Replacement Level in the Flesh


One of the most frequently used numbers here at Getting Blanked is Wins Above Replacement – a counting stat based on offensive production, defensive contributions, and durability (playing time) that is adjusted for position. It is a great way to compare across leagues, positions, and eras.

The components of this metric are familiar to most baseball fans though the concept of “replacement level” often throws new comers to this brave new world for a loop. Who or what is a ‘replacement level’ player?

Is it another way of saying average player? (No.) Would a team of replacement level players win zero games? (No.) Would they win 81 games? (Closer to 45, random chance is a real thing.) What gives? (Lots.) Can one excellent player really lead a ragtag bunch of stiffs to the promised land? (No.)

If we are searching for a great measure for both replacement level and the One True Star phenomenon, the terrible 2004 Arizona Diamondbacks are a nearly ideal test subject.

The 2004 Diamondbacks were truly dreadful. They lost 111 games, featured the league’s worst offense, one of the worst defensive teams in baseball, an awful bullpen and brutal pitching staff (with one exception.) They played in a very deep NL West (the second place Giants won 91 games) and, basically, sucked in a very real way.

Both the team’s position players and bullpen corps collectively finished with negative WAR, meaning their cumulative contributions were below replacement level. The Diamondbacks sent out multiple players that easily could’ve been replaced by free agents unsigned during the previous off season – guys no other team wanted! Danny Bautista and Alex Cintron were the greatest offenders, both of whom saw their name on the lineup card every day despite awful numbers across the board.

The starting rotation, however, managed 12 collective wins above replacement. 12 How could that be? Oh yeah, Randy Johnson.

Randy Johnson turned in possibly his greatest season in 2004. He pitched 245 innings in which he struck out 290 batters versus only 44 walks. He allowed a mere 18 home runs and posted a 2.60 ERA, 2.30 FIP and, ohbytheway, pitched a perfect game against the Atlanta Braves.

In total, his season was worth 9.9 Wins Above Replacement. Which is crazy. Johnson’s final win/loss tally for the year was 16-14, a fact that largely kept him from winning the Cy Young award.

The Diamondbacks of 2004 were one generational talent, two league average players (Brandon Webb & Chad Tracey), a few aging & injury prone former stars, and a whole lot of scrubs. The vast, vast majority of whom could easily be replaced — contract status notwithstanding — with any number of AAA talents with no appreciable impact on the team’s performance.

The concepts of replacement level don’t generally jibe with real life but the 2004 Diamondbacks come very close to representing what it is all about. So many bad players playing so badly with a sure-fire Hall of Famer in the middle. And you thought Felix Hernandez had it bad.

Comments (11)

  1. “Both the team’s position players collectively finished and bullpen corps with negative WAR”

    Did you mean:

    Both the team’s position players and bullpen corps collectively finished with negative WAR?

  2. His WAR was only 7.4 not 9.9 no?

    • 9.9 WAR on Fangraphs, which uses FIP (the defensive is preemptively taken out.) Rally WAR (the version employed by BR) uses runs allowed and tinkers with the defense after the fact. I like fWAR better, especially for pitchers.

  3. Absolutely love this blog, first off. Dunno if you get a salary or how you do it but please don’t stop.

    Second, Johnson was robbed in ’04. He pitched way more than Clemens and gave up the same number of runs, I mean come on. Win/Loss should never matter in the Cy, as far as I am concerned, and I don’t even see why it’s still debated, but what do I know?

    Thirdly, I don’t like WAR, I think it’s largely a meaningless stat that’s been way overvalued. So far nobody’s convinced me otherwise. The reason is that the stat, as I understand it, is based on completely unknown quantities. In other words, we can’t know precisely who would replace a player on a given team or how they might perform. Let’s say A Rod got hit by a bus. Chavez would probably replace him. Obviously, Chavez is not A Rod but is A Rod worth 6 or 7 times Chavez’s value? I rather doubt that. 2 or 3 times as good a player, maybe. One example, but to me it shows the dubiousness of the stat. Unless I am missing something, totally ready to be corrected.

    Thanks again for the great blog!

    • Thanks for the great comments.

      Replacement level is a general concept, culled from a large sample of players who played one season yet were unable to find work the following season. Chavez, in your example, is THE replacement player; not A replacement level player.

      From MLB Trade Rumors, I see that Joe Crede, Pedro Feliz, and Willy Aybar are among the third basemen still out there that a team could sign for free. Creating a weighted average of their 2010 stats would set replacement level for a third basemen – the measure against both A-Rod and Eric Chavez are measured.

  4. Ah gotcha. Makes sense.

  5. OK, that detail had escaped me: that the “replacement” player must be a player who couldn’t get signed! I still find it of dubious value. But thanks for taking the time to clarify.

    A Rod is probably worth 6 or 7 Willie Aybars, based on a glance at the stats for ’10!

    • @MKD – glad to help.

      I think one really underrated aspect of WAR is its ability to incorporate playing time. The more time you are on the field, the less time an inferior player is out there mucking things up.

  6. So sorry to belabor this, but see this definition of WAR and you can see why I had a different idea. It says the replacement player would come from the bench or the minors:
    http://www.fangraphs.com/library/index.php/misc/war/

    Now, again, my criticism is that some teams, like the Yankees for instance, have stronger benches and farms than others, so how can this stat be consistent without taking into account each and every variable (an impossibility)?

  7. Don’t apologize, I appreciate the exchange and I’m glad you’re so interested.

    I think the biggest stumbling block in your case is the application versus calculation. This is from the Fangraphs library you linked above:

    which is the value a team would lose if they had to replace that player with a “replacement” player – a minor leaguer or someone from the waiver wire

    This explains how replacement level is calculated. In your example, it is more about applying the actual number.

    If A-Rod went down and Eric Chavez stepped in, we can say they’re going from a 6 WAR guy in Rodriguez to a 2 WAR guy in Chavez. The Yankees lose 4 wins because they can pay/attract Chavez while the Jays have to use a guy like McCoy who is the quintessential replacement level guy.

    Hopefully this makes some more sense than my rambling above.

Leave a Reply

Your email address will not be published. Required fields are marked *