBIs with two outs
Posted by Andy on September 27, 2007
I wanted to use some league-wide 2007 batting splits to emphasize a couple of points.
First, all the data I am citing can be found on the 2007 AL batting splits page and the 2007 NL batting splits page. This type of data is among my most favorite at B-R.com because there's a lot you can glean from such large sample sizes, and yet this data isn't easily available in many places.
OBP from the leadoff spot
First, let's take a look at a very basic split: batting with empty bases versus batting with runners on.
In the NL, OBP while batting with the bases empty is .323 on over 53,000 plate appearances. While batting with runners on, however, NL OBP is .348 on over 44,000 PAs, a difference of 25 points.
In the AL, the picture is similar. With bases empty, the OBP is .326 on over 46,000 PAs, and with runners on the number switches to .352 on over 39,000 PAs, a 26 point increase. (If you happened to notice that the NL has more PAs, remember that's because it has more teams!)
So, you ask, what does this have to do with OBP from the leadoff spot? Well bear with me for a little more split data.
Let's look at OBP numbers from the 9th position by league. In the NL, 9th-place hitters have an OBP of just .241, since it's mainly pitchers and pinch-hitters. In the AL, 9th-place hitters have an OBP of .301, which is below league average but still way, way better than in the NL.
So what does this mean? It means that in the NL, leadoff hitters must hit with fewer men on base, on average, than leadoff hitters in the AL. Why? It's simple deduction. In the NL, 9th-place hitters make out more often, failing to get on base and/or ending the inning, which tends to bring leadoff hitters up with the bases empty more often. (Obviously, leaderoff hitters in both leagues always bat with the bases empty at the beginning of each game.) Since we saw above that production numbers are lower with bases empty versus runners on, we'd expect to find that NL leadoff hitters have slightly lower numbers than AL leadoff hitters.
And the numbers support this. OBP for NL leadoff batters is .342, while OBP for AL leadoff hitters is .350. It's "just" an 8-point difference, but it's the largest difference among the most important spots in batting order. Second-place hitters in the NL/AL are .339/.345 (6 points), third-place hitters in NL/AL are .368/.366 (2 points), fourth-place hitters in the NL/AL are .364/.362 (2 points), and fifth-place hitters in NL/AL are .339/.340 (1 point.)
See what I'm getting at here? There's a cascade affect. Pitchers in the NL create a hole in the lineup that make it slightly tougher for leadoff hitters to get on base, which in turn makes it very slightly tougher for 2-hole hitters to get on base, but by the time we get to the heart of the order (3-4-5) the problem has corrected itself.
All of this is a long-winded way of saying that NL leadoff hitters shouldn't be held to quite the same standard as AL leadoff hitters. And it's a really long-winded way of saying that Jimmy Rollins should win the NL MVP despite the fact that his OBP isn't as high as you might expect.
RBIs with two outs
Whew, that was a long one. Are you still with me?
Here's a much simpler analysis. You know how frustrating it is when you favorite team nearly gets out of an inning, registering two outs with runners on base, but then gives up the big hit that scores a couple of runs and loses the game? It's frustrating, isn't it? It seems like scoring with two outs is really hard, and that teams should be able to escape these innings more often.
Well, there's one big thing we have to remember about 2-out baseball: baserunners can run full speed. When a fly ball or line drive goes to the outfield, all baserunners are going full speed without having to wait to judge whether it's going to be caught. This is a huge advantage, and it's why two-out singles and doubles so often score runners from 1st or 2nd base. It's also why so many batters have such great numbers with 2 outs and RISP--because if you get a hit, you're guaranteed to rack up the RBIs.
Anyway, here is the split data that backs up what I am saying.
Looking first at the NL data, we see that plate appearances are almost evenly spread among outs: There have been 33,675 PAs with no outs, 32,459 with one out, and 31,586 with two outs. But there are many fewer hits as the outs increase: 8269 hits with no outs, 7719 with one one, and just 7075 with 2 outs. BA, OBP, and SLG all drop off as outs increase too.
Nevertheless, with fewer PAs and fewer hits, there are nearly as many runs scored with 2 outs as there are with 1 out, and way more than with 0 outs. The data are 2716 R with no outs, 4596 with 1 out, and 4523 with 2 outs. What I cannot discern is exactly how many RBI opportunities there were with each number of outs. It follows that on average there would be more baserunners with 1 or 2 out than with 0 outs, when many of the at-bats lead off an inning and therefore there are no baserunners.
The AL data is very similar and I'll leave it up to you to take a look at it.
The moral of the story, though, is that two outs & runners on is a very dangerous situation, and a lot of teams score a lot of runs in such cases.
September 27th, 2007 at 8:25 am
Great Article. If the Phillies make the playoffs JRoll should be the MVP.
Nothing better than the SOTD.
September 27th, 2007 at 8:40 am
What if they don't make it
September 27th, 2007 at 1:00 pm
Super use of stats to carefully analyze the game -- nice work!
September 27th, 2007 at 1:42 pm
Very well done on your first point. I was aware of many of those facts, but had never put them together to conclude that NL leadoff mean have an inherent disadvantage. Nice job.
On your 2nd point, I somewhat disagree. I don't think the advantage of having runners moving fullspeed with 2 outs is that important compared to having additional runners on base, period. As an inning progresses, batters either get out or get on. I think that overall there are far more RBI opportunities with 2 outs than with 0 outs, because most AB with 0 outs are leading off an inning. "two outs & runners on is a very dangerous situation," true, but still less dangerous than zero or one out & runners on.
September 27th, 2007 at 1:54 pm
Certainly I agree that I did not word that last phrase very well, and thanks for correcting it.
What I'd really like to see is a side-by-side comparison of various baserunning situations with 0, 1, or 2 outs. For example, doubling with the bases loaded. With 2 outs, this is almost always going to score 3 runs (unless it's a ground rule double or a small fraction of other cases) whereas with 0 or 1 outs, I would think that the incidence of scoring only 2 runs is higher.
I am willing to bet that virtually any scenario shows more runs scoring with 2 outs...but we don't have that kind of detailed data here. Folks like those over at THT would need to look into this.
September 27th, 2007 at 3:24 pm
"For example, doubling with the bases loaded. With 2 outs, this is almost always going to score 3 runs (unless it’s a ground rule double or a small fraction of other cases)"
Andy,
I think you're a little optimistic with "almost always going to score 3 runs". I looked up 20 instances at random in 2007 where a batter hit a bases loaded double (not including ground-rule doubles) with 2 outs, and only 9 of the 20 doubles plated 3 runs. I know this is a small sample size, but I'm not a subscriber to the play index, so this is about the best I could do.
September 27th, 2007 at 3:35 pm
Hey--that's a great idea. I used the PI Team Batting Event Finder to look up doubles with the bases loaded.
0 outs: 39 occurrences, 32 times it scored 2 runs, 7 times it scored 3 runs
1 out: 108 occurrences, 68 times it scores 2 runs, 40 times it scored 3 runs
2 out: 95 occurrences, 42 times it scored 2 runs, 53 times it scored 3 runs
As you can see, with 2 outs, it scores 3 runs far more often (56% of the time) than with 1 out (37%) or no outs (18%.)
I didn't even correct for ground-rule doubles, almost all of which score 2 runs (although there are some placed, including Tampa Bay I think, where a ground-rule double with the bases loaded bats in 3.)
Thanks for the suggestion.
September 27th, 2007 at 5:06 pm
Rollins has had a hot month. I was curious to see how he compared with other players in the NL pennant race. Holliday is the only other one to be having a great month. Whereas Reyes is slumping terribly.
Over the last 30 days:
*Rollins: 44 for 133 (.331) with 23 runs, 8 homers, 19 RBIs and 11 SBs.
*Holliday: 32 for 95 (.337) with 26 runs, 12 homers 29 RBIs and 2 SBs.
*Beltran: 32 for 106 (.302) with 23 runs, 7 homers, 25 RBIS and 5 SBs.
*Tulowitzki: 29 for 105 (.276) with 21 runs, 5 homers, 21 RBIs and 1 SB.
*Utley: 35 for 111 (.315) with 22 runs, 4 homers, 17 RBIs and 2 SBs.
*Helton: 41 for 103 (.398) with 19 runs, 3 homers, 18 RBIs and 0 SBs.
*Alou: 44 for 103 (.427) with 21 runs 4 homers, 16 RBIs and 0 SBs.
*Wright: 37 for 107 (.346) with 21 runs, 7 homers, 19 RBIs and 6 SBs.
*Howard: 27 for 103 (.262) with 16 runs, 9 homers, 22 RBIs and 0 SBs.
*Soriano: 34 for 123 (.276) with 20 runs, 13 homers, 23 RBIs and 1 SB.
*A Gonzalez: 31 for 115 (.270) with 18 runs, 4 homers, 17 RBIs and 0 SBs.
*Rowand: 36 for 117 (.318) with 21 runs, 5 homers, 15 RBIs and 0 SBs.
*Reyes: 25 for 114 (.219) with 19 runs, 3 homers, 9 RBIs and 7 SBs.
September 27th, 2007 at 5:26 pm
Good write-up on the OBP from the leadoff spot, but I still don't buy the Rollins love. With last night's 3-for-4, he's barely above the league, leadoff OPB average (still below it when you consider park effects). Yes, he has value and he's having a good (or great) year, but his value is in his SLG, not his OBP. So, not only is his OPB ill-suited for his spot in the line-up and hurting the Phils, his greatest value (SLG) is being wasted by having less runners on base for him. If he puts up his rate stats and the SBs from the third spot in the order, he's probably the NL MVP. But, not from the leadoff spot.
It's the same issues that's cause the SABR members to harp on Alfonso Soriano's employers for years. Put Rollins third in the line-up, where his power will be put to better use, his speed will still be an asset in front of the sluggers and his OPB won't hurt you. Leave the top of the order to the guys who get on base enough to NOT lead MLB in outs.
September 27th, 2007 at 6:00 pm
I totally agree that Rollins should bat third, and note that he did have 103 PAs hitting third this year when Utley was injured. (see here. He had a lousy .694 OPS hitting third, interestingly.
I can't hold it against Rollins that he was used improperly.
September 27th, 2007 at 10:46 pm
I haven't studied this too much, but I can't see how Rollins is hurting Philly by leading off. They don't really have a much better candidate on their team. Everyone who gets on base more than Rollins also has a lot of power. The one guy who might fit is Victorino, but if you bat him leadoff, you're only doing it because he seems to fit some kind of leadoff mold. There's 5 regulars who are clearly better hitters, so why choose him to get the most AB? I don't know if Rollins should be MVP, but he does have the most runs created on the best offense in the league, while playing SS, so that's gotta count for something. (By my BaseRuns calculation, Utley is only about 10 runs behind Rollins with over 100 fewer outs, so he certainly deserves consideration.)
September 28th, 2007 at 9:38 am
Minus the first inning, do players batting in the #1 spot really have more at-bats leading off an inning than players batting in other spots of the line-up?
September 28th, 2007 at 12:54 pm
I'm not sure this is easy to calculate, but I did something like it just for the Phillies. I figured out the total number of times in 2007 a given spot in the order has come up with no outs and nobody on base. There is no option I could see to specify "leading off the inning." Obviously, if a guy comes up after the true leadoff hitter hit a solo homer, then there are no outs and nobody on base. SO these numbers are not exact but do give a general sense.
Anyway, here are the number of such PAs for the Phillies in 2007, going by each spot in the batting order:
1st: 323
2nd: 152
3rd: 121
4th: 156
5th: 156
6th: 174
7th: 143
8th: 144
9th: 137
Obviously, so far this year there have been 159 times that the #1 hitter has come up to lead off the game, so taking those out leaves 164, still good for second on team. To me, that's a high number, since the leadoff hitter will very rarely lead off the second inning, and somewhat rarely lead off the third inning.
Let's do another NL team, and two AL teams for the heck of it.
I'll pick the Mets, a team with the same record as the Phillies at the moment:
1st: 318 - 159 = 159
2nd: 142
3rd: 115
4th: 174
5th: 158
6th: 170
7th: 149
8th: 142
9th: 121
The corrected leadoff value of 159 is 3rd on the team. By the way, you notice how rare it is for the 9th place hitter to bat with the bases empty and no outs. This is because even with two outs, pitchers often work around the 8th place guy to get to the 9th-place hitter. I always thought this was dumb...the 8th guy is usually not such a great hitter anyway (Hello, Alfredo Griffin!), and if you get him, you get the pitcher leading off the next inning.
Now, two AL teams:
First I'll take Detroit because they too have the same record:
1st: 307 - 159 = 148
2nd: 138
3rd: 128
4th: 161
5th: 156
6th: 153
7th: 158
8th: 135
9th: 149
Notice two things. Now the leadoff hitter (corrected) is down to 6th most PAs, and the 9th position is no longer much lower than the rest. (This all backs up my original argument above.)
And one other AL team...let's pick whoever is middle-of-the-pack for runs scored: I take Toronto:
1st: 295 - 159 = 136
2nd: 137
3rd: 126
4th: 179
5th: 181
6th: 143
7th: 143
8th: 148
9th: 140
Wow, their #1 hitters are next to last in no-out, no-runner PAs. That means their leadoff guy is hitting with runners on more often than just about anybody else on the team, which really helps their averages.
I'm going to conclude that my overall point has been proven.
September 28th, 2007 at 2:49 pm
Andy,
Maybe I am missing something here - I'm not a regular observer of baseball statistics. I'm not sure showing that the OBP with runners on is higher when men are on-base than when they are not proves that it's more difficult to hit with the bases empty.
Couldn't it be that OBP are higher with runners on precisely because the runners on-base OBP stat is overrepresented by the *better* bats in the lineup, while the bases empty stats are overrepresented by the weaker ones? The lineup is designed with such a situation in mind.
Is this what the 8% difference (as opposed to 6 points for 2nd batters) between NL and AL lead-off men is supposed to substantiate?
September 28th, 2007 at 2:55 pm
You've got to read the entire thread above to fully get it, but the difference in OBP with bases empty vs runners on is due in large part to effect such as larger holes in the infield (particular between 1B and 2B) when the defenders need to cover bases, and the fact that the pitcher needs to pitch out of the stretch, which is less effective for many pitchers, in terms of batting average against.
September 28th, 2007 at 3:23 pm
This is what I read: http://www.baseball-reference.com/sotd/archives/324
Is there more to the thread than this? I didn't see anything about holes in the infield or pitching from the stretch.
In any case, it may be true that it's easier to reach base with runners on for other reasons, but I'm not sure the OBP comparison provide good evidence of the claim. For instance, would a stat sheet representing OBP with runners on, where 65% of the list were at-bats by Ryan Howard and Aaron Rowand and 35% were at-bats by Abraham Nunez and Carlos Ruiz, compared to a stat-sheet which represented OBP with no runners on, where the percentages were reversed (*worse* hitters 65%), not be skewed by the varying proportions of good batters to worse batters in each? If 3-4-5 hitters get more opportunities with runners on, then it sounds like 3-4-5 hitters are represented more often in the OBP with runners on stats than they are in the OBP with no runners on base stat. I'd just think that would skew the comparison. They're better hitters.
September 28th, 2007 at 4:56 pm
Yeah, I think you're right. There are many things about my cursory analysis that are shortcuts, and that's one of them.
September 28th, 2007 at 6:12 pm
Generally speaking, pitchers are tougher against hitters when there is no one on base. There are no base-runners to distract the pitcher or catcher, and fielders can play to the batter's specific scouting report. With runners on base, infielders need to shift, outfielders might have to play in, etc.
September 28th, 2007 at 9:11 pm
"Couldn’t it be that OBP are higher with runners on precisely because the runners on-base OBP stat is overrepresented by the *better* bats in the lineup, while the bases empty stats are overrepresented by the weaker ones? The lineup is designed with such a situation in mind."
That's a very good point, and it probably accounts for some part of the difference. But if you look at individual players' stats with bases empty vs. runners on, I believe most hit better w/ runners on, so there is an inherent advantage to hitting in those situations.
September 29th, 2007 at 7:38 am
Very interesting. Batters in the #1 spot are less likely to be lead-off hitters in an inning than batters in the #4, #5, #6, and #7 spots. So being a "lead-off hitter" is a misnomer, and the "talent" is over-rated.
What batting in the #1 spot *does* provide as an advantage is more plate appearances. With that said, wouldn't it be better to have Frank Thomas (and the like) bat in the #1 spot?
September 29th, 2007 at 10:29 am
KT, that's true in the AL only, I think. And the issue with batting Frank Thomas first is: since Thomas is a power hitter, it's still probably better to let him bat later in the inning when his power hits might bat in more than just himself.
By the way, if you look simply at straight PAs (regardless of # of outs or anything else), the higher the position, the more PAs it gets.
For example, check out the Phillies batting splits so far this year:
http://www.baseball-reference.com/pi/bsplit.cgi?team=PHI&year=2007
I won't put all the numbers here, but suffice it to say that the #1 spot got 779 PAs and the #9 spot got 643 PAs, and it's a smooth linear scale in between. The rule of thumb I remember was that each spot in the order was worth 15 more PAs per season, and that's almost exactly correct in the Phillies' case.
Also, you may recall, in the past when certain teams had a player going for a seasonal record--such as getting 200 hits--if they were out of contention, they would sometimes move that player to the leadoff position for the rest of the game, just to get him more PAs.