Revenge of the Tandem: Why Starters and Relievers Need To Go (Part I)

Note: This pair of articles is intended to stoke discussions about tandem rotations, and the one in which I make the case for tandems should follow soon (this one is about starters and relievers as currently constructed). I started working on this pair of articles last spring, so all stats, unless specified otherwise, are 2015. The conclusion hasn't changed, however.

Back in 2013, the Astros employed tandem rotations for their minor league teams, an experiment previously used by the Colorado Rockies and previously employed by Tony La Russa in the early '90's. All three experiments were soon scrapped (the one used by the Astros was changed and then thrown out), and the same goes for other unidentified experiments with paired pitching. The first question is: why did they fail? One reason is the probable need of a thirteenth pitcher on every team (i.e., the mop-up guy). Another is more related to money: no one will be racking up wins or saves (but saves are obviously awful for baseball), leading to less money in the free-agent market. But we here at the Crawfish Boxes live in an ideal world. One where paired pitching is possible, J.D. Martinez is still an Astro, and Jeff Bagwell is in the Hall of Fame (man, I started this a while ago). So let's explore paired/tandem/piggyback pitching, beginning with why anyone in the world ever thought it was a good idea.

Starting pitchers get worse as games go on

Look no further than this chart of 2015 starters as they proceed through the order once, twice, and thrice (w/r/t tOPS+:<100=good for pitcher; >100=bad for pitcher, and it's performance relative to to doing better/worse than usual):

	BA	OBP	SLG	OPS	tOPS+
1st PA in G, as SP	0.251	0.308	0.4	0.709	96
2nd PA in G, as SP	0.26	0.318	0.413	0.731	102
3rd PA in G, as SP	0.268	0.328	0.436	0.764	111

BA rises by .009 for the second PA and .008 for the third, OBP rises by .01 for each, SLG goes up by .013 and .023...there's obviously a tangible impact felt each time through the order.

But don't take my word (or, I guess BBREF's word) for it--check out this super amazing article by Mitchel Lichtman over at BP, where he discusses wOBA-against in a multitude of ways (this is 1999-2002):

Times Through the Order	TBF	wOBA
1	163,900	0.345
2	158,872	0.354
3	124,603	0.362
4	22,221	0.354

Again, similar results. wOBA rises by about .009, then .008, then drops by .008 when faced with a 4th. But even that is not as it appears.

As stated by Lichtman, pitchers who make their way through the order for a 3rd (and even 4th) time are, on the whole, better than their counterparts. Therefore, when he judged for pitcher quality, these results appeared:

TTO	Pitcher quality	Batter quality	Expected wOBA	Obs. wOBA
1	0.349	0.347	0.353	0.345
2	0.349	0.348	0.353	0.354
3	0.348	0.35	0.354	0.362
4+	0.345	0.351	0.353	0.354

Judging by this chart, you can see that wOBA is below expected for the first time through the PA, and it jumps by .001 for the 2nd and 4th+ time. Yet for the third PA, it jumps by .008.

Now, 2000-2012 data:

TTTO	Pitcher quality	Batter quality	Adj. wOBA Obs.
1	0.346	0.34	0.34
2	0.345	0.34	0.35
3	0.343	0.343	0.359
4+	0.339	0.346	0.359

This data does have a larger sample size (but not necessarily more impactful, except for possibly 4+). As evidenced by this, wOBA jumps up to be more than expected by at least .01 for each time through the order.

Lichtman has quite a bit more to talk about, so check it out (along with his follow-up article on pitchers' repertoires). BP does in fact have more articles with similar topics. Ken Funck, for example, put up an article in 2009 about tandem pitchers, in which he referenced the performance of pitchers in trips through the order. The 2006-2008 stats of all starting pitchers:

TTO	PA	AVG/OBP/SLG	OPS	wOBA
1	130,882	.262/.326/.415	0.741	0.328
2	127,740	.273/.335/.436	0.771	0.339
3	96,999	.285/.347/.460	0.807	0.352

An obvious jump exists here as well, unsurprisingly. Both the wOBA and OPS jumps are pronounced. To put the OPS jump in context, it's like having Nick Markakis (2016 OPS: .744) turn into Carlos Correa (2016 OPS: .811) just by going through the order two extra times.

Funck wondered whether or not those stats were being influenced by "bad starters", so he checked out the stats while only including pitchers with 40+ total starts (143 of those pitchers) in the same era:

TTO	PA	AVG/OBP/SLG	OPS	wOBA
1	92,423	.258/.319/.406	0.725	0.322
2	90,893	.266/.327/.421	0.748	0.33
3	73,265	.279/.339/.449	0.788	0.344

This one's a little better, but take a look at the 1st & 3rd TTTO for both wOBA and OPS. They show a similar jump (.066 in the first chart and .061 in the second for OPS, and .024 in the first and .02 in the second for wOBA). This led to Funck's suggestion of implementing a tandem rotation (more on that later!)

Russell Carlton addressed a similar topic, this time wondering about how many pitchers can go late in games. First, he explored how deep starters went into games (in 2012):

Recorded 0-15 outs	Recorded 16-18 outs	Recorded 19-21 outs	Recorded 22-24 outs	Recorded 25 or more outs
30.60%	30.40%	27.00%	9.20%	2.80%

The mean number of outs is 17.4 and the median is 18, suggesting most pitchers make it through between 5-7 innings per outing. This is consistent with what most people assume about starters. Interestingly enough, Carleton discovered that, in 2012, more starts lasted four innings or fewer (12.7%) than lasted past the seventh inning (12.0%), suggesting that good starts are becoming less likely.

A tangent

Looking at pitcher quality and perseverance made me wonder about a point brought up in Scorecasting (and probably other places as well), which is that ace pitchers are usually defined as those who are able to make their way through more innings than the average pitcher without getting shelled. Here is how the pitchers with a SIERA of 3.00 and under performed in 2015 (sOPS+ has the same rules as tOPS+, except it's adjusted for the league):

Name	Team	xFIP	SIERA		BA	OBP	SLG	OPS	tOPS+	sOPS+
Clayton Kershaw		2.09	2.24	1st PA in G	0.179	0.221	0.229	0.45	74	29
	Dodgers			2nd PA in G	0.21	0.246	0.333	0.579	121	58
				3rd PA in G	0.197	0.249	0.287	0.536	106	42
Chris Sale		2.6	2.52	1st PA in G	0.226	0.255	0.342	0.598	83	68
	White Sox			2nd PA in G	0.229	0.297	0.39	0.687	111	88
				3rd PA in G	0.226	0.272	0.357	0.629	94	65
Max Scherzer		2.88	2.63	1st PA in G	0.194	0.229	0.362	0.591	96	65
	Nationals			2nd PA in G	0.186	0.223	0.312	0.534	79	46
				3rd PA in G	0.228	0.264	0.366	0.63	111	64
Carlos Carrasco		2.66	2.74	1st PA in G	0.257	0.285	0.438	0.723	122	102
	Indians			2nd PA in G	0.202	0.267	0.309	0.576	80	59
				3rd PA in G	0.231	0.292	0.367	0.659	105	73
Jake Arrieta		2.61	2.75	1st PA in G	0.16	0.212	0.255	0.466	84	32
	Cubs			2nd PA in G	0.211	0.254	0.287	0.541	114	49
				3rd PA in G	0.19	0.254	0.286	0.54	113	43
Dallas Keuchel		2.75	2.84	1st PA in G	0.236	0.287	0.322	0.61	112	73
	Astros			2nd PA in G	0.199	0.249	0.308	0.557	93	53
				3rd PA in G	0.21	0.242	0.311	0.553	91	45
Corey Kluber		3.05	2.98	1st PA in G	0.218	0.252	0.395	0.647	97	80
	Indians			2nd PA in G	0.221	0.273	0.378	0.651	100	77
				3rd PA in G	0.237	0.293	0.329	0.622	94	65
Jacob deGrom		2.92	2.99	1st PA in G	0.167	0.208	0.287	0.494	72	39
	Mets			2nd PA in G	0.236	0.276	0.335	0.611	114	68
				3rd PA in G	0.247	0.288	0.34	0.629	120	66
Madison Bumgarner		3.02	3	1st PA in G	0.196	0.248	0.306	0.554	82	57
	Giants			2nd PA in G	0.227	0.26	0.352	0.611	100	67
				3rd PA in G	0.266	0.296	0.43	0.726	136	89

And here's how they did (same order as above) relative to the rest of the league (as far as vs. OPS/tOPS+ for that trip through the order and league OPS/tOPS+ as a whole, respectively):

Lg. avg. for TTO	vs. lg. avg. for PA	Lg. avg. for TTO	vs. lg. avg. for PA	Lg. avg.	vs. lg. avg.	Lg. avg.	vs. lg. avg.
-0.709	-0.259	-96	-22	-0.732	-0.282	-102	-28
-0.731	-0.152	-102	19	-0.732	-0.153	-102	19
-0.764	-0.228	-111	-5	-0.732	-0.196	-102	4
-0.709	-0.111	-96	-13	-0.732	-0.134	-102	-19
-0.731	-0.044	-102	9	-0.732	-0.045	-102	9
-0.764	-0.135	-111	-17	-0.732	-0.103	-102	-8
-0.709	-0.118	-96	0	-0.732	-0.141	-102	-6
-0.731	-0.197	-102	-23	-0.732	-0.198	-102	-23
-0.764	-0.134	-111	0	-0.732	-0.102	-102	9
-0.709	0.014	-96	26	-0.732	-0.009	-102	20
-0.731	-0.155	-102	-22	-0.732	-0.156	-102	-22
-0.764	-0.105	-111	-6	-0.732	-0.073	-102	3
-0.709	-0.243	-96	-12	-0.732	-0.266	-102	-18
-0.731	-0.19	-102	12	-0.732	-0.191	-102	12
-0.764	-0.224	-111	2	-0.732	-0.192	-102	11
-0.709	-0.099	-96	16	-0.732	-0.122	-102	10
-0.731	-0.174	-102	-9	-0.732	-0.175	-102	-9
-0.764	-0.211	-111	-20	-0.732	-0.179	-102	-11
-0.709	-0.062	-96	1	-0.732	-0.085	-102	-5
-0.731	-0.08	-102	-2	-0.732	-0.081	-102	-2
-0.764	-0.142	-111	-17	-0.732	-0.11	-102	-8
-0.709	-0.215	-96	-24	-0.732	-0.238	-102	-30
-0.731	-0.12	-102	12	-0.732	-0.121	-102	12
-0.764	-0.135	-111	9	-0.732	-0.103	-102	18
-0.709	-0.155	-96	-14	-0.732	-0.178	-102	-20
-0.731	-0.12	-102	-2	-0.732	-0.121	-102	-2
-0.764	-0.038	-111	25	-0.732	-0.006	-102	34

As a whole, that group performed rather well, and here are the averages of those groups versus the rest of the league:

Avg. for TTO (OPS) (1, 2, 3)	Avg. for TTO (tOPS+) (1, 2, 3)	Avg. vs. lg. (OPS)	Avg. vs. lg. (tOPS+)
-0.1386666667	-4.666666667	-0.1616666667	-10.66666667
-0.1368888889	-0.6666666667	-0.1378888889	-0.6666666667
-0.1502222222	-3.222222222	-0.1182222222	5.777777778

Now, in the interest of fairness, I then chose the nine worst (as opposed to the above nine best) in 2015, at least as far as SIERA. Behold:

Name		Team	xFIP	SIERA	BA	OBP	SLG	OPS	tOPS+	sOPS+	Lg. avg. for TTO	vs. lg. avg. for PA	Lg. avg. for TTO	vs. lg. avg. for PA	Lg. avg.	vs. lg. avg.	Lg. avg.	vs. lg. avg.
Yovani Gallardo	1st PA	Rangers	4.31	4.59	0.263	0.34	0.385	0.726	99	106	-0.709	0.017	-96	3	-0.732	-0.006	-102	-3
	2nd PA				0.25	0.299	0.368	0.667	83	83	-0.731	-0.064	-102	-19	-0.732	-0.065	-102	-19
	3rd PA				0.298	0.356	0.466	0.822	124	115	-0.764	0.058	-111	13	-0.732	0.09	-102	22
John Danks	1st PA	White Sox	4.65	4.63	0.27	0.319	0.463	0.783	95	119	-0.709	0.074	-96	-1	-0.732	0.051	-102	-7
	2nd PA				0.263	0.316	0.437	0.753	88	105	-0.731	0.022	-102	-14	-0.732	0.021	-102	-14
	3rd PA				0.312	0.374	0.513	0.888	122	132	-0.764	0.124	-111	11	-0.732	0.156	-102	20
Marco Estrada	1st PA	Blue Jays	4.93	4.64	0.203	0.285	0.347	0.632	101	79	-0.709	-0.077	-96	5	-0.732	-0.1	-102	-1
	2nd PA				0.212	0.275	0.398	0.674	112	83	-0.731	-0.057	-102	10	-0.732	-0.058	-102	10
	3rd PA				0.204	0.23	0.365	0.595	86	54	-0.764	-0.169	-111	-25	-0.732	-0.137	-102	-16
Tom Koehler	1st PA	Marlins	4.58	4.67	0.211	0.285	0.341	0.626	70	78	-0.709	-0.083	-96	-26	-0.732	-0.106	-102	-32
	2nd PA				0.285	0.35	0.472	0.822	121	124	-0.731	0.091	-102	19	-0.732	0.09	-102	19
	3rd PA				0.272	0.367	0.405	0.772	110	105	-0.764	0.008	-111	-1	-0.732	0.04	-102	8
Chris Tillman	1st PA	Orioles	4.58	4.69	0.228	0.29	0.374	0.664	74	87	-0.709	-0.045	-96	-22	-0.732	-0.068	-102	-28
	2nd PA				0.291	0.364	0.452	0.816	115	124	-0.731	0.085	-102	13	-0.732	0.084	-102	13
	3rd PA				0.281	0.34	0.468	0.808	111	111	-0.764	0.044	-111	0	-0.732	0.076	-102	9
Mark Buehrle	1st PA	Blue Jays	4.46	4.71	0.293	0.312	0.484	0.796	110	122	-0.709	0.087	-96	14	-0.732	0.064	-102	8
	2nd PA				0.288	0.318	0.415	0.733	97	100	-0.731	0.002	-102	-5	-0.732	0.001	-102	-5
	3rd PA				0.245	0.294	0.418	0.712	90	86	-0.764	-0.052	-111	-21	-0.732	-0.02	-102	-12
R.A. Dickey	1st PA	Blue Jays	4.72	4.76	0.24	0.297	0.422	0.719	102	102	-0.709	0.01	-96	6	-0.732	-0.013	-102	0
	2nd PA				0.223	0.292	0.346	0.637	82	76	-0.731	-0.094	-102	-20	-0.732	-0.095	-102	-20
	3rd PA				0.262	0.312	0.432	0.744	110	94	-0.764	-0.02	-111	-1	-0.732	0.012	-102	8
Aaron Harang	1st PA	Phillies	4.99	4.87	0.246	0.29	0.414	0.704	75	97	-0.709	-0.005	-96	-21	-0.732	-0.028	-102	-27
	2nd PA				0.288	0.346	0.416	0.763	93	110	-0.731	0.032	-102	-9	-0.732	0.031	-102	-9
	3rd PA				0.287	0.355	0.585	0.941	131	142	-0.764	0.177	-111	20	-0.732	0.209	-102	29
Alfredo Simon	1st PA	Tigers	4.78	4.88	0.279	0.338	0.482	0.82	102	130	-0.709	0.111	-96	6	-0.732	0.088	-102	0
	2nd PA				0.267	0.353	0.463	0.815	103	123	-0.731	0.084	-102	1	-0.732	0.083	-102	1
	3rd PA				0.275	0.335	0.466	0.801	98	109	-0.764	0.037	-111	-13	-0.732	0.069	-102	-4

And now, the averages (exact same format as the chart above):

Avg. for TTO (OPS)	Avg. for TTO (tOPS+)	Avg. vs. lg. (OPS)	Avg. vs. lg. (tOPS+)
0.009888888889	-4	-0.01311111111	-10
0.01122222222	-2.666666667	0.01022222222	-2.666666667
0.023	-1.888888889	0.055	7.111111111

Some of the above are, of course, subject to small sample size, but as a whole, the data seem to support the hypothesis that, in fact, better pitchers tend to perform better than lesser-skilled pitchers if left in longer. Now, back to the topic at hand.

Why relief pitchers are inconsistent

Everyone moans about it. Relief pitchers are inconsistent, don't succeed from year to year, and pop up out of nowhere to have shutdown seasons before disappearing into the ether. And that is what makes this line (from 2015) so tempting yet so awful:

	BA	OBP	SLG	OPS	tOPS+
1st PA in G (RP)	0.244	0.315	0.384	0.699	94

Yup, that's an OPS which is .01 better than a starter's first TTTO, and .065 better than their third.

If only relief pitchers could somehow sustain this achievement. An OPS below .700! Only 38 starters matched or beat that mark, and yet the mark above is an average of all relievers in baseball (that includes pitchers like Tanner Scheppers and Jason Motte). Of course, as stated above, that is but a mirage. Relievers are less consistent than the Astros' ability to score runners from third base. Our sister site Athletics Nation summed it up rather succintly:

Relievers, as can be seen by the R^2 factor of 0.02 (a perfect correlation is 1), are wildly inconsistent from season to season. Even the top relievers are inconsistent. Twenty-five relievers earned 2+ fWAR in at least one season between 2011-2015:

2015	Name	WAR		2014	Name	WAR		2013	Name	WAR		2012	Name	WAR		2011	Name	WAR
1	Cody Allen	2.6		1	Dellin Betances	3.1		1	Koji Uehara	3.1		1	Aroldis Chapman	3.3		1	Craig Kimbrel	3.2
2	Aroldis Chapman	2.5		2	Wade Davis	3		2	Greg Holland	3		2	Craig Kimbrel	3.3		2	Jonathan Papelbon	3
3	Dellin Betances	2.4		3	Aroldis Chapman	2.7		3	Mark Melancon	2.5		3	Fernando Rodney	2.4		3	Sean Marshall	2.6
4	Zach Britton	2.1		4	Jake McGee	2.6		4	Kenley Jansen	2.4		4	Greg Holland	2.1		4	David Robertson	2.6
5	Carson Smith	2.1		5	Sean Doolittle	2.4		5	Joe Nathan	2.3		5	David Hernandez	2.1		5	Mariano Rivera	2.4
6	Andrew Miller	2		6	Craig Kimbrel	2.2		6	Trevor Rosenthal	2.3		6	Jake McGee	2		6	Sergio Romo	2
7	Trevor Rosenthal	2		7	Andrew Miller	2.2		7	Craig Kimbrel	2.2		7	Kenley Jansen	2
8	Ken Giles	2		8	Kenley Jansen	2.2		8	Nate Jones	2
9	Wade Davis	2		9	Greg Holland	2.2
				10	Mark Melancon	2
				11	Steve Cishek	2

Of those twenty-five pitchers, only three made the top five twice (Craig Kimbrel, Aroldis Chapman, and Kenley Jansen), and none made it thrice. In fact, only three pitchers even reached 2 fWAR thrice.

Teams, not surprisingly, show no more consistency than do individual relievers. Bullpens fluctuate constantly, and BTBS put out an article in September 2014 detailing the fluctuation of team FIP- from month to month:

2014	Mar/April	May	Ch	June	Ch	July	Ch	August	Ch	Average FIP-
ARI	83	102	19	101	-1	90	-11	89	-1	93
ATL	65	85	20	95	10	75	-20	97	22	83.4
BAL	117	104	-13	91	-13	62	-29	71	9	89
BOS	74	89	15	92	3	101	9	100	-1	91.2
CHC	115	74	-41	98	24	76	-22	81	5	88.8
CHW	107	112	5	102	-10	97	-5	124	27	108.4
CIN	129	101	-28	90	-11	77	-13	101	24	99.6
CLE	94	98	4	104	6	78	-26	85	7	91.8
COL	94	103	9	111	8	100	-11	102	2	102
DET	109	92	-17	104	12	106	2	117	11	105.6
HOU	134	81	-53	105	24	110	5	101	-9	106.2
KC	72	95	23	86	-9	101	15	94	-7	89.6
LAA	126	110	-16	113	3	65	-48	75	10	97.8
LAD	96	119	23	80	-39	101	21	89	-12	97
MIA	109	99	-10	80	-19	86	6	55	-31	85.8
MIL	78	110	32	98	-12	90	-8	117	27	98.6
MIN	92	89	-3	101	12	74	-27	135	61	98.2
NYM	124	104	-20	96	-8	105	9	123	18	110.4
NYY	98	74	-24	107	33	94	-13	91	-3	92.8
OAK	83	84	1	105	21	95	-10	97	2	92.8
PHI	129	103	-26	83	-20	81	-2	62	-19	91.6
PIT	102	120	18	86	-34	139	53	98	-41	109
SD	82	103	21	79	-24	105	26	82	-23	90.2
SEA	108	84	-24	64	-20	85	21	74	-11	83
SFG	98	98	0	96	-2	78	-18	117	39	97.4
STL	81	90	9	86	-4	117	31	118	1	98.4
TB	111	133	22	97	-36	52	-45	78	26	94.2
TEX	101	81	-20	86	5	97	11	97	0	92.4
TOR	111	96	-15	95	-1	109	14	101	-8	102.4
WAS	81	83	2	74	-9	88	14	84	-4	82

I mean, just take a look at the Astros from April through June. A drop of 53 from April to May, with a sudden jump of 24 in June? And poor Minnesota. After losing 27 FIP points in July, they jumped up a staggering 61 points in August. They had the fourth-best bullpen in July, and, ignominiously, were the flat-out worst in August.

To more easily understand the volatility, see this accompanying graph:

I then decided to check out team SIERA in 2015, and sure enough, more change was at hand:

Team	SIERA	SIERA		SIERA		SIERA		SIERA		SIERA
	Mar-Apr	May	May Change	June	June Change	July	July Change	August	August Change	Sept-Oct	Sept-Oct Change	Average SIERA
Angels	3.63	3.34	-0.29	3.68	0.34	3.64	-0.04	3.91	0.27	3.08	-0.83	3.546666667
Athletics	3.71	3.78	0.07	3.24	-0.54	3.23	-0.01	3.46	0.23	4.41	0.95	3.638333333
Mariners	4.26	3.99	-0.27	3.32	-0.67	3.54	0.22	4.18	0.64	3.52	-0.66	3.801666667
Rays	3.87	3.56	-0.31	3.55	-0.01	4.1	0.55	2.71	-1.39	4.21	1.5	3.666666667
Rangers	4.09	3.94	-0.15	4.05	0.11	3.95	-0.1	3.28	-0.67	3.28	0	3.765
Blue Jays	3.29	3.34	0.05	3.03	-0.31	2.78	-0.25	2.86	0.08	3.53	0.67	3.138333333
Diamondbacks	4.06	3.34	-0.72	4.13	0.79	3.57	-0.56	3.67	0.1	3.58	-0.09	3.725
Braves	4.1	3.94	-0.16	3.76	-0.18	3.5	-0.26	4.3	0.8	4.41	0.11	4.001666667
Cubs	3.06	4.08	1.02	3.64	-0.44	3.54	-0.1	3.23	-0.31	2.66	-0.57	3.368333333
Reds	3.56	3.97	0.41	3.96	-0.01	4.25	0.29	3.92	-0.33	3.56	-0.36	3.87
Rockies	3.25	4.11	0.86	3.7	-0.41	4.26	0.56	4.78	0.52	3.82	-0.96	3.986666667
Orioles	4.06	3.02	-1.04	3.17	0.15	3.33	0.16	3.28	-0.05	3.41	0.13	3.378333333
Marlins	3.58	3.09	-0.49	3.46	0.37	2.88	-0.58	3.95	1.07	4.39	0.44	3.558333333
Astros	2.76	2.42	-0.34	3.7	1.28	3.75	0.05	2.9	-0.85	3.28	0.38	3.135
Dodgers	2.79	3.05	0.26	3.24	0.19	3.57	0.33	2.74	-0.83	2.95	0.21	3.056666667
Brewers	3.72	2.99	-0.73	2.92	-0.07	3.29	0.37	3.02	-0.27	3.29	0.27	3.205
Nationals	3.3	3.51	0.21	3.39	-0.12	2.89	-0.5	3.85	0.96	3.54	-0.31	3.413333333
Mets	2.93	3.2	0.27	4.6	1.4	3.64	-0.96	3.32	-0.32	3.33	0.01	3.503333333
Phillies	4.4	3.7	-0.7	3.47	-0.23	3.06	-0.41	3.63	0.57	3.94	0.31	3.7
Pirates	2.98	3.66	0.68	3.07	-0.59	4.24	1.17	3.09	-1.15	3.33	0.24	3.395
Cardinals	3.34	3.68	0.34	3.29	-0.39	4.04	0.75	3.65	-0.39	3.18	-0.47	3.53
Padres	4.51	2.99	-1.52	3.64	0.65	3.06	-0.58	3.12	0.06	3.28	0.16	3.433333333
Red Sox	3.77	4.17	0.4	3.7	-0.47	3.45	-0.25	4.59	1.14	3.97	-0.62	3.941666667
Giants	4.16	3.51	-0.65	3.61	0.1	3.24	-0.37	3.3	0.06	3.43	0.13	3.541666667
White Sox	3.04	3.77	0.73	3.63	-0.14	3.5	-0.13	2.86	-0.64	3.77	0.91	3.428333333
Indians	3.68	3.41	-0.27	3.05	-0.36	3.1	0.05	3.28	0.18	3.35	0.07	3.311666667
Tigers	4.17	3.8	-0.37	4.12	0.32	3.82	-0.3	4.26	0.44	4.39	0.13	4.093333333
Royals	3.16	4	0.84	3.42	-0.58	3.14	-0.28	3.1	-0.04	4.05	0.95	3.478333333
Twins	4.65	3.87	-0.78	4.38	0.51	3.73	-0.65	3.66	-0.07	3.88	0.22	4.028333333
Yankees	3.08	3.49	0.41	3.22	-0.27	2.67	-0.55	3.4	0.73	3.7	0.3	3.26

The Astros may have, by a hair, had the best bullpen in baseball, but that didn't stop them from, after having an average of 2.59 over the first two months, averaging 3.725 over the next two. And despite the Padres having the second-worst bullpen in April, theirs was suddenly tied for second-best in May.

So, to sum it all up, pitchers don't last that long and relievers are rather inconsistent. Now, how can paired pitching fix both of these problems? That's for the next installment.