Announcement

Collapse
No announcement yet.

FiveThirtyEight: Marlins Better Than Record, But Not Elite

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Statistical Analysis: FiveThirtyEight: Marlins Better Than Record, But Not Elite

    Sometimes, early in the season, a team will jump out to a division lead despite preseason predictions calling for it to finish in last place. In many of those cases, sabermetrics will poke holes in the team’s record — perhaps its run differential lags far behind its win-loss tally, maybe it’s had an easy schedule, or maybe it’s just been lucky.

    The 2014 Miami Marlins fall in the opposite category. Their current 16-15 record is good for last place in the NL East. But the Marlins — pegged in the preseason to win about 69 games if we average together the various predictions made about them — also lead all of Major League Baseball in an important statistic: schedule-adjusted run differential. If anything, their 16-15 mark has been unlucky so far; they’ve won two fewer games than would be expected from their runs scored and runs allowed. And they’ve played 80.6 percent of their games against teams with .500 records or better, the third-highest proportion of any MLB team in the early going.

    Traditionally, run differential has been a better indicator of a team’s skill level than its win-loss record, particularly in small samples of games. But perhaps there’s something about the way in which the Marlins have scored (and prevented) runs that betrays their abilities.

    We can check for this by comparing the Marlins’ actual run differential to what would be predicted by a run estimator. According to Base Runs, Miami “should” have scored about 149.6 runs and allowed 120.6 runs so far in the 2014 season. But compare that differential (plus-29) to their actual run differential (plus-24), and the Marlins haven’t been overly lucky at all — if anything, they’ve been unlucky in that department as well.

    Essentially, we’ve just gone step by step through the same process as Baseball Prospectus when it generates “third-order” wins and losses, and sure enough, the Marlins lead the National League in that category by a wide margin. So, given all of this, does it follow that the Marlins are truly one of the best teams in baseball?

    Well, no, not really.

    For one thing, their hitters also lead the league in batting average on balls in play (BABIP)
    , meaning they’ve gotten exceptionally lucky at having the ball fall in for a hit when they’ve made contact. Interpreting hitter BABIP is more complicated than it is on the pitching side because some batters deviate from the norm by quite a bit even over large sample sizes, but Miami’s BABIP is out of step with its batters’ established baselines. Using the Marlins’ actual 2014 balls in play totals, combined with their preseason ZiPS projections (a statistical projection system created by Dan Szymborski), we would have expected the Marlins’ non-pitchers to put up a BABIP of .305 (compared to their actual rate of .335). Remove those 22 extra hits, and the Marlins lose more than a quarter of their run differential.

    Another component of team luck (Power Point) simply comes from players performing over their heads (call it a hot streak, or even a career year if it persists long enough). Compared to their preseason ZiPS projections, the Marlins as a team have about 4.8 more wins above replacement than would be expected at this point in the season. Perhaps the projections were wrong, and Miami’s players were substantially more talented than their prior histories would indicate, but it’s early in the season to jump to that conclusion. A more reasonable take — and certainly one more grounded in Bayesian philosophy — would acknowledge the Marlins’ early overachievement to some degree, as it is a small bit of evidence that the projections were too pessimistic, but also heavily weigh the prior assumption that Miami is not a good team. The needle moves slightly in the Marlins’ favor, but not by too much.

    And in essence, that’s what Fangraphs does in its projected standings table. It lists the Marlins as projected to win 78 games over the entire 2014 season, up from the 74 wins its Steamer projection system called for before the year, but that number also includes the 16-15 record they’ve already banked (16-15 is about two more wins than we’d have expected through 31 games from their preseason projection). Over the remainder of the season, Steamer expects Miami to play like a 77-win team — just a three-game upgrade over the team’s original expectations, despite the sensational sabermetric indicators mentioned above.

    As analysts, we often identify teams primed for a fall — in spite of a strong record — because their fundamentals are weak. But these Marlins are the rare case of a team with a mediocre record and strong fundamentals that wouldn’t be poised for an improvement in the standings. That makes them a good lesson in why it’s important to take into account a team’s prior talent assessment, especially this early in the season.
    http://fivethirtyeight.com/datalab/h...e-the-marlins/
    Originally posted by Madman81
    Most of the people in the world being dumb is not a requirement for you to be among their ranks.
    Need help? Questions? Concerns? Want to chat? PM me!

  • #2
    That's pretty much exactly how I feel.

    Comment


    • #3
      Yeah I read that earlier. Good article.

      Comment


      • #4
        Great article, MK. Thanks for posting.

        Regression to the mean might come in the form of a west coast road trip.

        Unless Stanton goes ALL OVER BEEFS FACE

        Comment

        Working...
        X