Are you worried about the right thing?
Posted by Andy on April 1, 2010
As the season nears, it's always interesting to hear fans talk about their own teams. I hear lots of talk about hitting, such as whether the team can beat last year's home run total, and lots of talking about starting pitching and the closer. I almost never hear anybody talk about middle relief.
You might want to think a little more about your team's middle relief.
Check out the following plot showing number of runs scored by inning in MLB in both 2009 and the average over the years 2000-2009. This sort of data is available here.
Before I continue the discussion of middle relief, let me address some basic things about this plot:
- Note that I scaled the y-axis to include only 1500 to 3000 runs.
- The run scoring is always highest in the 1st inning, which makes sense since a team's best hitters always bat in the 1st.
- The run scoring is always lowest in the 9th inning, although this is mainly an artifact. Teams that win at home (other than walk-off wins) don't bat in the 9th so the run total is artificially low. I bet if I looked at run scoring by just the visiting team, scoring would still be a little low in the 9th inning thanks to the occasional presence of the home team's closer, who is usually one of the best pitchers on the team.
- After the 9th inning, the lowest-scoring inning is always the second inning (this is true in every individual season 2000 to 2009.) This is because most often, the lower part of the batting order comes up here. By the time we get to the 3rd inning, there is a lot of randomness in terms of which players are batting. Think about it this way--when does a team's leadoff hitter usually bat for the second time? If his team scored a couple runs in the first two innings, he'll probably bat in the second inning. If not, he'll usually bat in the third inning. if his team didn't get anybody on base at all, he won't bat until the 4th inning. So it's tough to predict who will bat when once you get past the first couple of innings.
Here's the entire point of this post: after the 1st inning, the inning that always has the most scoring is the 6th inning. That's true for every season 2000 to 2009. What are the reasons for this? There are three obvious ones I can think of:
- Starters tire significantly by the 6th inning and performance starts to worsen
- By the 6th teams are more likely to use a pinch hitter when a run-scoring chance presents itself, meaning a better hitter is likely to bat
- Once the starter leaves, the guys who relieve in the 6th inning are the worst pitchers on the team.
Most readers are probably familiar with Bill James' argument that closers should not be saved for the 9th inning but rather should be used in key situations in the 6th through 8th inning, when many games are decided. Instead, most teams are trotting out guys with ERAs well over 4.00.
Here's a different angle that approaches this same issue: think of teams to win the World Series in the last 15-20 years. Chances are you can name some of their middle relievers. How about Duane Ward in 1992, Greg McMichael in 1995, Mike Stanton and Jeff Nelson in 1998, Brendan Donnelly in 2002, Mike Timlin in 2004, Neil Cotts in 2005, and lots of other guys.
The bottom line is this--teams with lousy relievers allow more runs when it counts. The difference isn't huge, but it's a real trend. The middle relief guys probably make the difference of 3-5 wins per year--doesn't sound like a lot, but it could be the difference between your team winning 87 games and going home and winning 92 games and going to the playoffs as the wild card.
So, are you worried about your team's middle relief?
April 1st, 2010 at 1:43 pm
Very revealing graph. A clarification, which I think is more than a nitpick or semantics: most of those relievers you listed would be considered "setup" men, rather than "middle" relievers. I think they most often entered the games in the 7th or 8th, not the 5th or 6th. A lot of the times (though certainly not all) that the team's worst relievers are being used, the game may already be out of hand, or at least a comeback is a bit unlikely.
Is there a way to search how many high-leverage situations occur in the 6th inning? I can't think of one but there's so much stuff on the site now I may be overlooking something.
April 1st, 2010 at 1:50 pm
For what it's worth I checked how many 6th-inning appearances those relievers above made in those seasons (regular season only):
McMichael 12
Cotts 10 (I had completely forgotten about him, this was his only good season)
Timlin 7
Donnelly 6
Stanton 4
Nelson 4
Ward 0
All made more appearances in the 7th and 8th, and some even in the 9th.
Not any of this negates the point that teams can gain a great advantage by having good middle relief, because most don't.
April 1st, 2010 at 1:54 pm
Point taken JT. I picked some names pretty quickly and most of the are 8th inning guys.
I don't see a way of breaking out LI by inning but I might be missing it--I, too, have not integrated a lot of the new stuff yet.
April 1st, 2010 at 3:50 pm
The problem is two-fold, I think.
Money and laziness.
It's much easier for a manager to only use the closer in 'save' situations cause it's easier to explain why he didn't use him in a more crucial part of the game.
The other issue is now relievers get big contracts based on saves, so the closers and their agents only want them pitching in save situations. Plus pitching in the 9th (esp on the road) artificially lowers one's ERA since if you can't give up more than the winning run.
April 1st, 2010 at 4:07 pm
There is an "aLI" stat (Average Leverage Index) which you can sort by on the Pitching Win Probability page. The general trend seems to be the closers are at the top, then the setup men, then the starters, then the mop-up guys. There's a lot of names on the list, but spot checking a couple of my favorite teams verifies this trend (I think).
My thinking is that these mop-up situations are generally in the middle innings. The worst relievers on the roster have to pitch at some point or the better relievers will get worn out. I think it makes sense that these bad pitchers are generally used in this role.
I do agree with your main point that key make-or-break situations can occur as early as the 6th inning and its a shame when a team trots out someone like Bobby Keppel for those.
April 1st, 2010 at 4:22 pm
You know, questions like "Who was the guy who most often pitched the sixth for the 2003 Braves?" would be easily answered if there were a reliever depth chart, akin to the ones for the lineups. Sean would remember them from the old BBBA, but if you want to see which relievers came on in the seventh when leading by one, you could look.
April 1st, 2010 at 5:12 pm
Yes, closers probably could be used more efficiently, but as David points out they do still pitch the most important situations overall (as measured by LI), and their aLIs are not much different than top relievers of the Fireman Era. Now, part of this may be because if Gossage entered a sticky situation in the 7th and got out of it, he most likely would finish the game, even if his team then went up by 5 runs. This lowers his aLI. Modern 1-inning relievers don't have as much opportunity to face high- and low-pressure situations in the same game. Comparing the LI only at the point they entered the game might be more instructive here. I don't _think_ we can do that on B-R, but you can compare the number of games in which a reliever entered "high," "medium," and "low" leverage situations.
April 1st, 2010 at 8:08 pm
I think the point is also about a given team's worst relievers. If you throw washed up vets, unprepared rooks, or career scrubs (Quad-A guys) in that role, then you're going to have some bad middle innings. What would happen if a team had 2 or 3 of those guys on a roster? If your "worst" pitcher is significantly better than everyone else's worst, you gain a big advantage. And, while we've seen our share of over-inflated reliever contracts AND we know there is a certain volatility to performance, it certainly would seem to behoove a team to invest money into several top-notch MRs than one overhyped closer.
April 2nd, 2010 at 12:24 am
I don't think there really is such a thing as a top-notch MR. Anyone who performs well in that role gets "promoted" to a setup role, or maybe to the starting rotation.
April 2nd, 2010 at 5:19 am
Twisto-
I think that's because the mindset is to just dump a scrub there. Why not just develop/sign several top setup men, if the value is there? That's like saying your #8 hitter has to be a scrub, because if he was a 40HR guy he'd be your #3 or 4 hitter. Well, if you have 6 40HR guys, wouldn't that be more desirable? Not necessarily do-able, but it's foolish to concede a roster spot to a sucky guy just because that's what logic dictates if you have the means and ability to do more with it.
April 2nd, 2010 at 5:39 am
I don't think you've understood JT's point quite correctly. There aren't enough good pitchers in baseball to have a star at every position on the staff. Some teams might achieve that by overspending and we've seen the Yankees try to do it. They have had some good #8 and #9 hitters in recent years and they had excellent relievers during their dominance of the late 1990s. But there simply aren't enough good pitchers for all teams to have good middle relief. Remember my original point, though. The middle relief guys are always the worst on the team. Managers put their better players on the starting staff or as closers or setup guys. This doesn't mean the MRs are "scrubs", just the worst on the team.
April 2nd, 2010 at 8:26 am
[...] Are you worried about the right thing? (Baseball Reference). Andy serves up some good food for thought: “…teams with lousy relievers allow more runs when it counts. The difference isn’t huge, but it’s a real trend. The middle relief guys probably make the difference of 3-5 wins per year–doesn’t sound like a lot, but it could be the difference between your team winning 87 games and going home and winning 92 games and going to the playoffs as the wild card.” [...]
April 2nd, 2010 at 8:27 am
One thing we sabermetric types all need to remember is that the mentality of the managers and players enters into the equation here, and quite significantly at that. It's easy to look at past games and say when the closer should have been used. In practice, however, the manager can't possibly know when best to warm up his closer--for the 6th, 7th, 8th or 9th. Warmup pitches put strain on pitchers, and some guys such as Mariano Rivera, must be used when they warm up. He either comes into the game immediately or ends up throwing on the side and then is done for the night. Players' success also depends in large part on them understanding and preparing for their roles. We stat guys like to think it shouldn't matter, but we know that the stat show that it DOES matter. Some pitchers consistently do better starting an inning clean instead of coming in with runners on. Some guys do better when the score is close than when it isn't. My point is that you can't expect a pitcher to do optimally when his role is less well-defined--it goes against human nature. There comes a point when the optimal usage of pitchers based purely on statistics breaks down in the face of the realities of managing and psychology.