In defense of replacement-level players?

Ryan Rowland-Smith of the Mariners was 3.6 wins below replacement in 2010, the most of anyone in the majors.
Ryan Rowland-Smith of the Mariners was 3.6 wins below replacement in 2010, the most of anyone in the majors.

[Update: There’s an interesting discussion of this piece at the Baseball Primer Newsblog.]

The really fun thing about poking around in baseball numbers is the same as the really fun thing about watching baseball. If you do it enough, then no matter how long you’ve been doing it you still see things all the time that you’ve never seen before. You still get surprised.

Or, as Bill Cosby used to say at the beginning of “Fat Albert”: If you’re not careful, you might learn something.

When I was poking around recently looking for illuminating factoids to hold up my side of a learned confab about the value of league average players, which I wrote about here and here, the most surprising thing I found wasn’t about league-average players, it was about replacement-level players.

I’m not going to tell you that they’re “pretty good” too, but they’re better than I thought. I bet they’re better than you thought.

First, what is a replacement level player? It’s a tough concept to get your mind around if you’re not familiar with it. You can read some good definitions on Tom Tango’s website, at the Hardball Times and on the Mariners blog Lookout Landing. That last piece is part of Lookout Landing’s excellent Sabermetrics 101 from a few years ago.

Replacement level is a mathematical construct that represents a baseline against which all players can be compared. It’s the level of play expected from the least valuable players who are still good enough to play in the majors.

So, to put it another way, if you had a team full of replacement-level players, you would have a contender for the worst team of all time. Baseball history tells us that the very worst teams of major leaguers are going to win just under 50 games. Occasionally teams actually win even fewer. In the last half century the 1962 Mets won 40 and the 2003 Tigers won 43, but those totals probably included some bad luck. Their Pythagorean winning percentages — what they figured to have won and lost based on how many runs they scored and allowed — had the Mets winning 50 games and the Tigers 49.

A good way to think about it is that replacement level is what’s expected from a player who can be acquired at little or no cost: called up from Triple-A, plucked off the waiver wire, picked up in a trade for a player to be named later, that sort of thing. It describes the bottom layer of players who are good enough to play in the big leagues. But there isn’t some clear demarcation line between big leaguers and minor leaguers. At any time there are plenty of guys at Triple-A who are just as good as many big leaguers. The circumstances just haven’t broken right for them to be in the majors at the moment.

But just because replacement level is the lowest level of play teams can expect from a big league player, that doesn’t mean they’ll get it. There are always some guys who perform below replacement level. Their Wins Above Replacement, or WAR, is a negative number.

You know he's not going to hit it: Pedro Feliz's -2.3 WAR was the worst in the big leagues in 2010.
You know he's not going to hit it: Pedro Feliz's negative 2.3 WAR was the worst in the big leagues in 2010.

That can happen for a number of reasons. A guy who’s shown himself to be capable of perfectly good play has an extended slump or a down year. A rookie who was thought to be ready isn’t. A veteran who was thought to have something left in the tank doesn’t. A marginal player, a replacement-level guy, who has a string of bad luck — an unusual number of hard-hit balls right at people, let’s say — will end up below replacement level, even if his actual talent level is still the same.

Consider the Giants starting rotation, which we’ve been talking about around here lately. The Giants’ signed Todd Wellemeyer in the offseason for $1 million. Fangraphs does the math every year and figured out that this year teams paid, on average, about $4 million to free agents for each marginal win, or win above replacement. The minimum salary is $400,000. So the Giants were expecting Wellemeyer to be a little bit, a very little bit, better than replacement level.

He’d been 2.3 wins below replacement in 2009 with the Cardinals, using Baseball-Reference’s version of WAR, but he’d been a 2.3-win player in 2008 and the Giants were hoping for a rebound.

They didn’t get it. Through June 10 Wellemeyer made 11 starts and a relief appearance and had an ERA of 5.52. He struck out 6.3 batters per nine innings, but he also walked 5.4 and gave up 1.8 home runs, both staggering figures. He was a half a win below replacement, which is kind of an accomplishment in only a little over a third of the season. WAR is a counting stat, and by the end of the season only seven National League pitchers had accumulated -1.5 WAR or worse.

Wellemeyer went on the disabled list for two months with a quad injury, made one relief appearance in August and was released.

In 2008 the Giants had a guy whose first 11 starts were almost identical to Wellemeyer’s in 2010. He had an ERA of 5.53. He was barely striking out five batters per nine innings, and walking as many as he struck out. But he didn’t go on the disabled list or get released. He made 21 more starts. At the end of the year his ERA was 5.15, and he’d been 0.6 wins below replacement. That’s because his name was Barry Zito, he’d been a good pitcher for half a decade in Oakland and he was a year and a half into a seven-year, $126 million contract.

The Giants also got negative WAR numbers this year from Bengie Molina, who was washed up but keeping the chair warm for Buster Posey; from Ryan Rohlinger, a journeyman minor leaguer who was forced into 18 plate appearances worth of duty by injuries; from Mark DeRosa, who tried unsuccessfully to play through a wrist injury. There are a lot of ways a guy can get to a sub-replacement performance.

He showed up: Denard Span provided 705 plate appearances' worth of replacement level play for the Twins.
He showed up: Denard Span provided 705 plate appearances' worth of replacement level play for the Twins.

But here’s the thing I found that really surprised me, as I finally roll around to the point of this sub-replacement-level blog post. I’ve always thought of replacement level as basically the bottom of the major leagues, the worst you can play and still be a big leaguer. I think if you’d asked me I’d have acknowledged that it’s actually the worst you can be expected to play and still be in the big leagues, and that there’s a difference between that and the worst level at which some players actually play.

What I didn’t realize was how big a difference. Because there is a lot of sub-replacement play in the big leagues.

In the major leagues in 2010, 24.5 percent of all innings were thrown by pitchers who ended the season with a negative WAR. Almost one out of every four innings. More than two innings of every game, on average, were tossed by sub-replacement-level pitchers. Every team had at least 100 innings thrown by sub-replacement guys, and if you throw out the Twins’ 103, every team had at least 140 innings. Only the Giants, Twins and Rockies averaged fewer than one of every nine innings thrown by sub-replacement pitchers.

The Diamondbacks got 816 and a third innings from pitchers who were below replacement level for the year. That’s more than half.

Meanwhile, sub-replacement hitters accounted for 18.4 percent of all plate appearances not made by pitchers. Almost one in five times a guy walked to the plate in the majors this year, he was a guy who would end the season with a negative WAR. That’s about seven plate appearances per team per game, and remember we’re not talking about pitchers hitting.

In 2009 the numbers were almost exactly the same: 18.5 percent of all plate appearances and 25.2 percent of all innings were by sub-replacement players, guys who ended the year with a negative WAR.

So it’s misleading to think of a replacement-level player as the bottom of the league, because a whole big pile of players isn’t even that good. Almost a quarter of the league does not play as well as the guy who is supposedly “freely available.”

I didn’t know that. I wasn’t careful. I learned something.

* * *

Rowland-Smith photo: Jane Gershovich/Flickr Creative Commons
Feliz photo: SBoyd/Flickr Creative Commons
Span photo: Bethany Khan/Flickr Creative Commons