Previous entries:

Pens Envy: Baseball's Best Bullpens

Now let's use our new toy, Relief Wins, to determine the best (or at least most valuable) bullpen. Here are the teams with 10+ RWins:

You'll note that the vast majority of these teams are from the last ten years. That's because use of middle relievers over that time has skyrocketed. More innings mean more value.

The 2003 Dodgers (Gagne, Quantrill, Mota, Shuey, and Martin) come out on top. I still prefer the 1990 A's pen. I had compared them to the "Nasty Boys", the 1990 Reds pen, in the Nineties section and came down in favor of the A's. This method supports that opinion as the Reds rank 59th at 8.62 RWins.

Now here are the worst pens:

That's all good and well, but are the bullpens today overrated because of the extra innings? What if we divide the Relief Wins by innings pitched? Let's see (min. 50 IP):

So the 2003 Dodgers still end up on top. The '90s A's move up to second, and a Connie Mack team moves into the top 10. Now the worst:

Again pens have improved over time as teams devote better talent to the role. This is a good argument against the accepted position that pitching has become too diluted in baseball today, but that's the subject of another study.

Finally, here are the best and worst bullpens by decade:

Here is a statistical breakdown for the best pens:

2003 Los Angeles Dodgers:

1990 Oakland Athletics:

1995 Cleveland Indians:

2002 Atlanta Braves:

1979 Texas Rangers:

1979 Baltimore Orioles:

2002 Anaheim Angels:

1995 St. Louis Cardinals:

1998 Colorado Rockies:

1926 Philadelphia Athletics:

2003 Anaheim Angels:

1982 Boston Red Sox:

1997 New York Yankees:

1981 New York Yankees:

2001 Anaheim Angels:

1999 Cincinnati Reds:

Now the worst:

1899 Cleveland Spiders (Surprise):

Pen Notes, Future Studies, And Other Bull

2003 NL Cy Young Revisited:

Let's take one last look at the past season's Cy Young candidates based on the wins above average calculations from this section. Here are the candidates with at least 5 RWins or SWins:

Now you can view this either a validation of Gagne's Cy Young worthiness or an indictment of the system I've laid out. As for me, I might be ready to say, "Uncle," and acknowledge Gagne was the best candidate. Almost.

Recommendations:

I've already made these recommendations within this tome, but I just wanted to collect them at the end:

1) Change the Save rule:

SAVES FOR RELIEF PITCHERS
10.20
Credit a pitcher with a save when he meets all three of the following conditions: (1) He is the finishing pitcher in a game won by his club; and (2) He is not the winning pitcher; and (3) He qualifies under one of the following conditions: (a) He enters the game with a lead of no more than three runs and pitches for at least one inning; or (b) He enters the game, regardless of the count, with the potential tying run either on base, or at bat, or on deck (that is, the potential tying run is either already on base or is one of the first two batsmen he faces); or (c) He pitches effectively for at least three innings. No more than one save may be credited in each game.

One thing is clear to me after doing this study: The save stat is probably the most meaningless one that can be dreamt up. Bobby Thigpen's record-setting 1990 season (57 saves) was nowhere near the top reliever seasons (126th with 4.33 Rwins as a Robb Nen-type reliever). One could use that as an argument against this study, but I think most learned fans will agree that Thigpen's 1990 season, aside from the save record, wasn't all that spectacular. It was a very good season to be sure but not among the best for a reliever. Randy Myers 38-save 1992 season results in negative Relief Wins (-1.35). A 4.23 ERA (17% worse that the league average) will do that. Myers' 38-save, 3.88-ERA 1995 season wasn't much better (0.24 Rwin for an adjusted ERA 5% better than the league average). Jeff Reardon's career save totals are not taken that seriously because of season like his 35-save 1986 campaign: -0.57 RWin because of a 3.94 ERA, 6% worse than the park-adjusted league average).

You get the point. An interesting study may be to determine if saves correlate to any tool like RWin that measures relief pitcher effectiveness. I didn’t run it here because I think I know what the results would be. It's clear to me that saves are the result of pitcher usage more than performance.

The rule can be improved to measure results better. First, either get rid of or segment off the three-inning automatic save. It does measure endurance but does not necessarily capture the same idea as the rest of the save rule. Next, get rid of the one-inning, three-run-lead save.

I propose that the save rule be changed so that the pitcher either enters the game at the start of a half inning with a one-run lead or enters a game mid-innings with the tying run either on base or at bat to earn a save. That's it. It's a rule informed by Bill James' research. And the historical stats should be changed retroactively.

There is a precedent. The rule was tweaked before the 1973, 74, and 75 seasons. When the rule was established in 1969, a reliever was credited if his team maintained a lead while he was pitching. In 1973, the rule was changed so that the pitcher was "protect" the lead, i.e., that it didn't change hands. In 1974, the three-inning rule was put into effect. Also, for the first time a pitcher would receive a save only if the score was "close" when he entered. In '74 that meant that the tying run was either on base or at the plate in mid-inning appearances. (I have to dig through my pre-'74 Encyclopedias to determine when the three-run lead rule—10.20.3a above—came into affect). In 1975 the rule was enlarged to include the tying run on deck for mid-inning appearances.

Anyway, the rule is an anachronism and will become more so as teams average four or five pitchers per game. We now have 30+ years of real historical data with the save rule. It's clearly a failure. Baseball should fix it now to revitalize the stat. The only argument I see against this position is Bobby Thigpen's hurt feelings.

2) Establish an Official Historic Hold Rule: It's time for MLB to add an official stat to help evaluate the myriad non-closer relievers on the planet. It should be added retroactively to the statistical record.

Credit a pitcher with a hold when he enters a game in which his team leads by one run or the score is tied at the beginning of a half-inning or when the tying run is at the plate in a mid-inning appearance and he a) records an out and b) maintains the lead/tied status. A pitcher should not get a hold for walking the bases full with a one-run lead and then being saved by a reliever that replaces him. Also, it should not be credited unless any runners he allows are taken into account. Let's say pitcher A enters with a runner on first, one out, and a one-run lead. He strikes out the first batter and then allows a walk to the second. He is taken out and the new pitcher (B) gives up a home run. Pitcher A should not get a hold because his runner was the go-ahead run. Let's say B strikes out the last batter to save A's bacon, I'm on the fence as to whether A should get a hold or not. B should, but without the effort from B, A would have blown the lead. But then again the lead didn't actually change hands and A did have something to do with that (the strikeout).

These are the sorts of things that need to be figured out, but an official stat that is clearly thought-out is a must.

3) Record historic relief statistics officially: All of the calculations throughout this section have been based on prorating the relief stats for those pitchers that started and relieved games (more than 50% of all pitcher seasons). Baseball has chosen to lump all pitcher stats together. This is fine when a pitcher is a pure starter or pure reliever, but not for swingmen.

Baseball should go back through the historical record and either separate pitching stats by starting and relieving or create a separate set of statistics for each role while still keeping all pitching statistics lumped together. I'm sure this is high on Bud's to-do list.

4) Investigate Middle Relievers More Fully: Middle reliever seasons comprise about 75% of all reliever seasons. I have used Bill James' reliever archetypes for this study and those archetypes are based solely on relief aces or closers. There were no middle relief archetypes. Therefore, I created one based on a combination of the starter and the Client Brown-type reliever archetypes.

I feel that the results for middle relievers (e.g., Paul Quantrill and Mark Eichhorn) overvalued their appearances. This may not be an issue for middle relievers from over 20 years ago, given that they were a sub-par crew for the most part. But as teams assigned better and better pitchers to this role, the stats started to look more inflated. Further studies and possibly an explicit middle reliever archetype is needed to determine if this is indeed the case.

Sources

Data: Sean Lahman's 2003 and 2004 MS Access baseball databases, Baseball-Reference.com. (I don't recommend doing any real database querying with Access. I prefer the dead/moribund FoxPro and Paradox dbs, but these are the conditions that prevail. It’s Microsoft's planet; we're just living on it.)
Game Logs: Retrosheet.org (historical), ESPN.com (2003).

The End?

—Last frame of The Blob (Not to be confused with Terry Forster, "The Big Tub of Goo").

Let the end try the man.

—"Hammerin'"Henry IV by William "Author" Shakespeare

Not every end is a goal. The end of a melody is not its goal: but nonetheless, had the melody not reached its end it would not have reached its goal either. A parable.

—Friedrich "Fat Freddie" Nietzsche