Warning: This post is a mess; proceed at your own risk…

I was thinking about the bullpen a while back and got to wondering a few things:

1. How have Padres relievers performed this year as compared to the starters?
2. How does this compare with league average?
3. How does this compare with recent Padres teams?
4. How does this compare with all Padres teams?
5. What were some shared characteristics of the good bullpens? Of the bad ones?
6. How was workload distributed (i.e., was it spread fairly evenly or did two or three guys carry most of the burden)?
7. How often were relievers used?
8. How hard were they worked?

As is often the case, I got distracted. I ended up tackling (well, maybe not tackling, more like tapping on the shoulder) #3 and #4, and ignoring the rest. Basically I created a spreadsheet that contains select starter/reliever splits for every Padres team from 1969 to 2008.

I found some interesting things, most of which had nothing to do with my initial inquiry. For example, in 1991, there was a dramatic shift in the way the Padres utilized their bullpen. In 1998, there was another:

Padres Individual Relief Pitcher Usage: 1969-2008

	IP/G			PA/G
Years	Avg	Min	Max	Avg	Min	Max
Statistics are courtesy of Baseball-Reference and are through games of September 2, 2008.
1969-1990	1.56	1.45	1.68	6.69	6.11	7.13
1991-1997	1.26	1.21	1.32	5.43	5.21	5.56
1998-2008	1.10	1.04	1.14	4.73	4.32	4.86

This table shows us the number of innings per reliever per game for each of these three eras. Note that not only have the averages declined, but also there is no overlap between the minimum of one era and the maximum of the next. In other words, these aren’t just small dropoffs, these are indicative of a fundamental change in how things are done.

Naturally, this leads to further questions:

1. Were there similar leaguewide shifts at each of these points, or is this phenomenon confined to the Padres?
2. Was there anything specifically going on with the Padres (e.g., managerial change) that might have served as a catalyst for change?
3. Why might these shifts in usage be occurring?

I haven’t attempted to answer the first question because it seems too much like actual work. Someone should study this.

As for catalysts within the organization, well… Greg Riddoch took over for Jack McKeon midway through the 1990 season and continued in 1991. It’s possible that he may not have felt comfortable asserting his ideas on proper bullpen usage with a team that he inherited. Then again, Riddoch never managed before or since at the big-league level, so it’s hard to gauge what those ideas might have been.

The second shift came smack in the middle of Bruce Bochy’s tenure as manager. Bochy was many things (most of them positive), but I’ve never heard him accused of being a catalyst for change.

As for the final question, my first guess is that there has been increased specialization over the years. But before we assert that with anything resembling certainty, we need to approach the problem from a slightly different angle.

To this point we’ve been looking at individual reliever usage, e.g., noting that the average Padres reliever from 1969 to 1990 worked 1.56 innings per game, while the average Padres reliever from 1998 to 2008 works 1.10 innings per game. But how about aggregate usage? If this shift is primarily driven by greater specialization, we wouldn’t expect the overall innings pitched totals for relievers to change. In other words, it would be a matter of more pitchers using up the same amount of innings. To the numbers:

Padres Aggregate Starting and Relief Pitcher Usage: 1969-2008

Years	IP(R)/IP(S)	PA(R)/PA(S)
IP(R), innings pitched by relievers; IP(S), innings pitched by starters; PA(R), plate appearances against relievers; PA(S), plate appearances against starters.
Statistics are courtesy of Baseball-Reference and are through games of September 2, 2008.
1969-1990	0.46	0.46
1991-1997	0.50	0.50
1998-2008	0.51	0.51

The way to read this is like so: From 1991 to 1997, Padres relievers worked half as many innings (and faced half as many batters) as did their starters. There are two important points worth noting here:

1. There was an increase in this number at both shifts.
2. The increase was pretty small and of no great consequence.

In other words, our hypothesis about specialization seems to hold, i.e., more pitchers are being used to do the same job.

At this point, it might be fun — if a little time-consuming — to dig deeper and look at individual usage. Maybe we could make a list of everyone who worked, say, 5% or more of a team’s bullpen innings and see how the workload is distributed among them, and whether the distribution has shifted over time. For example, grabbing one team from each era (making sure that the reliever/starter ratios are similar so as not to introduce another bias), we get this:

1987

Lance McCullers: 24.4% of relief innings
Greg Booker: 13.5
Mark Davis: 12.3
Rich Gossage: 10.3
Craig Lefferts: 10.2
Keith Comstock: 7.1
Other: 22.1

1993

Gene Harris: 11.9% of relief innings
Trevor Hoffman: 10.9
Roger Mason: 10.0
Kerry Taylor: 8.0
Mark Davis: 7.7
Tim Scott: 7.6
Tim Mauser: 7.6
Pedro Martinez: 7.4
Jeremy Hernandez: 6.9
Rich Rodriguez: 6.0
Pat Gomez: 5.7
Other: 10.3

A few points:

• Who the heck is Pat Gomez? This is right around the time I started following the Padres, and I have no recollection of such a person.
• The ’87 closer (McCullers) worked twice as hard as the ’93 closer (Harris).
• In ’87, only six relievers broke our arbitrary 5% threshold; in ’93, that number jumps to 11.
• Each of the top five relievers in ’87 worked harder than his counterpart in ’93, but after that, the ’93 guys bore more of the brunt; in other words, there was a more even distribution of work.

2002

Jeremy Fikac: 14.0% of relief innings
Trevor Hoffman: 12.0
Steve Reed: 8.3
Brandon Villafuerte: 6.5
Alan Embree: 5.8
Jason Boyd: 5.7
Jason Middlebrook: 5.4
Other: 42.3

Yuck, what an awful bullpen.

Anyway, we’re kind of cherry picking here. I don’t know if these particular teams are representative of a particular era. The proper way to do this would be to repeat this exercise for all the Padres teams and look for patterns. If this article were more study and less mess, then I might even try it myself. Regardless, you get the idea.

Oh, and speaking of the awful 2002 ‘pen, I remember now what inspired this line of inquiry: the awful 2008 ‘pen. Actually, it’s not quite as bad as we might think. Don’t get me wrong, it’s certainly not good, but it’s not historically bad. Here’s historically bad:

1997: 4.99 ERA, 775 OPS against
2003: 4.72, 775
1974: 4.66, 790
1995: 4.61, 731
1970: 4.42, 746

None of this is adjusted for era or park, but it gives you some idea. Part of why we think this year’s ‘pen is so awful (and let’s not be too forgiving; it does check in with the ninth highest ERA and 14th highest OPS against in franchise history) is that last year’s bullpen was one of the best:

1982: 2.78 ERA, 635 OPS against
1988: 2.78, 651
1992: 3.05, 645
2007: 3.06, 648
1983: 3.12, 657

This also marks the first time the bullpen’s ERA hasn’t improved from one season to the next since 2002-2003:

2003: 4.72 ERA, 775 OPS against
2004: 3.75, 738
2005: 3.49, 687
2006: 3.45, 692
2007: 3.06, 648
2008: 4.24, 729

The ‘pen kept improving — to the point of historically good levels — and then fell apart one year. Bummer.

I don’t really know how to end this. I guess there are two points I would make:

1. Kevin Towers has demonstrated an ability in recent seasons to assemble a strong corps of relievers; for whatever reason (and this may be worth analyzing as well), he had an off year.
2. Be thankful that we don’t have to watch the ’74 bullpen, which was worst in the National League by a lot.

So yeah, at least we’ve got that going for us…

Tagged as: baseball reference, bullpen, dramatic shift, fundamental change, initial inquiry, league average, relief pitcher, splits, starters, statistics