We dropped the ball on this series, but we've now got all the data compiled and we'll get through it in quick order over the next few days.
For refreshers, you can read Part 1 here and Part 2 here.
We'll start here with a little added historical context, as it will become important to the unfolding story for us to know that the workload level for relievers undergoes its own rise and fall over the 56 years being examined. Here in part three (1985-1998) we will see the beginning of the great drop in workload maximums for relievers; as the chart at right reveals, the five-year running percentage of 100+-inning relievers falls off a cliff and heads toward absolute zero at something analogous to "terminal velocity."
The average IP for the highest workload reliever drops from 139 in 1971-84 to just 113 in 1985-98. The percentage of teams with 100+ IP relievers drops from 59% to 25%. (To show how this accelerates, we will preview the fact that this percentage drops to just 3% from 1999-2012.) By 2007, we will start to see the rapid extinction of the 90-IP reliever.
Performance-wise, it looks as though the rapid decline of workload stems the tide of "next year" decline: whereas next-year ERA+ values were 77% of the previous year in 1957-70, and 80% in 1971-84, that figure is 87% during 1985-98.
More than a fourth of these high-inning relievers (29%) had better ERA+ values the following year in this timeframe, as opposed to 18% in the previous periods (1957-70, 1971-84). Nothing earthshaking in those numbers, but they tend to support the idea that reducing the workload will mitigate the next-year decline for these pitchers.
And most of these pitchers began to do so in 1990: the nineties would see the greatest concentration of "better the next year" ever, with nine out of eighteen (we'll leave 1994 out for this...), or 50%. That percentage will shrivel to just 15% in 2000-09.
The last great workhorse reliever is Mark Eichhorn (157 IP in 1986 for the Blue Jays). At the time, it wasn't seen as being quite as aberrant as it now appears: Willie Hernandez had a Cy Young season in 1984 with his 35 saves and 140 IP. The year before, Bob Stanley had thrown 145 IP for the Red Sox.
(You may be noting that we've included the 1994 season in the data set. And wondering how much that inclusion affects the averages. When we take those two partial seasons out, the average IP shifts to 115. There is no shift in terms of IP for the "season after.")
It became clear that those workhorse guys would eventually break down. The nineties experimented with lowering workload, which (as noted) seemed to stem the tide of next-year decline.
The specializing trend where a workhorse middle reliever would move to a lower-inning closer role was probably best exemplified by Mariano Rivera, who went from being John Wetteland's setup man to the greatest closer in baseball history. But interestingly, Mo has never had more WAR than he had in 1996, when he threw 25% more innings as a setup man than he would ever have as the Yankees' shutdown ace.
Fear not...the more interesting (and disturbing) data is coming up, in Part 4. Stay tuned.