No announcement yet.

Quantifying the Timeline Adjustment

  • Filter
  • Time
  • Show
Clear All
new posts

  • Quantifying the Timeline Adjustment

    Here's an interesting question - who was the greater pitcher, Walter Johnson Or Roger Clemens?

    Purely statistically, it's Johnson - 146 ERA+ (Clemens is 144), 4817.7 IP ( Johnson has 5914.7 ). Johnson was more effective in more innings. He also had the better peak.

    However, there's one major problem to tackle here - Roger Clemens has played his years in the best leagues ever, while Johnson played his entire career in pre-integration baseball. It stands to reason that baseball has improved since the first professional league was formed in 1871 - the question is, how much?

    Bill James, in his New Historical Baseball Abstract, used a linear timeline adjustment - baseball improves a set amount every year. Obviously, I think that's false:

    -During the 1800's, baseball improved more than it did from 1910 to 1940, due to more and more people being turned onto the sport, the advent of professional leagues, the development of the sport, the invention of new equipment and ways of doing things, improving umpires, improved record-keeping (allowing teams to make better decisions about who to sign or drop, etc), more experience (leading to better decisions, etc), and so on.

    -During the League Wars between the NL and AL, and the NL/AL and the Federal League, the quality of play dropped, as marginal players got jobs, soon to be eliminated.

    -During the wars (World War I, World War II, and the Korean War), the quality of play dropped, as players were sent off to war (Cobb and Mathewson; Feller, DiMaggio, and Williams; Mays and Newcombe).

    -During the late 1940's and early 1950's, the quality of play improved drastically, as MLB was integrated (1947 wasn't much better than 1946, but by 1949-1951, the league had improved a ton).

    I'm sure there are other periods of decline or accelerated improvement, but these are the biggest ones I've identified as of yet.

    So, now, how can we account for this? Can we say that players from 50 years ago weren't good enough to play in MLB today? I don't think we can - I'm fairly certain Phil Rizzuto would make it to the majors today, to say nothing of Ted Williams or Babe Ruth. Rizzuto might be comparable to Neifi Perez, and Williams and Ruth might be comparable to Manny Ramirez or Gary Sheffield, but they'd be able to maintain a job in baseball today. Remember, the Wilt Chamberlain argument doesn't quite work, as baseball's been around for a lot longer than basketball or football, and so baseball of the 1920's might be comparable to the NBA of the 1960's or 1970's.

    Any thoughts?

  • #2
    Really tough to say.... Any hypotheitcal like this has so many variables...

    Remember, when comparing players from different era's and stating that today players have better training, better food, better accomadations, if yesterdays players were playing today, they would also benifit from these same things......


    • #3
      I maintain my position that league quality adjustment is (a) not linear (b) not the same for different skill strata among players (a great player is a great player in ANY league...good or bad) and (c) best accomplished by advanced curve fitting combined with a minor linear adjustment to account for extreme changes in league settings (such as vast influxes of players with short careers caused by rapid expansion, world war, poor interleague relations, the formation of additional leagues and the stagnation in scouting techniques that occured in the 1900s through 1920s). Tango seems to think the entire problem is solved by the odds ratio method, which demands that league average be the sole descriptor for league quality, which is why I don't believe that's a complete answer. Others still hold to the idea of a constant league quality adjustment that is subjectively defined and applied similarly to all talent levels. I doubt we'll ever agree 100% and I doubt anyone will be able to "answer" the question perfectly...doesn't mean I won't try to come up with something that comes close.


      • #4
        Probably the easiest way to come up with a timeline adjustment would be to compare the top 10 players in OPS+ and ERA+ for that year to the league average OPS and ERA for that year, and see how many standard deviations the league leaders were from league average. Would this work? Or would the player who led the league always be +3.00 SD from the mean.


        • #5
          In the 2001 "By the Numbers" Newsletter, Rob Wood points out the flaws in the methods Bill James uses to rank the greatest players. Regarding the Timeline Adjustment that Bill James employs, Rob writes:

          "Let me tread carefully here. There are basically two approaches to proceed given these beliefs. One way is to work within the standard sabermetric framework and derive an “internal” adjustment mechanism that attempts to account for the improved quality of play. The other way is to step outside the standard sabermetric framework and make an “external” adjustment. I prefer the first approach; James chooses the second approach."

          Basically James employs an ad hoc adjustment that isn't based on any objective data, but is based solely on his own personal beliefs about the quality of play over time.

          I have posted Rob's article from the "By The Numbers" Newsletter on the timeline adjustment in a Microsoft Word file. The file also includes a few posts that Rob made on Baseball Think Factory and it includes the results from his own Timeline Adjustment he devised.
          Attached Files


          • #6
            The failing in this study (which is a cleverly devised study...don't get me wrong...I like it) is that it doesn't show the robust fluctuations that occured when sudden changes to the game took place. It looks greatly smoothed. WWII is the not the same as 1940 or 1946...


            • #7
              I like this study a lot. As James stated in one of his books, if you don't account for the issue of earlier players dominating the competition to a greater extent than modern players, a disproportionate number of the best players will be pre-1950. I think that Wood is on the right track.

              I think that another way to look at the quality of play in certain eras is to compare how each age group performed versus the rest of the competition over time. Normalize the performance of players at each age and compare how well players born in each year fared versus the competition. We should be able to see some general patterns over time.
              "Batting slumps? I never had one. When a guy hits .358, he doesn't have slumps."

              Rogers Hornsby, 1961


              Ad Widget