RBI by batting order position
Posted by Andy on March 17, 2010
I got curious about RBI by batting order position so here's some data along those lines. I got all the data from 2009 splits, such as right here for the Phillies.
To understand the above plot, look first at just the black line. This line is for all NL teams (I eliminated the AL from this analysis so as not to create a discrepancy for the 9th position in the batting order.) The data takes the number of RBI coming out of each batting order position and divides it by the total number of RBI for the entire lineup. So on average, #1 hitters drove in close to 9% of their team's runs while #3 hitters were around 15% and #5 guys were close to 13%. That black line is pretty smooth and shows what we would intuitively expect--the #3, #4, and #5 hitters drive in the most runs, with the cleanup guy leading the way.
A note on this graph--of course it should really be a bar chart and not a line graph as the numbers 1 through 9 are discrete. I just couldn't come up with a way of displaying all the data on a single bar chart that made it easy to see, so I did it this way.
The other 5 lines are for the NL East teams. I chose this division because of the range of teams in terms of their final records. Here are some notes about the plot above:
- The Phillies have the breakdown most resembling the NL average. (This is true by eyeball test but also in fact they have the lowest RMS error of this group of five--if you don't know what I'm talking about, don't worry about it!) They have a big peak at the #4 position--I call this "The Ryan Howard Effect." 🙂
- Second-closest is the Nationals, who of course had a much worse record than the Phillies. So I doubt that this sort of data can tell us much when viewed in a vacuum.
- The Marlins are the only team in this group not to get the largest production from their #4 hitter, instead peaking with #3, a spot held primarily by Hanley Ramirez. Jorge Cantu got 98 of his RBI in the 4th position but the Marlins got 105 RBI from Ramirez batting 3rd, as well as another 10 from Dan Uggla and a handful of others. The got only 6 RBI from guys other than Cantu in the 4th spot. There's a little value in this sort of analysis--Ramirez had 106 RBI overall and Cantu had 100. That seems to be a pretty balanced 3-4 combo but in fact the lineup wasn't quite as balanced as that. The biggest factor, though, might be that Ramirez was just so good last year. Cantu can't drive in runners that Ramirez has already cleared off the bases...
- What's going on at #6 in the order? It's odd that all 5 teams are clustered right around 11%. However, since all 5 are below the league average I'm sure that all the rest of the teams aren't also clustered right in that spot, so this is probably just a coincidence.
- The Mets and Braves both have a strange bump at #7, jumping way above what the rest of the NL did in that position. My first thought was that they each had a good player in that spot who took advantage of his RBI chances. Not so. The Mets didn't get more than 27 RBI from any one guy batting 7th. The Braves didn't have a single guy with more than 23! Recall that since each team is normalized to its own team total for RBI, this effect can be as much about poor performances in other batting order spots as it is about good performances in the #7 spot. Notice that the Mets did well below average in the #2 and #9 spots, and the high #7 total compensates. The Braves had a black hole at #3 (hello Chipper Jones!) and so the rest of the percentages were boosted elsewhere.
March 17th, 2010 at 9:30 am
Try showing the OBP and SLG or ISO as you've done above to see which teams most closely follow theories about batting order. You could also
plot the results against runs scored (or runs created ... Etc.) to see which teams order selection process worked best.
March 17th, 2010 at 12:17 pm
The Mets and Braves both have a strange bump at #7, jumping way above what the rest of the NL did in that position. My first thought was that they each had a good player in that spot who took advantage of his RBI chances. Not so.
My first thought was "Francoeur?" Again, not so — he wasn't very good in the #7 spot for the Braves (13 RBI), and he didn't bat in that position even once as a Met.
March 17th, 2010 at 5:44 pm
It looks like the Marlins are actually a bit higher at #9 than at #1 and comparatively much higher than any of the other four teams at #9. Did they use line-ups with the pitcher batting elsewhere (maybe 7th, since they are lower than the other four in that slot)? Or was one of their pitchers a pretty good hitter? (Dontrell's gone, so he can't be the one.) Or did they get great production out of their pinch hitters? Hmm, maybe the opposition should have pitched around that guy in the 9-hole to face the lead-off man with runners in scoring position!
March 17th, 2010 at 7:54 pm
@3
Actually, the Marlins hit worse than league average in the #9 slot (sOPS+ 82)... their pitchers were pretty bad (sOPS+ 68). On the other hand their leadoff batters (mostly Coghlan) had a decent year (sOPS+ 106).
But RBI's are situational, so even though Coghlan hit well overall (.321/.390/.460) and with RISP (.352/.448/.477), he didn't have that many opportunities because the pitchers were never on base so his RBI totals projected to a paltry 59 per 162 which is pretty low for someone with that much pop.
As for the #9 slot, the Marlins had above average production from the #8 position (sOPS+ 122) so there was more people on base there than usual. Also, they pinch-hit well (sOPS+ 140)... not sure how often that occurred in the #9 slot but anecdotally, you would guess that happens quite a bit with RISP in the late innings.
That's the trouble with RBI's. They often measure the performance of the guys in front of a batter rather than just the performance of that batter.
March 18th, 2010 at 8:38 am
Yeah, keep in mind that poor performance by the #7 and #8 hitters can lead to more RBI opportunities by the #9 hitters. Even if the #9s don't hit well, if there are enough extra RBI opportunities, they could get more RBI than another team's #9 guys.
I used the Batting Event Finder to get some numbers.
Marlins #9 hitters had 183 plate appearances with runners in scoring position, yielding a .165 batting average with 37 RBI.
Phillies #9 hitters had 189 PAs with RISP, .160 BA and also 37 RBI.
Mets: 160 PAs with RISP, .185 BA with 23 RBI
Braves: 179 PAs with RISP, .185 BA with 33 RBI
Nats: 159 PAs with RISP, .239 BA with 37 RBI
Here are the seasonal RBI totals for each team (from all batting order positions):
FLA 727
PHI 788
NYM 631
ATL 700
WSN 685
All the teams all had very similar numbers of PAs and batting average (except for the Nats, whose higher BA didn't seem to produce more runs in this small sample.) The Mets generated a smaller number of RBI but they had fewer overall RBI as well so the percentage is roughly equivalent. The Marlins tied for the most RBI from the 9th hole with RISP but with a a smaller number of RBI than the Phillies as well as below league-average performance at all other holes except #3 and #5, their #9 percentage is boosted.
The Nats are weird, despite also tying for the most RBI in those circumstances, their percentage is quite low. This is probably due to well above-average production they got from #2 and #8, causing #9 to form a smaller percentage.