Team On-Base Percentage and a Balanced Lineup
Teams that get on base often score more runs than those that don’t. We know this, and it comes as no surprise. In 2013, the Red Sox had the highest team OBP (.349) and also scored the most runs in MLB. The Tigers had the second-highest team OBP (.346), and they scored the second-most runs. Team OBPs can tell us a lot about the effectiveness of an offense (obviously not everything), but they can also be misleading if proper context isn’t applied.
The Cardinals scored 783 runs in 2013, good enough for third in MLB. The rival Reds scored 698 runs, 85 fewer than the Cardinals. There are many reasons for this gap in runs scored, but I would like to examine just one of them. The Cardinals had a team OBP of .332 while the Reds had a team OBP of .327. On first look, it appears that the Cardinals and Reds got on base at a similar rate. But a major difference exists below the surface. Take a look at the chart below of the top eight hitters by plate appearance for both teams (Chris Heisey gets the nod over Ryan Hanigan as to not have two Reds’ catchers on the list).
| Reds OBP | Cardinals OBP |
| Joey Votto .435 | Matt Carpenter .392 |
| Shin Soo Choo .423 | Matt Holliday .389 |
| Jay Bruce .329 | Allen Craig .373 |
| Todd Frazier .314 | Yadier Molina .359 |
| Brandon Phillips .310 | John Jay .351 |
| Devin Mesoraco .287 | David Freese .340 |
| Zack Cozart .284 | Carlos Beltran .339 |
| Chris Heisey .279 | Pete Kozma .275 |
The difference is quite evident. The average OBP in 2013 was .318. Seven of the top eight Cardinal hitters got on base at an above-average clip. Besides the pitcher, there is one easy out in that lineup. The Cardinals maintained a ridiculous batting average with RISP, but that matters much more because they always had people on base.
On the other hand, the Reds had two on-base Goliaths. Joey Votto and Shin-Soo Choo camped out on the bases. They became one with the bases. The problem was that the Reds had only one more player with an above-average OBP, Jay Bruce at .329. The other five players struggled to get on base consistently. Three of them had OBPs under .300.
So while the Cardinals achieve a high team OBP through balance, the Reds had two hitters who significantly raised the team OBP. Take Votto and Choo away, and the other six Reds on this list have a combined OBP of .305. That is a staggering low number for six of the top hitters on a playoff team.
What does this teach us? Well, team OBPs do not provide insight into how balanced a lineup a team has. The Reds would be foolish to think they have a lineup that gets on base enough to be an elite offense. With the loss of Choo, the Reds offense may struggle to produce runs at a league-average clip as Votto and Bruce could be stranded on base countless times.
A balanced lineup was a major factor in the Cardinals scoring the most runs in the National League. Their team may have had an excellent .332 OBP, but their top eight hitters by plate appearance had a .355 OBP. As a group they were excellent. The Red Sox were similar in that their top eight hitters by plate appearances all had above-average OBPs with Stephen Drew coming in eighth at .333. Think about that! The Red Sox eighth-best hitter at getting on base was 15 points above league average.
Even though the Reds finished 6th in team OBP in 2013, their on-base skills were lacking. While the Cardinals had only a five-point advantage in team OBP over their rival, they were much more adept at clogging the bases. Team OBPs are great, they just don’t always tell the whole story.
Interesting argument, I wonder if this holds up over a larger sample. And if so, just how important is it?
This is similar to the concept in football where not only is the average yards per carry important, but the lower the standard deviation of a team’s yards per carry the better. If this is indeed true–that low standard deviations in team OBP cause for a team to outperform their average–we’ve just unearthed quite a little gem here.
You’re right that more research is needed here. This limited study bore out of my frustration from hearing Reds fans blame Joey Votto for any offensive shortcomings the Reds did have in 2013. It is at least interesting to note that a team can finish in the top six in team OBP and NOT actually be good at getting on base as a starting lineup.
I’ve done some preliminary research and so far it seems that this effect–using all teams from 2013–does in fact exist. Would you mind if I did a more extensive write-up on this?
What I’ve found so far is basically that (spoiler alert) 22 out of the 30 teams that had an above average/below average standard deviation of OBP had a corresponding above/below average runs scored number. The sample of 30 isn’t conclusive, but it does hint and nudge and point towards this being a real thing.
Go ahead! Sounds like you have some useful information toward a good article. I appreciate you asking.
So if I’m understanding correctly, your argument is this: Teams that have less variation among their primary hitters’ OBP will score more runs than we might expect based on their aggregate performance (and vice versa). Is that fair?
Because I tested that and I don’t see it.
Firstly, I calculated the residual of a team’s actual runs scored compared to the number of runs we would have expected them to score given their team OBP. (obviously other factors affect run scoring, but introducing them would be questions about their correlations with OBP). The Cards scored 36 runs more than their OBP predicts; the Reds 24 runs fewer.
Secondly, I then took the Standard Deviation of OBP, by team, for all batters with 300+ PA with that team. The cards were more or less average, with a SD of .038 compared to the league average of .034. The Reds were off the chart, perhaps historically so, at .063.
So, the question is: Is there a correlation between a team’s OBP variation and it’s run scoring residual. If yes, it would mean that more (or fewer) variation produces more (or fewer) runs than we would think based on their aggregate performance.
So I then regressed that OBP Standard Deviation against the OBP-Predicted Runs residual. The result? Essentially zero correlation (an R2 .0009). And even removing the Reds, since they’re such an outlier, leaves an R2 of .0044.
Here’s the full chart: http://i.imgur.com/GYCZNwI.jpg (if link does post, I’ll be cross-posting on the RedsZone forum where the image will show up)
Now, we could certainly run the SD against a better runs estimator, but I don’t think you’re going to see the numbers change much. Using OPS as the runs estimator, the R2 skyrockets to a still paltry .0243.
Let me know if I’m doing something I shouldn’t here, but unless I really missed something, there’s no detectable effect here. We could run it with more data, but I’d be shocked if a significant effect reared it’s head. If the Reds, one of the least balanced teams of this generation don’t under-perform by much, it seems unlikely to be a real thing.
Ultimately, I agree, team OBPs don’t tell the whole story. But if you’re looking for the rest of the story, your next stop should be SLG. “Balance”, as it were, is likely pretty far down that list.
I think you are late to the game on this one. Brandon Firstname already covered the variance issue quite well. http://www.fangraphs.com/community/team-construction-obp-and-the-importance-of-variance/.
My main point was way more simplistic: team OBP can be misleading. The Reds had a similar OBP to the Cardinals and close enough to the Red Sox that you would believe the Reds were good at getting on base when really they weren’t that great at it. Two players were VERY good at getting on base, but the lack of OBP elsewhere was staggering.
I did say “A balanced lineup was a major factor in the Cardinals scoring the most runs in the National League.” But it was more of a hypothesis than anything else that Brandon did a fuller study on. Essentially, I agree with you on the variance issue. I just didn’t intend to say something that definitive with this article, but I did intent to suggest it would be a nice study based on the outcomes of these three teams.
Thanks for the clarification and link to Brandon’s follow-up. I made the mistake of not noticing that your article was a bit old — followed a link from a new post on our forum. Your basic point is well taken and Brandon definitely covered the hypothesis you posed.