The end of the bench

Sometimes, the little things matter. The ability to turn the double play. Taking the extra pitch. Moving the runners over. Sometimes those little things, which can be overlooked easily, add up. Done once or twice, they don’t mean much. But over the course of the season … yeah, they can add up. Sometimes, they can even make the difference between going to the postseason and staying at home.

Those little things, however, are all little things that happen during the game. There are other little things that matter, like the back end of the roster. Every team has 25 roster slots, and we all know that all those players get used, but it’s easy to overlook the impact of that 25th man.

Oh, everyone understands the importance of starters. They’re out there every day. It’s hard to miss them. And we all recognize the importance of the five main starting pitchers in the rotation. Everyone is aware how important the closer is—and beyond him, the main set-up men. But what about the other players?

What can we say about the rest of the roster? What impact does it have on the course of the season? This is something I looked up in midseason, and now that the year is over, I want to take a look back at how it played out over the course of the full year.

Let’s look at how much production all 30 teams got from the back end of their benches and then look at how it affected them over the course of the season. There are three things we’re looking at here: 1) marginal position players, 2) marginal starting pitchers, and 3) marginal relievers. In all cases, they are defined primarily by what they are not rather than by what they are.

A marginal player is someone not in the core of the roster. A core position player is one of the eight guys listed as a starter for his team on (or nine guys, in a DH league). All other position players are marginal.

A core starting pitcher is a guy who is one of the top five in the team in games started. All other starts come from marginal position players.

A core reliever is one of the top five arms in relief innings. All other relief work comes from marginal relievers.

Let’s go group-by-group and then combine them all. Obviously, by definition marginal players are below average. You’d expect them to be below average. Aye, but they’re not all equally below average. Some teams, in fact, get lucky and have benches that actually do a bit better than expected. Other benches are terrible, even by the standards of fringe players.

Marginal position players

Let’s use the old Bill James stat Runs Created for these guys. More specifically, we’ll see how many runs each bench created versus how many outs each created. Then we’ll compare that to how many runs they would’ve created if the marginal players hit like league-average players, which typically deliver 20 runs worth of value.

This system has the benefit of not only seeing how they did, but seeing how they did in the context. Oh, and it adjusts for park (of course). Finally, please note we’re excluding pitcher batting. Position players are supposed to hit as part of their job. Pitchers aren’t.

Again, we’d expect benches to hit worse than average. After all, by definition, backups should be below average. But the question is how much worse than average some of these benches are.

Well, do all the appropriate mathematical hokey-pokey, and here are the results (listed in terms of runs):

Team	Bench
BOX	 63.0
CLE	 48.1
CHC	 37.3
PIT	 36.1
SDP	 25.6
TEX	 24.9
TBR	 22.2
ARI	 19.9
NYM	 16.2
ATL	 13.5
LAA	 12.9
OAK	 10.4
KCR	 -3.8
SEA	 -3.9
MIN	 -7.6
DET	 -7.8
LAD	 -7.9
SFG	 -8.7
STL	 -8.8
PHI	-10.5
TOR	-14.0
CIN	-14.9
MIL	-15.1
MIA	-16.4
COL	-17.0
BAL	-25.6
NYY	-25.7
HOU	-32.6
DCN	-42.5
CWS	-45.3

Seven teams had above-average benches offensively (and Tampa was within a run of average). I wouldn’t have guessed that many, but there you go. The Cubs, please note, actually had better offensive production from their bench players than their starting players. That’s a sign of a really good bench … or really bad starters. Or both.

As for the playoff situation, all three clubs vying for the two AL Wild Card slots had terrific batting benches, but the performance of Cleveland’s backups may have made the difference for the Tribe.

If you were to just look at benches in a RC/27 outs basis—that is, as a rate stat and without regard to playing time—the Nationals would have the worst bench in baseball. Luckily for them, they didn’t have to use their bench as much as Houston or the White Sox relied on theirs.

Homestretch: The 1967 AL Pennant Race, Part 3
A tight race shows no signs of letting up.

If you add it up, benches combined to create 606 fewer runs than league-average hitters would create in that same span. Thus, an average bench was 20 runs below average. So Boston gained about six wins more than a typical team would’ve thanks to its bench plus (plus four instead of minus two).

Marginal starting pitchers

For these guys, the base stat is ERA. Take all starts by pitchers not in the top five in games started for their team, and figure out their combined ERA. Then figure out how many runs league-average pitchers would allow in the same number of innings that these guys actually pitched. And, as always, adjust for park.

Key note: we’re only looking at how someone did as a starting pitcher. Many marginal pitchers spend time as a spot starter and occasional reliever, but we’re just looking at starter stats here. (Thus, it’s possible for someone to be considered a marginal starter and a core reliever, or vice versa.)

Teams vary greatly in how much use they have for their marginal starting pitchers. At the low end, the Tigers needed just six outings from marginal starting pitchers. At the other extreme, the Marlins had 55 starts from their marginal arms.

Anyhow, here is how they stack up in terms of actual earned runs allowed versus how many earned runs league-average pitchers would’ve allowed in the same number of innings. (Positive means better):

Team	   SP
TBR	 19.0
KCR	  9.5
ATL	  4.5
CHC	  2.0
HOU	  1.6
DCN	  1.1
CIN	 -1.1
CWS	 -1.4
DET	 -2.8
OAK	 -3.8
NYY	 -7.6
STL	 -8.9
ARI	-14.5
BOX	-14.5
SEA	-14.7
SDP	-17.9
LAD	-20.7
PIT	-20.7
MIL	-21.9
CLE	-25.3
TEX	-26.5
LAA	-28.4
SFG	-28.9
NYM	-29.0
MIA	-30.7
COL	-38.8
TOR	-41.2
BAL	-42.8
PHI	-58.2
MIN	-58.6

Only six teams had better-than-average pitching from their marginal starters. One of them, the Cubs, had a similar superior performance from its bench batters. In both cases, the Cubs’ mainline talents did worse than its backups. Tampa Bay scores the highest here, and they were within a run of average with their backup batters.

The Phillies and Twins easily had the worst pitching performances by marginal arms. In reality, the Phillies were the worst of all. At least the Twins starters took 48 turns to finish 58 runs below average. Philadelphia’s spot starters needed just 36 starts to do it.

Combined, all spot starters allowed 521 more earned runs than league-average pitchers would have in the same number of innings. That’s a bit more than 17 runs per team. In other words, the Phillies and Twins starters were still several games worse than one would expect.

Marginal relievers

Okay, one last group to check, relievers. The formula is the same as it was for starting pitchers, actual earned runs allowed versus what league-average pitchers would surrender in the same number of innings. Of course, make sure to adjust for park. And again, we’re looking just at how pitchers did in this role. If a hurler split his time between the rotation and bullpen, only his bullpen numbers matter.

Here is how they shake out:

Team	Bullpen
KCR	 31.3
MIL	 26.4
ATL	  8.6
MIN	  6.7
TOR	  6.5
TBR	  3.5
OAK	  2.7
ARI	 -1.8
BAL	 -3.0
CIN	 -4.0
SFG	 -5.4
TEX	 -7.7
NYY	 -9.7
DCN	-10.5
PIT	-10.5
CWS	-11.1
DET	-12.4
SDP	-15.4
CHC	-15.6
LAD	-16.4
MIA	-18.7
CLE	-19.4
LAA	-23.9
COL	-24.1
NYM	-29.9
BOX	-32.1
STL	-34.3
PHI	-37.9
SEA	-41.4
HOU	-44.2

Two teams really stand out here, Milwaukee and Kansas City. The Royals, please note, had the best overall bullpen ERA by an AL team since the 1990 A’s. Their bullpen was fantastic on top and on bottom. The Brewers had good pitching from their core relievers (3.49 ERA), but even better work from their marginal arms (2.78 ERA).

Overall, these guys surrendered 343 more runs than league-average pitchers would in their same playing time.

Adding it up and creating a new baseline (from league average to marginal average)

Okay, we’ve looked at all three parts. Now let’s put it all together and see which teams got the most/least from their benches.

Hold on a second, though. Rather than just add them up, it might help to shift our center of gravity. So far, we’ve compared all players to league-average performers. But by definition, these players are marginal. Part of the goal here is to see how much more or less some teams got from their bottom of the roster compared to others. Now that we have the numbers, let’s see what an average bottom of the roster would do and compare that to how each team did.

For example, all bench position players in the AL produced 2,449 Runs Created while making 17,421 outs. That’s a league total of 3.80 Runs Created per 27 outs. For the NL, it’s 3.61 Runs Created by 27 outs. How did our benches do compared to those numbers? (I guess we could combine them into one supersized total, but park factors are neutralized to each league, not all of baseball, so we’ll keep it this way.)

With pitchers, AL backup starting arms had a total ERA of 4.85, while marginal AL relievers had a 4.44 ERA. In the NL, the ERAs were 4.68 for spot starters and 4.31 for extra relievers.

The numbers won’t be the same because 1) we’ve change the baseline from league average to marginal average, and 2) there’s a difference between adjusting from playing time back to playing time.

In other words, if the Twins and Phillies come out about the same with their starting pitchers, despite the Phillies having many fewer starts, when you flip it backwards, they’ll do worse here because their rate stats were so much worse.

Well, change the baseline, add it up, and here is how it turns out. Here are how much each team gained or lost from its bench compared to the rest of the league:

Team	Starters SP   Bullpen	Total
TBR	 22.2	 36.7	 11.7	 70.6
CHC	 37.2	 26.0	 -1.5	 61.7
KCR	 -3.8	 22.1	 39.5	 57.8
ATL	 13.5	 17.3	 17.8	 48.6
BOX	 63.0	 -0.3	-21.7	 41.0
PIT	 36.1	 -2.4	  2.7	 36.4
SDP	 25.6	  9.1	 -1.6	 33.1
ARI	 19.9	  2.8	  9.3	 32.0
OAK	 10.4	  4.2	 10.7	 25.3
CLE	 48.1	-14.1	 -9.9	 24.1
MIL	-15.1	 -2.9	 40.6	 22.6
TEX	 24.9	 -3.8	  1.4	 22.5
CIN	-14.9	  8.6	  4.8	 -1.5
NYM	 16.2	 -5.8	-13.8	 -3.4
LAA	 12.9	-10.0	-12.8	 -9.9
DET	 -7.8	 -0.1	 -3.7	-11.6
TOR	-14.0	-19.5	 19.6	-13.9
SFG	 -8.7	-15.0	  9.7	-14.0
LAD	 -7.9	 -1.0	 -6.3	-15.2
STL	 -8.8	  9.0	-21.5	-21.3
MIA	-16.4	  1.3	 -8.6	-23.7
NYY	-25.7	  1.9	 -1.0	-24.8
MIN	 -7.6	-35.6	 17.5	-25.7
DCN	-42.5	 15.4	 -0.3	-27.4
CWS	-45.3	 17.1	 -1.3	-29.5
HOU	-32.6	 26.4	-27.2	-33.4
SEA	 -3.9	  0.9	-31.5	-34.5
BAL	-25.6	-23.5	  5.4	-43.7
COL	-17.0	-20.2	-10.7	-47.7
PHI	-10.5	-41.1	-21.5	-73.1

Want to know why Tampa Bay was able to snag that last playoff spot? It was because its roster was so deep with talent. Alternately, roster depth killed Baltimore. The Orioles finished seven wins behind the Rays in the AL East and received 110 fewer runs from the end of their roster.

Does this mean that Tampa’s bench gave them seven extra wins? Not exactly; it depends on how you look at things. The average bench is 49 runs below average, so being 70 runs above the typical bench is just two wins above average. In other words, while the average bench would’ve cost their team five games, Tampa’s bench gave them two wins. So it’s either two or seven wins, depending on how you look at it.

The most striking thing here is the gap between the positives and the negatives. Though only 12 teams did better than the typical batch of marginal players, they all were 20 runs better or more. The teams that had depth had plenty of it.

References & Resources
Info comes from

Print This Post
Sort by:   newest | oldest | most voted
Ian R.
Ian R.

“Finally, please note we’re excluding pitcher batting. Position players are supposed to hit as part of their job. Pitchers aren’t.”

In a piece that’s all about adding up marginal contributions, that seems like an odd choice. Sure, pitchers aren’t expected to be good hitters, but a pitcher who IS a (relatively) good hitter makes a positive contribution to his team. It’s also worth pointing out that some especially good-hitting pitchers are used as pinch-hitters (albeit usually for other pitchers), making them marginal position players in a sense.

David P. Stokes
David P. Stokes

The real problem I see with trying to study this issue lies in defining who is a marginal player.  The method used here seems suspect to me.  A team that platoons at 3 or 4 positions is going to have its platoon players counted as marginal, when really they are expected to hit as much as the “regulars” (after all, that’s why you platoon them, to get the platoon advantage).  This method would also classify established stars who missed a lot of time with injuries as “marginal”, even if they performed very well when available.


Great stuff.  Plus, it’s understandable by the likes of non-statheads like me, a Nationals fan still trying to understand what happened.


There seems to be no correlation to anything. You cherrypicked the Rays and Orioles but for example I could look at who was directly behind the Rays, the Cubs and they had a horrible season. The Royals who missed the playoffs were behind them, and the Cardinals and Dodgers who squared off in the NLCS with the Cardinals tying for the best record in baseball were 18th and 19th. Detroit and LA won thier respective divisions and made CS appearances while being negative in all 3 categories.

D Leaberry
D Leaberry

A big difference between the Rays and the Orioles is that the Rays thrive on cheaper situational players while the Orioles trotted out the same line-up quite often.  The Orioles top six field players- Davis, Jones, Hardy, Wieters, Markakis and Machado- played 160, 160, 159, 148, 160 and 156 games, Wieters starting 139 games as catcher. A seventh Oriole, McLouth, started 127 games.  The Tampa bench received more production because the Rays are DESIGNED that way. Obversely, the Orioles have one of the best starting line-ups in baseball so they used their bench less.

Barney Coolio
Barney Coolio
It’s not just about platooning.  Using B-Ref’s “starting 8” or nine, is incomplete.  What if a team has a guy start 82 games at 3b, then trade him and call up a prospect for the last 80 games?  The first guy is listed as the starter, but the youngster really is not a “bench” player.  Also, Bref just lists the guys who played the most at any given position.  Nick Hundley played 58 games for the 2012 Padres, but that was more than any other catcher so Hundley is listed as the catcher.  Sometimes a guy will play all 3… Read more »
Barney Coolio
Barney Coolio

In 2013, Chris Denorfia is listed as a bench player despite 520 plate appearances (more than all 3 Padre starting outfielders.) Daniel Nava had 536 PAs and is listed on the bench despite more PAs than 2/3 OFers and 5/9 “starters.” 

I think a better way to do this would be to exclude any player receiving X amount of plate appearances, say 400.  Anyone getting more than 400 plate appearances really isn’t “marginal,” or at “the end of the bench.”

Barney Coolio
Barney Coolio

I checked out the first 8 teams on this list, and besides SD and Bos, the only teams with a “bench” guy getting 400 plate appearances were Pittsburgh (440), and Tampa (481).  Although Cleveland had three guys with 394, 384, 322.

Chris J.
Chris J.

Ian – no teams hires pitchers based on their hitting ability.  Ultimately, hitting ability is one factor in determining which position players make the roster.  Pitching ability affects which pitchers make the roster.

Hitting ability isn’t just marginal to a pitcher’s place on the roster. It’s irrelevant to his ability to stay on the roster.

David – True, there are ways this method is imperfect.  But you can’t find a perfect system.

Also – no teams platoon 3-4 positions.  Hell, in the days of a 12-13 pitching staff, not that much platooning goes on at all.

Chris J.
Chris J.
budd – I never said there was a correlation.  There doesn’t need to be a correlation.  If Tampa got 110 more runs from their marginal players than Baltimore did, then that’s one reason they got the wild card over Baltimore.  You wanna say there is no conclusive evidence that the end of the bench is always a key factor in determining pennant races?  Congrats – you’re attacking a straw man.  The article never says that.  But not all teams get the same quality from the ends of their benches.  Barney – I already noted this system isn’t perfect—and if you… Read more »