Marginal wins

DMZ · November 21, 2004 at 5:35 pm · Filed Under Mariners 

Ben Murphy, who is a fine fellow and highly recommended, wrote up this year’s Marginal Wins article at Baseball Prospectus, continuing the work of Doug Pappas, who I hung out with at the Mariners game where the power went out, and was a fine fellow himself.

To spoil the article for you: marginal wins is a measure of how well a team spent their money. It’s cool because it doesn’t care where you get your wins, or how you went about constructing the team. All that matters is how many wins you got, and how much money it took you go get those wins, compared to how much money you’d have to spend on a team of minimum-salary guys who’d lose many, many games.

The Mariners were second-worst in the majors, with $5,079,433/marginal win. The A’s spent $1,212,858. The average team spends about $1,855,682 for every marginal win.

There are some drawbacks to marginal wins in evaluating the worth of a front office, particularly in that a GM can inherit a lot of really bad contracts and that’ll show up (and conversely, a GM taking over a team with a good, deep farm system will have quality, cheap players they may not have been responsible for developing). But even then, it’s an excellent metric for how relatively good an organization has done assembling their team and running their operations.

And the 2004 Mariners were awwwwwwwwwwwwwful.


15 Responses to “Marginal wins”

  1. EA on November 21st, 2004 5:46 pm

    Yeah, but those 2001 Mariners were amazing. They were at what, 750K/win? Ahh, those were the days.

  2. Idahobob on November 21st, 2004 6:02 pm

    This is still an issue worth debating? People, we are officially “BORED”.

  3. planB on November 21st, 2004 6:12 pm

    Erm, that’s a joke article at the Leone for Third blog, not an article from, before anyone freaks out.

  4. Ty on November 21st, 2004 6:14 pm

    I had to look, just to make sure 😛

  5. adam on November 21st, 2004 6:15 pm

    That was scary…..

  6. DMZ on November 21st, 2004 6:19 pm

    Dammit, I didn’t stamp that one out before people saw it. Sorry, folks, I can’t be on the comments real-time (at least, not until someone wants to pay me to be on the comments real-time).

  7. David J Corcoran on November 21st, 2004 6:56 pm

    DMZ or whoever wants to answer:

    On a not unrelated note, would it be worth the money to pay Corey Koskie 5 million a year, when we have Justin Leone who can post similar numbers available for the league minimum?

  8. DMZ on November 21st, 2004 7:14 pm


  9. Jeff Sullivan on November 21st, 2004 7:29 pm

    At least we were seven times as efficient as the Diamondbacks.

    Seriously, their figure will go down as one of the worst in history.

  10. Evan on November 21st, 2004 7:42 pm

    The 2003 Tigers scored infinitely poorly, remember.

  11. Jeff Sullivan on November 21st, 2004 7:51 pm

    Oh yeah.

    Go Diamondbacks!

  12. Digger on November 22nd, 2004 8:57 am

    Murphy’s previous piece derived a simple formula that predicts how much a free agent will be paid based on his previous year’s salary and VORP. Of the 5 signings so far this year, it’s very close on 3 and over-predicts the other 2. Looks good for Benson, too.

  13. Jeff Sullivan on November 22nd, 2004 9:40 am

    Through my contact with him over the past month or so, I’ve been told that Murphy (and friends) are working on slightly improving the model, and they’re going to come out with a new article projecting free agent contracts sometime soon.

    With a little calculation, you’ll come up with the following figures for this year’s premier free agents:

    Beltre: $6.28m
    Beltran: $4.99m
    Drew: $5.34m
    Clement: $6.29m

    The model has a few kinks to work out, because it’s underestimating (significantly, in a few cases) this year’s unusually young class of A-list free agents.

  14. sidereal on November 22nd, 2004 2:03 pm

    As long as Marginal Wins breaks whenever you get near the number of replacement-level wins (hello Arizona), or even goes negative below (we payed negative money! Yay!), I won’t like it. There’s probably an arithmetic fix, but I think the whole concept is flawed. You already have Runs as the most reliable metric for team performance. Just calculate dollars spent per run created. I’m not sure how you tackle the whole replacement-level issue, and that was Doug’s point with the system. I’m not sure you need to. There are no replacement-level teams and there won’t be.

  15. DMZ on November 22nd, 2004 2:39 pm

    You’d get more or less the same thing using $/run except that it would show a much narrower band because you’re not setting the bar high enough.

    And if you think replacement level if flawed as a concept, well… good luck to you with that.