Sunday, October 01, 2006

Trevor Hoffman

Trevor Hoffman recently notched his 479th career save, breaking Lee Smith's former record of 478 to become the all-time career saves leader. So is Trevor Hoffman a Hall-of-Famer?
I don't know, and I'm going to go to great lengths to try and find out. I don't have the time to do it in one sitting, so I'll set up my research and then hit the books (or the webpages) to piece out the data myself and come up with a satisfactory answer.

But first, some background . . .

The roots of relief pitching are very modest. The earliest relief pitchers were simply backups who would come in whenever the main pitcher needed a day off, or if he got in really bad trouble and couldn't finish the game. This is back when starters completed something like 95% of their starts. Relievers were utility players, no more notable than a backup shortstop or second-string catcher.
As time wore on, the team's overall workload started to get spread out. By the turn of the century, the 3-man rotation was basically standard, and a team would also need a couple other pitchers to fill in on the still-rare occasions when a starter couldn't finish. The main pitcher had been replaced by a 3-man rotation, but the relief work was either handled by a starter who wasn't pitching that day or by a spare pitcher. These spare pitchers weren't stars at all and no more celebrated than a bench player, perhaps even less so.
The first true relief ace was a man named Firpo Marberry. Marberry came up with the Washington Senators as a starting pitcher and showed a good deal of promise. I don't know the exact circumstances, but in 1924, Marberry was used mainly as a reliever, only making 15 starts out of 50 appearances. He threw 195.1 IP with a 3.09 ERA (a great 130 ERA+, meaning 30% better than the league average). The Senators went on to win the pennant and their only World Series.
The next season, Marberry pitched exclusively in relief, which was all but unheard of at the time. No one had taken a good pitcher and had them specialize as a reliever; relievers were supposed to be fringe starters or members of the rotation getting in some innings in between starts. Although Marberry would never pitch a full season exclusively in relief, he would become the game's first true "relief pitcher," making 551 career appearances and only 187 starts. More importantly, he was a darn good pitcher, lasting 14 seasons with a career 3.63 ERA (116 ERA+).
The use of Marberry started making others in baseball sit up and take notice. Pitchers were no longer completing 95% of their games anymore, or anything even close. As a greater percentage of a team's innings were allocated to relief pitchers, it became important to have someone on your club who specialized in relief to come in and pitch well in the late innings. Managers also no doubt realized that innings become exponentially more important as the game goes on, making the 9th inning a crucial one. More managers wanted to have an ace set aside for just such late-inning crisis situations.
Relievers were much, much different than they are now. The modern closer -- the guy who only comes in to finish a ballgame, usually in the 9th -- didn't exist. As you can tell from Marberry's innings logged, he was more than a one-inning man. He could come in in the 2nd inning if the starter got bombed and go the distance. Or he could come on in a pinch to snuff out a rally in the 8th. Relievers of Marberry's era were more of this ilk; they resembled a multi-purpose modern swingman more than a closer. With few exceptions, they were not true "aces," and many were indeed failed starters, but their versatility in the middle innings won them a role on every major league club.
The next reliever to change the way we perceive the role was Hall-of-Famer Hoyt Wilhelm. Wilhelm came up in 1952 with the Giants and made an immediate impact; in his first season, he made 71 appearances, threw 159.1 IP, and posted a 2.43 ERA (152 ERA+). He finished second in the league to Brooklyn's Joe Black and amazingly managed a 4th-place finish in the MVP voting. This more than anything shows the shift in perception associated with relief pitchers; they were no longer the rubber-armed failed starters, but were perceived more and more as aces. You need look no further than 1950, when Phillies relief ace Jim Konstanty became the first relief pitcher to win the MVP Award. This was back in the days before the Cy Young, but that still represented a major shift for the game.
The biggest thing that separates Wilhelm and pitchers of his era (Lindy McDaniel, Stu Miller, Elroy Face) from modern relievers is of course the number of innings pitched. No modern pitcher throwing exclusively in relief would even come close to throwing over 150 innings, as Wilhelm did. And yet, Wilhelm threw more than 100 innings per year in relief more often than not. The same was true of Stu Miller and Lindy McDaniel, although Roy Face was closer to the modern model in specialization and the lower number of innings pitched.
As time went on, relievers started to become more and more specialized. A new term entered the baseball lexicon: "closer." This was the guy whose job was to close out ballgames. The relief ace wouldn't be making any innings-eating appearances in the 2nd inning anymore; it was the 7th inning or later. As starting rotations shifted from four to five men, teams naturally began carrying more pitchers to accomodate the innings not picked up by starters. Bullpens began to grow, lending them to specialization. Whereas Marberry was a multi-purpose Swiss-Army knife, baseball in the expansion era found a pitcher to fit every role.
It was in this context that what I call the "true" closers emerged. They were primarily concerned with the 9th inning, which is what sets them apart from their forebears. This trend started in the late 60's, with the most famous example being Mike Marshall's rubber-armed 1974 campaign with the Dodgers when he threw 208.1 IP in 106 games -- all in relief. It took Marshall a few years after that before he managed another dominant year (thus establishing the limits of a pitcher's durability), but it was the pitcher like Marshall who became the "true closers."
Other than Marshall, here is a partial list of the "true" closers from this era: Goose Gossage, Rollie Fingers, Bruce Sutter, Sparky Lyle, Dan Quisenberry, Tug McGraw, Willie Hernandez, Gene Garber, Rawly Eastwick, and others. These were the "true closers," pitchers whose role was to come in whenever the late-inning crisis hit and pitch multiple innings if necessary. Forgive me if I seem to romanticize this era, but I do so because it was about to end abruptly.
It ended because of a statistic.
Now here I am, arguing that the use of a statistic has been bad for the game. But the "save" statistic came to mean much more than a textbook definition. In the same sense that a "Win" has become a magical, individual achievement rather than an arbitrary rulebook measure, a "Save" has become the same. The best pitcher is the pitcher with the most "saves," (or so people thought). Because he "saved" the most games.
And thus the save statistic grew far out of proportion to its actual importance.
Pitchers noticed than in arbitration and on the free agent market, a lot of saves meant more money. Saves were the only stat a relief pitcher could point to to prove that they had a good season. Consequently, pitchers began to pitch more exclusively in save situations. And for some reason, traditional baseball wisdom followed down this path, thinking it the right way.
A "save" is like a win; it represents an arbitrary ruling by the MLB rules committee, not any concrete accomplishment toward a team's wins. A pitcher who "won" 24 games did no such thing in the literal sense, what he did was pitch so as to accumulate 24 games where he met the criteria for the statistic. But even moreso than starting pitching, relief pitching has become a "slave to the save," and the game has suffered for it.
The generation of relievers after the Gossage/Quisenberry era were short-timers; that is, they didn't pitch nearly as many innings as their forebears. A relief ace nowadays pitches around 70-75 innings, whereas it wasn't at all unusual for a "true" closer to go 100 innings. Relievers were less inclined to enter the game earlier than the 9th inningl; they just wanted the save, no matter how much the team might need them in the 8th.
I will admit that it's possible that pitching fewer innings is a good thing. It's possible that this lighter workload is easier on a pitcher's arm; except that there's no evidence that relief aces are getting injured any more or less frequently than they were 20 years ago. Many relievers bore the brunt of consecutive 100-inning seasons and went on to long careers. So the only way that pitching more innings could be valuable is if they're pitching more valuable innings. Sure, guys like Gossage pitched 100+ innings. But maybe Trevor Hoffman is saving all of his stuff for the 9th, which is the most important inning. He's also saving it for save situations, which are the most important for his team. Therefore, Hoffman pitches fewer innings, racks up more saves, and is more valuable than Gossage.
Except that I everything I just said is a lie.
"Saves" are not necessarily the most valuable situations in which to use a reliever. That's the basic fallacy that everyone assumes anytime they use saves as a legitimate measure of a pitcher's value; the assumption that saves are the best thing a relief pitcher can do.
The best thing a relief pitcher can do is to help his team win games. They can do that by getting the save, sure, but there are a lot of very valuable situations that do not qualify for a save. There are also a lot of less valuable situations that do. If we want saves to be the most crucial situations a team faces, then we're counting the wrong innings. Saves are not set up to value the most crucial innings a team faces; rather, they're set up in an entirely arbitrary way to reward someone who pitches the 9th inning when his team is ahead. By extension, we can say that Trevor Hoffman, and most relievers of his era, are pitching the wrong innings. If, for some reason, pitchers like Hoffman are indeed limited to 70 innings (Hoffman has only topped 75 innings twice in 14 seasons), then they're pitching the wrong innings.
Here's a situation: tied ballgame, 9th inning. Home team is batting with the bases loaded and nobody out. Their cleanup hitter is on deck. A win is almost assured. The visiting team brings on its closer, who strikes out the side.
No save.
Here's another situation: visiting team ahead by three runs in the bottom of the 9th. The visiting team's closer comes on to pitch. After getting two ground ball outs, he allows two walks, a double, and a wild pitch that brings the lead down to 1. The final hitter hits a long fly ball to the deepest part of the ballpark, where the outfielder catches it at the warning track.
Those represent the two extremes, but it also represents the fact that a save does not reward a "good" performance with much consistency. A situation where the game is tied is much, much more important than one where your team is ahead by three. You could conceivably employ a closer with an ERA of 18.00 (allows two runs per inning) and still have him accumulate 20+ saves by only using him with 3-run leads. Indeed, the list of poor pitchers who have accumulated lots of saves (and subsequently lots of money) is a long one.
So what should teams do? Teams shouldn't value a statistic over the reality. The reality is that using your closer in tied ballgames, or in games where you're one run behind, is much more important than in the 9th inning with a 3-run lead. Teams should also be willing to use their closer as early as the sixth inning, to stifle anything that looks like a crucial rally. Teams should not toss out their closer at the first sign of trouble; that would be foolish. But equally foolish is what managers currently do; bring out vastly inferior pitchers to pitch in trouble spots in the late innings, all the while saving your closer for the 9th inningemergency that may never come.
I could cite numerous studies done, both anecdotal and statistical, stating that the best model for closer usage is the Gossage/Quisenberry model. The best attempt at putting the matter to rest is in the Baseball Prospectus book Baseball Between the Numbers. Bill James also makes a reasonable and valid argument in his Historical Baseball Abstract.

So what does all of this have to do with Trevor Hoffman?
It means that we can take 479 saves and throw them out the window. Don't get me wrong, they're important. Saves aren't the best indicator of quality, but anyone who gets this many is doing something right. Just like wins and losses aren't the best way to measure a starting pitcher, but anyone with 300 wins is probably pretty darn good nonetheless.
I don't have the facilities to do a regression analysis; such a thing would give a correlation showing the relationship of a pitcher's saves from year to year. There would be a strong correlation of strikeout rate, walk rate, and home run rate, but a lower correlation in saves and wins, or winning percentage. In place of that, I will list, in order, each league's league leader in saves since 1969, the first year of the divisional era. Obviously, if saves were directly connected (or correlated) to some inner talent -- that is, if pitcher's actually had the ability to get saves, then you would see most the same pitchers on the list year-to-year. A list of home run leaders varies, yes, but is generally stable and indicative of a true inner talent. Let's take a look at saves leaders:
American League: Ron Perranoski ('69-'70), Ken Sanders ('71), Sparky Lyle ('72), John Hiller ('73), Terry Forster ('74), Goose Gossage ('75), Sparky Lyle ('76), Bill Campbell ('77), Goose Gossage ('78), Mike Marshall ('79), Gossage & Quisenberry (tied, '80), Rollie Fingers ('81), Dan Quisenberry ('82-85), Dave Righetti ('86), Tom Henke ('87), Dennis Eckersley ('88), Jeff Russell ('89), Bobby Thigpen ('90), Bryan Harvey ('91), Eckersley ('92), Jeff Montgomery & Duane Ward ('93), Lee Smith ('94), Jose Mesa ('95), John Wetteland ('96), Randy Myers ('97), Tom Gordon ('98), Mariano Rivera ('99), Derek Lowe & Todd Jones ('00), Mariano Rivera ('01), Eddie Guardado ('02), Keith Foulke ('03), Mariano Rivera ('04), Bob Wickman & F. Rodriguez ('05)
National League: Fred Gladding ('69), Wayne Granger ('70), Dave Giusti ('71), Clay Carroll ('72), Mike Marshall ('73-74), Rawly Eastwick & Al Hrabosky ('75), Eastwick ('76), Rollie Fingers ('77-78), Bruce Sutter ('79-82), Lee Smith ('83), Bruce Sutter ('84), Jeff Reardon ('85), Todd Worrell ('86), Steve Bedrosian ('87), John Franco ('88), Mark Davis ('89), John Franco ('90), Lee Smith ('91-92), Randy Myers ('93), John Franco ('94), Randy Myers ('95), Jeff Brantley & Todd Worrell (tied, '96), Jeff Shaw ('97), Trevor Hoffman ('98), Ugueth Urbina ('99), Antonio Alfonseca ('00), Robb Nen ('01), John Smoltz ('02), Eric Gagne ('03), Armando Benitez & Jason Isringhausen (tied, '04), Chad Cordero ('05)

I must admit that there is a volatility inherent in relief pitching that makes a pitcher's unpredictable and uneven in general. This partially shows up here. But I would imagine that in other measures of relief performance, core measures such as strikeout rate and home run rate, you would find a much higher consistency than the above lists. So let's compare the above list of saves to one of the "truest" skills a pitcher can have, strikeout rate. Here are the league leaders in strikeouts per 9 innings since 1969 (I'll eschew the years and give it in more of a shorthand style). Marvel at the consistency:

American League: Sam McDowell, Sam McDowell, Vida Blue, Nolan Ryan, Ryan again, Ryan again, Frank Tanana, Ryan, Ryan, Ryan, Ryan, Len Barker, Barker again, Dave Righetti, Floyd Bannister, Mark Langston, Floyd Bannister, Langston, Langston, Roger Clemens, Ryan, Ryan, Ryan, Randy Johnson, Johnson, Johnson, Johnson, Clemens, Johnson, Clemens, Pedro, Pedro, Hideo Nomo, Pedro, Pedro, Johan Santana, and Santana again.
National League: Tom Griffin, Tom Seaver, Seaver, Seaver, Seaver, Seaver, John Montefusco, Seaver, Jerry Koosman, J.R. Richard, Richard again, Mario Soto, Steve Carlton, Soto, Carlton, Dwight Gooden, Sid Fernandez, Mike Scott, Nolan Ryan, Ryan again, Mark Langston, David Cone, Cone, Cone, Jose Rijo, Andy Benes, Hideo Nomo, John Smoltz, Pedro, Kerry Wood, Randy Johnson, Johnson, Johnson, Johnson, Kerry Wood, Oliver Perez, Mark Prior.

In this case, all but 14 of these guys appear on the list more than once. This is a list that reflects a true, underlying and repeatable talent. There are some flukes (John Montefusco?), but they had less to do with luck; these pitchers had the real thing, but for some of them it didn't last long. If you're wondering, the years immediately before 1969 are dominated by "Sudden" Sam McDowell and some guy named Koufax.

On the saves list, the superstars do manage to pop up several times, but not as much as you'd think. Mariano Rivera, one of the greatest closers of all time, only led the league in saves three times. What if I told you that Babe Ruth only led the league in homers three times, or that Randy Johnson only led the league in strikeouts three times? It would be even more difficult to swallow if the Babe and the Unit had been beaten out by some of the utterly marginal players that appear on these lists. You will get your flukes on the home run list, yes, but not an Antonio Alfonseca-sized fluke.

My point (and I do have one) is that we're going to have to look elsewhere for Trevor Hoffman's Hall-of-Fame case. 479 saves isn't enough. It is a good start, just like a pitcher's win-loss record is a good start, but I simply need more information. Here's what I need to know:

1. Was Trevor Hoffman used to his full advantage as a closer? In other words, did he get a lot of cheap, 3-run lead saves (called "platter saves"), or did he pitch more often in more important roles? Was he a strictly one-inning man, like many of his brethren, or was he more versatile?
2. How does Hoffman compare to his contemporaries? If Hoffman compares very favorably to his contemporaries, then that makes the historical differences less of an issue. If Hoffman's pitched brilliantly compared to his peers, but had a manager that used him in all the wrong innings, we could perhaps forgive that. Not entirely, but somewhat. Because it doesn't just matter how good you were, it matters how much you helped your team win. And a big part of that, unfortunately, is having a manager that uses you to your full potential.
3. Is the difference between Hoffman and the "true" closers so significant? Most of what I've said so far has been in generalities, but it's possible that Hoffman is not used in accordance with his era.
4. Is Hoffman really as amazingly durable as he seems? Because that could be, in and of itself, a point in his favor. Hoffman could get extra credit for prospering for a long time in a position that sees very few consistently prosper, as the above lists show all too clearly. Few relievers dominate for any length of time. If Hoffman, used right or not, was an uncommonly durable pitcher for his position, that's a point in his favor. We would give extra credit to a catcher who managed to stay at the position for 20 years, because catcher is a position that sees a lot of attrition, and a catcher who can still catch and produce at an advanced age is valuable to his team, because it gives them uncommonly good and consistent production from a position that is short on it. The same could be said for Hoffman, putting together a consistently good career in a role that usually chews guys up and spits them out within 3-5 years.

I plan to go pretty in depth into this one and track Hoffman's career as best I can. I'll report back when I have more data.

To the hunt!

No comments: