FanGraphs Baseball

Comments

RSS feed for comments on this post.

  1. The problem with that approach is that projections vary so much from system to system. I mean, I can pull numbers out of a hat and call that a projection. It would be a bad projection, but a projection nonetheless. So it’s hard to validate the credibility of such a methodology, especially when most projection systems are black boxes. Using past performance as a future indicator, meanwhile, at least grounds the numbers in reality.

    Comment by Andy S — June 29, 2010 @ 10:07 am

  2. Cool stuff, Jack — any chance you could plot the BtB rankings on the same chart? I’m just curious.

    Comment by Sky — June 29, 2010 @ 10:36 am

  3. I totally agree with this article. I find power rankings in general absurd, though they don’t necessarily have to be useless.

    I recall seeing the Padres as a top 3 or 4 team around the second or third week of the season. I thought to myself: how can they possibly put them there? Do you really consider them the third or fourth strongest team out there?

    These major companies know better than to base their rankings on the past week’s performance–it’s practically the fantasy industry’s motto (which obviously CBS and ESPN are heavily involved in).

    Power rankings could be interesting if they reflected a team’s chances (from this point forward) to make the playoffs–the ultimate goal of the regular season anyway.

    Comment by Seth — June 29, 2010 @ 10:39 am

  4. Hear, hear. I think each power ranking should have as its intro a word about its goal or methodology. Is it aiming to be predictive about W-L records over the remainder of the season? Indicative of chances of making the playoffs or winning the World Series? Making a statement regarding the likelihood of team x beating team y in a significant sample of games?

    I find it interesting (though not particularly telling) that the Angels range from 6th (FanHouse) to 17th (CBS Sports).

    Comment by Jason B — June 29, 2010 @ 10:58 am

  5. what is the standard of “success?” ideally you’d want your rankings today to be correlated with future wins, right? how do the “simple” rankings discussed in this piece do by that standard?

    Comment by notdissertating — June 29, 2010 @ 11:14 am

  6. If a power ranking is trying to predict future success, you really just want an updated projection.

    There are many things a power ranking could be trying to do. I’m ok with a lot of them. I just get annoyed when they try to do more than one thing (it all just gets jumbled) or they simply align with Wins and Losses (like so many media power rankings, do.)

    Comment by Sky — June 29, 2010 @ 11:27 am

  7. Not all projection systems are black boxes.

    You can use previous years’ predictions vs. actual results to evaluate the systems. Most sites already do this. So yes, you can pull numbers out of a hat “and call it a projection”, but it’s not a particularly robust prediction, and upon future retrospective analyses it would be shown to be worthless.

    Prediction systems DO use past performance as a future indicator. They just do it in a more sophisticated manner.

    Comment by Travis L — June 29, 2010 @ 11:52 am

  8. PS – is there any feature available where I can get an email if someone replies to one of my comments? I find it hard to remember to return to continue the conversation.

    Comment by Travis L — June 29, 2010 @ 11:52 am

  9. Interesting point. But then you would really be mixing in league and especially divisional effects, which (as stated in the article) are something that Power Rankings are generally trying to flatten.

    Comment by Travis L — June 29, 2010 @ 11:53 am

  10. Nobody has mentioned it yet, so I will.

    BaseballProspectus just runs a formula to grab their power rankings. Although their projection systems are more proprietary than most, they are upfront about how they rank teams.

    Oh and they generally keep AL power rankings separate from NL.

    Comment by Travis L — June 29, 2010 @ 11:55 am

  11. BPro’s rankings are nice because they’re objective, I agree. Mixing wins with pythag wins with WARP2 with WARP3 just seems like a pointless mish-mash, to me, however. Maybe I’m missing something.

    Comment by Sky — June 29, 2010 @ 11:59 am

  12. BPro and coolstandings.com have playoff odds updated daily, in case you’re not familiar. BPro even does it two ways: assumes a team’s true talent is along the lines of how well they’ve played so far this year or assumes a team’s true talent is heavily influenced by pre-season projections (I prefer the latter).

    Comment by Sky — June 29, 2010 @ 12:06 pm

  13. In my opinion, by far the best way to produce a power ranking is to simply use Vegas lines. They know as well as anyone how good teams are, and tracking lines over the course of a season can provide a very accurate snapshot of both how good a team is right now, and how good we should expect them to continue to be.
    I’ve actually been doing this all season myself and I’ve been extremely satisfied with the rankings it has produced. Also, it’s much simpler than trying to produce a projection, since you’re only tracking one thing.

    Comment by degen — June 29, 2010 @ 2:50 pm

  14. Do you post those anywhere, degen? Mind shooting me an email: skyking162@gmail.com — sounds like a really cool project.

    Comment by Sky — June 29, 2010 @ 4:09 pm

  15. Yeah, I really like this idea.

    Comment by Jack Moore — June 30, 2010 @ 12:30 am

  16. On the 25th “Who You Face Matters” was posted by Pat Andriola. His premise was that the statistics of pitchers should be adjusted for the skill of opposing competition. Maybe a similar adjustment could be applied to team performance and power rankings. What part of team performance can be attributed to favorable or tough pitching matchups? I’m not sure that games against Halladay or Jimenez necessarily evens out during the season. It would be a nice project to include pitching matchups in a strength of schedule measure to make a better RPI for baseball.

    Comment by lucas — June 30, 2010 @ 7:03 am

  17. I’ve sent some explantion to Sky, but for others interested here are what I’d call the current Vegas ratings for all teams.
    NL
    STL .571
    PHI .561
    ATL .540
    LAD .532
    COL .520
    SFG .519
    CHC .511
    CIN .505
    NYM .500
    FLA .499
    MIL .496
    WAS .486
    SDP .485
    ARI .469
    HOU .433
    PIT .421

    AL
    NYY .615
    TBR .582
    BOS .572
    TEX .551
    LAA .538
    MIN .535
    DET .527
    CHW .512
    OAK .509
    SEA .503
    TOR .486
    KCR .456
    CLE .450
    BAL .422

    Comment by degen — June 30, 2010 @ 2:01 pm

  18. Yeah, Jay Jaffe’s ‘Hit List’ columns each week are a system where “power is actually ranked.” Individual successes, failures, and pertinent injuries are used to make interesting points about the status of the teams in addition to the rankings. I’m somewhat surprised a major periodical hasn’t added the Hit List to their Sunday editions for a consistent “stathead” angle.

    Comment by Walter L — July 1, 2010 @ 6:12 pm

  19. The problem with Vegas lines is that they are influenced by perception. If Phillies-Braves is a true talent even matchup, but more people think the Phillies are better, then the Vegas oddsmakers will make the payout smaller for the Phils than the Braves, and pocket the difference- even if they know they are evenly matched. That will make it appear that Vegas also thinks the Phils are better.

    It could be that wisdom of the crowd cancels this out, but I doubt it does. It seems likely that there’s a fair amount of structural bias involved (bias towards favorite teams, thus favoring larger markets, bias towards franchises with recent or frequent success, etc.).

    Comment by Todd — July 3, 2010 @ 3:20 am

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


Close this window.

0.250 Powered by WordPress