8 thoughts on “Run Shares: measuring offensive value in a team context”

1. John Autin March 14, 2014 at 11:33 pm

   Cool beans, Doug. A question: Is the wRC formula team-dependent? The Bill James Runs Created formula, as I understood it, was meant to be mostly independent of team performance.

   I ask because, while most of these Run Share leaders are truly great seasons, a few are just good years for horrible offenses, e.g.:

   — Red Murray, 1908: 77 Runs Created (B-R formula) for a Cardinals team that scored 2.42 R/G. Murray was 4th in NL RC; Wagner led with 126.

   — Amos Strunk, 1916: 85 RC for an A’s team that scored 2.90 R/G. Strunk was 6th in AL RC; Ty Cobb led with 125.

   — Nick Etten, 1941: 98 RC for a Phils team that scored 3.23 R/G. Etten was 8th in NL RC; Dolph Camilli led with 118.

   Anyway, if I’ve massaged the text correctly, here are your top 44 league-leading Run Shares — all 21% and above:

   Rshare — Player, Year
   25.6% — Babe Ruth, 1923
   25.0% — Barry Bonds, 2001
   24.7% — Rogers Hornsby, 1928
   24.6% — Nap Lajoie, 1910
   24.2% — Babe Ruth, 1924
   23.9% — Chuck Klein, 1933
   23.6% — Ted Williams, 1947
   23.6% — Rogers Hornsby, 1924
   23.5% — Stan Musial, 1948
   23.5% — Barry Bonds, 2002
   22.8% — Babe Ruth, 1920
   22.8% — Gary Sheffield, 1996
   22.6% — Babe Ruth, 1921
   22.6% — Red Murray, 1908
   22.6% — Ralph Kiner, 1949
   22.5% — Babe Ruth, 1919
   22.4% — Stan Musial, 1951
   22.3% — Wally Berger, 1931
   22.1% — Joe Jackson, 1912
   22.1% — Jimmie Foxx, 1935
   22.1% — Ted Williams, 1942
   22.0% — George Stone, 1906
   22.0% — Topsy Hartsel, 1901
   22.0% — Rusty Staub, 1969
   21.9% — Joe Jackson, 1911
   21.8% — Ty Cobb, 1917
   21.8% — Frank Thomas, 1994
   21.7% — Babe Ruth, 1926
   21.7% — Mickey Mantle, 1957
   21.7% — Dick Allen, 1972
   21.7% — Mark McGwire, 1998
   21.6% — Lou Gehrig, 1934
   21.6% — Rogers Hornsby, 1922
   21.6% — Barry Bonds, 2004
   21.4% — George Burns, 1918
   21.3% — Amos Strunk, 1916
   21.3% — Willie Mays, 1957
   21.3% — Derrek Lee, 2005
   21.2% — Joe Jackson, 1913
   21.1% — Tris Speaker, 1914
   21.1% — Tony Gwynn, 1994
   21.0% — Ted Williams, 1946
   21.0% — Babe Herman, 1932
   21.0% — Nick Etten, 1941

   Reply ↓
   1. Doug Post authorMarch 15, 2014 at 1:06 am

      I don’t know the wRC derivation is. This is what FanGraphs says.

      http://www.fangraphs.com/library/offense/wrc/

      It sounds like wRC+ is normalized for run scoring environment and park effects and what not, but that wRC is not. Like the difference between OPS+ (normalized) and OPS (not).

      I noticed that the highest values tended to be from the early years (I only counted 8 seasons in your list above later than 1972), so my hunch is there are different calculations of the metric based on available data (Bill James did this a lot in his original work and justified it on the notion that having a reasonable approximation was better than having nothing at all). So, perhaps the approximations used in the absence of data that became available later could tend to “inflate” the results, at least for the top seasons. But, that’s just a guess.

      Anyway, I stuck with rankings for this piece because I was suspicious about the magnitude of the % totals being skewed towards the earlier data.

      As to the comment that some of the seasons are among the best ever, and others are just good seasons by players on bad teams, my response is “correct”. In both cases, you have one player head and shoulders above all of his teammates, so his “share” of the team’s runs will be magnified. What you won’t see here are very good seasons on very good teams because most of those very good teams will have more than one player with a very good season, which will tent to flatten out the run share for any one player (case in point is Joe Morgan whose best years were with a strong supporting cast, including Pete Rose; neither of those two ever ranked higher than 5th in Run Share).

      If I see a drawback to this method, it’s that “the one good player on a very bad team” may tend to be a bit overdone, so that you see cases like Frank Howard with the Senators and Rusty Staub with the Expos and Dale Murphy with the 80s Braves all leading their league in consecutive seasons. But, I’m not complaining too much because Mantle, Mays, Musial, Bonds, Pujols, Ruth, Hornsby, Foxx, Williams, etc. all have healthy representation, so it’s not like it’s “missing” the truly dominant players.

      Reply ↓
      1. John Autin March 15, 2014 at 4:13 pm

         Just thought I’d update my top-44 list of league-leading Run Shares, to include the team’s Runs Created (B-R formula)

         25.6% — Babe Ruth, 1923 NYY — Team 784 RC
         25.0% — Barry Bonds, 2001 SFG — Team 891 RC
         24.7% — Rogers Hornsby, 1928 BSN — Team 649 RC
         24.6% — Nap Lajoie, 1910 CLE — Team 521 RC
         24.2% — Babe Ruth, 1924 NYY — Team 765 RC
         23.9% — Chuck Klein, 1933 PHI — Team 636 RC
         23.6% — Rogers Hornsby, 1924 STL — Team 724 RC
         23.6% — Ted Williams, 1947 BOS — Team 713 RC
         23.5% — Barry Bonds, 2002 SFG — Team 846 RC
         23.5% — Stan Musial, 1948 STL — Team 709 RC
         22.8% — Babe Ruth, 1920 NYY — Team 746 RC
         22.8% — Gary Sheffield, 1996 FLA — Team 713 RC
         22.6% — Babe Ruth, 1921 NYY — Team 891 RC
         22.6% — Ralph Kiner, 1949 PIT — Team 673 RC
         22.6% — Red Murray, 1908 STL — Team 387 RC
         22.5% — Babe Ruth, 1919 BOS — Team 523 RC
         22.4% — Stan Musial, 1951 STL — Team 687 RC
         22.3% — Wally Berger, 1931 BSN — Team 562 RC
         22.1% — Jimmie Foxx, 1935 PHA — Team 719 RC
         22.1% — Joe Jackson, 1912 CLE — Team 624 RC
         22.1% — Ted Williams, 1942 BOS — Team 756 RC
         22.0% — George Stone, 1906 SLB — Team 496 RC
         22.0% — Rusty Staub, 1969 MON — Team 596 RC
         22.0% — Topsy Hartsel, 1901 CHC — Team 499 RC
         21.9% — Joe Jackson, 1911 CLE — Team 678 RC
         21.8% — Frank Thomas, 1994 CHW — Team 660 RC
         21.8% — Ty Cobb, 1917 DET — Team 568 RC
         21.7% — Babe Ruth, 1926 NYY — Team 828 RC
         21.7% — Dick Allen, 1972 CHW — Team 551 RC
         21.7% — Mark McGwire, 1998 STL — Team 862 RC
         21.7% — Mickey Mantle, 1957 NYY — Team 728 RC
         21.6% — Barry Bonds, 2004 SFG — Team 878 RC
         21.6% — Lou Gehrig, 1934 NYY — Team 810 RC
         21.6% — Rogers Hornsby, 1922 STL — Team 842 RC
         21.4% — George Burns, 1918 PHA — Team 396 RC
         21.3% — Amos Strunk, 1916 PHA — Team 469 RC
         21.3% — Derrek Lee, 2005 CHC — Team 778 RC
         21.3% — Willie Mays, 1957 NYG — Team 645 RC
         21.2% — Joe Jackson, 1913 CLE — Team 596 RC
         21.1% — Tony Gwynn, 1994 SDP — Team 526 RC
         21.1% — Tris Speaker, 1914 BOS — Team 545 RC
         21.0% — Babe Herman, 1932 CIN — Team 631 RC
         21.0% — Nick Etten, 1941 PHI — Team 531 RC
         21.0% — Ted Williams, 1946 BOS — Team 771 RC

         The three Bonds years listed are among the all-time top 10 in B-R’s Runs Created (all 200+ RC), as are three of the Babe’s listed years and Hornsby ’22.

         Reply ↓
         1. Lawrence Azrin March 17, 2014 at 1:56 pm

            @3/JA,

            Comparing individual Runs Created to team Runs Created, makes more sense than comparing individual Runs Created to actual team runs – better to use the same metric for both numbers in a computation.

            I do recall that in making the case for George Brett as the 1985 MVP, Bill James in his 1986 Baseball Abstract showed that the % of actual runs Brett contributed to the Royals total was the highest of any AL player.
            This was:
            (player Runs Produced {Runs + RBI – HR})/ (team Runs Scored)

            It’s good to see that in this case, advanced stats (as used above) agree with a method using mainstream stats.

            Speaking of mainstream stats – I believe that the record for highest % of RBI by one player is still held by Nat Colbert of the 1972 Padres:
            111 RBI / (488 Padres runs) = 22.74% of the team total

            Incredibly enough, 40 years after his last season for the Padres, Colbert _STILL_ holds their career HR record, 163 (two more than AGon, and 28 more than that noted slugger Tony Gwynn…). Chase Headley might surpass that in 3/4 years, if he regains his 2012 HR stroke (and stays with the Padres…).

            Reply ↓
            1. Richard Chester March 17, 2014 at 2:13 pm

               As of the end of the 2011 season Colbert did hold that record of 22.74% of his team’s RBI.

               Others with at least 20% of their team’s RBI.
               Wally Berger, 1935 Braves, 22.61%
               Ernie Banks, 1959 Cubs, 21.25%
               Sammy Sosa, 2001 Cubs, 20.59%
               Jim Gentile, 1961 Orioles, 20.40%
               Bill Buckner, 1981 Cubs, 20.27%
               Bill Nicholson, 1943 Cubs, 20.25%
               Frank Howard, 1968 Senators, 20.23%
               Babe Ruth, 1919 Red Sox, 20.18%
               Frank Howard, 1970 Senators, 20.13%
            2. Doug Post authorMarch 17, 2014 at 2:33 pm

               Fair comment, Lawrence, about mixing derived and actual metrics. I did, though, do that on purpose because I wanted to give a fair shake to the unfortunate stars on woeful teams which, despite their star’s contribution, would tend to be inefficient run scorers that would likely undershoot their derived runs created. That’s just a surmise, but one that would seem to be validated by the frequent representation among the league leaders of lone stars on weak teams.
            3. Lawrence Azrin March 17, 2014 at 4:37 pm

               @6/Doug,

               Good point, you don’t wish to penalize good players too much for inefficient team offenses.

               I think this is one case where using the actual individual Runs Produced and team Runs Scored totals has some value (as I tried to point out in #4 with George Brett/1985).
         2. Richard Chester March 17, 2014 at 5:38 pm

            John: Just a point. If you use the PI Split Finder it shows 783.6 RC for the 1923 Yankees, which agrees with your list. But if you go to the 1923 Yankees team page and search under More Stats it shows 822 RC for them.

            Reply ↓