With alll the recent posts on WAR (including the NBA discussion) I wanted to take a look at how to measure replacement level (for MLB) by focusing on a WOWY. I doubt this is a completely original approach but I wanted to share what I found with the group.

I looked at 40 good to great (mostly HOF’ers) players that had a pattern of lots of games in season N-1, many fewer games in season N, and many games in season N+1.  The drop off had to be at least 30 games and also the player had to stay with the same team all 3 years.  I then looked at the records of the teams in all 3 seasons. Obviously there is a lot of noise + regression in year to year records, more than can be captured by changes in playing time for one player but looking at 40 would hopefully reduce (some of) the noise.  I used the drop off in games, then recovery in games played to hopefully isolate players who lost playing time due to injury or suspension versus a decline in playing playing in ability or selective benching (e.g. being sat against left-handers).

Season “N” is the one in which the player had significantly less playing time.

Here is the summary (average of the 40 players) of what I found (normalized to 162 game season by extrapolating from winning percentage)

Games Season N-1                 Games Season N                         Games Season N+1
147                                     83                                         143

Team Wins Season N-1             Team Wins Season N                   Team Wins Season N+1
89.0                                 85.0                                       88.9

Player WAR Season N-1           Player WAR Season N                   Player WAR season N+1
5.5                                   2.1                                       4.5

Below is the list of players I used (formatting may not carry over from Excel, sorry).  Getting more position players of this quality that meet the criteria I used may be tough. Many great (or even very good) players did not meet my selection criteria (e.g Aaron, Mays, F Robinson).

Player Age N Team Season N-1 Season N Season N+1
Garciaparra 27 BOS 2000 2001 2002
Williams 31 Bos 1949 1950 1951
Rice 34 Bos 1986 1987 1988
Hornsby 34 CHC 1929 1930 1931
Dawson 34 CHC 1988 1989 1990
Thomas 33 CHW 2000 2001 2002
Greenberg 25 DET 1935 1936 1937
Kaline 27 Det 1961 1962 1963
Morgan 24 HOU 1967 1968 1969
Molitor 27 MIL 1983 1984 1985
Carew 24 Minn 1969 1970 1971
Killebrew 29 Minn 1964 1965 1966
Oliva 33 Minn 1971 1972 1973
Gcarter 22 MON 1975 1976 1977
Raines 28 MON 1987 1988 1989
Terry 27 NYG 1925 1926 1927
Ruth 27 NYY 1921 1922 1923
Randolph 28 NYY 1982 1983 1984
Jeter 29 NYY 2002 2003 2004
Mattingly 29 NYY 1989 1990 1991
Mantle 31 NYY 1962 1963 1964
DiMaggio 34 NYY 1948 1949 1950
A Rodriguez 35 NYY 2010 2011 2012
Nettles 35 NYY 1979 1980 1981
Clemente 24 Pitt 1958 1959 1960
Mazeroski 32 Pitt 1968 1969 1970
Griffey 25 SEA 1994 1995 1996
Bonds 34 SFG 1998 1999 2000
Osmith 29 STL 1983 1984 1985
Allen 31 CHW 1972 1973 1974
Jackson 37 CAL 1982 1983 1984
Snider 31 LAD 1957 1958 1959
Gehringer 28 DET 1930 1931 1932
Henderson 28 NYY 1986 1987 1988
Appling 31 CHW 1937 1938 1939
Evers 23 CHC 1904 1905 1906
Wright 28 NYM 2010 2011 2013
Herman 24 CHC 1933 1934 1935
Fisk 36 CHW 1983 1984 1985
Kell 25 DET 1947 1948 1949

Nice job.

What you can also try to do is only focus on Runs Created, so look at guys with a high wRC+, and look only at runs scored for the team.

You could do the same thing for pitchers.  I wonder if there’s an even bigger team effect, since trying to replace ~200 IP of a HoFer could produce a cascade effect in the bullpen.

Thanks for the feedback and suggestions.  Does the overall size of the impact make sense - missing about 60 games from a great player and the team drop offs about 4 games?

I think so—the average player in your sample went from 5.5 wins to 2 wins.  That’s 3.5 games right there.  If it seems like 3-4 wins in a 60-game stretch is a lot, you’re right, it is, but there’s probably lingering effects on their performance from whatever injury forced them to miss two months.

I finally got around to applying this analysis to pitchers, since Jeff Passan made that comment about WAR not counting IP and I’m still curious about “innings-eaters”.  But I didn’t find anything as interesting as for batters, and I’m not sure why.

To be included in my sample, pitchers had to have at least 5 rWAR in the first year and 4 rWAR in the third year.  The innings pitched in their second year had to be at least 18% less than the first year, and the IP in the third year had to be no worse than 10% less than their first year.  So we want All-Star level pitchers who got injured or something the following year, then rebounded the year after that (while pitching for the same team).  There were 33 pitchers in the sample, all starters:
Christy Mathewson (NYG, 1901-1903)
Togie Pittinger (BSN, 1902-1904)
Bill Donovan (DET, 1903-1905)
Ed Killian (DET, 1905-1907)
Christy Mathewson (NYG, 1905-1907)
Eddie Plank (PHA, 1905-1907)
Addie Joss (CLE, 1905-1907)
Cy Young (BOS, 1905-1907)
Harry Howell (SLB, 1905-1907)
Jack Chesbro (NYY, 1905-1907)
Orval Overall (CHC, 1907-1909)
Walter Johnson (WSH, 1908-1910)
George McQuillan (PHI, 1908-1910)
Eddie Plank (PHA, 1909-1911)
Ray Collins (BOS, 1910-1912)
Babe Adams (PIT, 1911-1913)
Jim Scott (CHW, 1913-1915)
Dutch Leonard (BOS, 1914-1916)
Jeff Pfeffer (BRO, 1917-1919)
Eddie Cicotte (CHW, 1917-1919)
Walter Johnson (WSH, 1919-1921)
Bob Shawkey (NYY, 1920-1922)
Wilbur Cooper (PIT, 1920-1922)
Waite Hoyt (NYY, 1921-1923)
Walter Johnson (WSH, 1922-1924)
Jack Quinn (BOS, 1922-1924)
Dolf Luque (CIN, 1923-1925)
Pete Alexander (CHC, 1923-1925)
Eddie Rommel (PHA, 1923-1925)
Dazzy Vance (BRO, 1925-1927)
Willis Hudlin (CLE, 1927-1929)
Rube Walberg (PHA, 1929-1931)
Watty Clark (BRO, 1929-1931)
George Earnshaw (PHA, 1929-1931)
Ted Lyons (CHW, 1930-1932)
Lefty Stewart (SLB, 1930-1932)
Ed Brandt (BSN, 1931-1933)
Hal Schumacher (NYG, 1933-1935)
Tommy Bridges (DET, 1934-1936)
Carl Hubbell (NYG, 1934-1936)
Schoolboy Rowe (DET, 1934-1936)
Whit Wyatt (BRO, 1941-1943)
Tex Hughson (BOS, 1944-1946)
Dizzy Trout (DET, 1944-1946)
Johnny Sain (BSN, 1946-1948)
Johnny Schmitz (CHC, 1946-1948)
Harry Brecheen (STL, 1946-1948)
Warren Spahn (BSN, 1947-1949)
Ken Raffensberger (CIN, 1949-1951)
Murry Dickson (PIT, 1950-1952)
Ellis Kinder (BOS, 1951-1953)
Mike Garcia (CLE, 1952-1954)
Billy Pierce (CHW, 1953-1955)
Johnny Antonelli (NYG, 1954-1956)
Warren Spahn (MLN, 1954-1956)
Whitey Ford (NYY, 1956-1958)
Frank Lary (DET, 1956-1958)
Billy Pierce (CHW, 1956-1958)
Frank Lary (DET, 1958-1960)
Warren Spahn (MLN, 1959-1961)
Ernie Broglio (STL, 1960-1962)
Bob Friend (PIT, 1960-1962)
Jim Bunning (DET, 1960-1962)
Sandy Koufax (LAD, 1961-1963)
Bob Gibson (STL, 1962-1964)
Don Drysdale (LAD, 1962-1964)
Whitey Ford (NYY, 1962-1964)
Jim Maloney (CIN, 1963-1965)
Gaylord Perry (SFG, 1964-1966)
Mel Stottlemyre (NYY, 1965-1967)
Chris Short (PHI, 1965-1967)
Juan Marichal (SFG, 1966-1968)
Bob Gibson (STL, 1966-1968)
Phil Niekro (ATL, 1969-1971)
Bob Gibson (STL, 1970-1972)
Jim Palmer (BAL, 1970-1972)
Bert Blyleven (MIN, 1971-1973)
Joe Coleman (DET, 1971-1973)
Catfish Hunter (OAK, 1972-1974)
Phil Niekro (ATL, 1972-1974)
Jon Matlack (NYM, 1972-1974)
Steve Carlton (PHI, 1972-1974)
Jim Palmer (BAL, 1973-1975)
Steve Rogers (MON, 1973-1975)
Mickey Lolich (DET, 1973-1975)
Jon Matlack (NYM, 1974-1976)
Luis Tiant (BOS, 1974-1976)
Dennis Eckersley (CLE, 1975-1977)
Luis Tiant (BOS, 1976-1978)
Bruce Sutter (CHC, 1977-1979)
Jon Matlack (TEX, 1978-1980)
Vida Blue (SFG, 1978-1980)
Steve Rogers (MON, 1980-1982)
Jim Clancy (TOR, 1980-1982)
Ron Guidry (NYY, 1983-1985)
Dan Quisenberry (KCR, 1983-1985)
Rick Rhoden (PIT, 1984-1986)
Bret Saberhagen (KCR, 1985-1987)
Orel Hershiser (LAD, 1985-1987)
Jimmy Key (TOR, 1985-1987)
Charlie Leibrandt (KCR, 1985-1987)
Bret Saberhagen (KCR, 1987-1989)
Chris Bosio (MIL, 1989-1991)
Bret Saberhagen (KCR, 1989-1991)
Mike Moore (OAK, 1989-1991)
Nolan Ryan (TEX, 1989-1991)
Kevin Appier (KCR, 1990-1992)
Mark Langston (CAL, 1991-1993)
Charles Nagy (CLE, 1992-1994)
Mike Mussina (BAL, 1992-1994)
Roger Clemens (BOS, 1992-1994)
Roger Clemens (BOS, 1994-1996)
Pat Hentgen (TOR, 1994-1996)
Randy Johnson (SEA, 1995-1997)
Mike Mussina (BAL, 1995-1997)
David Cone (NYY, 1997-1999)
Greg Maddux (ATL, 1998-2000)
Tom Glavine (ATL, 1998-2000)
Al Leiter (NYM, 1998-2000)
Freddy Garcia (SEA, 1999-2001)
Kenny Rogers (TEX, 2000-2002)
Roger Clemens (NYY, 2001-2003)
Javier Vazquez (MON, 2001-2003)
Mike Mussina (NYY, 2001-2003)
Mark Mulder (OAK, 2001-2003)
Randy Johnson (ARI, 2002-2004)
Roy Oswalt (HOU, 2002-2004)
Mark Buehrle (CHW, 2002-2004)
Brandon Webb (ARI, 2003-2005)
Roy Halladay (TOR, 2003-2005)
Curt Schilling (BOS, 2004-2006)
Mike Mussina (NYY, 2006-2008)
Roy Halladay (TOR, 2006-2008)
Carlos Zambrano (CHC, 2006-2008)
Josh Beckett (BOS, 2007-2009)
Mark Buehrle (CHW, 2007-2009)
Johan Santana (NYM, 2008-2010)
Josh Beckett (BOS, 2009-2011)
Justin Verlander (DET, 2009-2011)
Tim Lincecum (SFG, 2009-2011)
CC Sabathia (NYY, 2009-2011)
Felix Hernandez (SEA, 2010-2012)
Cliff Lee (PHI, 2011-2013)
Johnny Cueto (CIN, 2012-2014)

But there’s not much of an effect on team performance:

YR | avgIP | WAR | TM W | TM RA
———————————————————-
1 | 259.7 | 6.8 | 89.2 | 673.9
2 | 153.0 | 1.8 | 87.2 | 679.7
3 | 264.3 | 6.3 | 88.8 | 648.9

Despite losing 4+ wins on average, teams are only two wins worse in the middle year, and pitching staffs don’t give up nearly as many runs as you’d expect.  Is this a sample size issue, or is there a better stat I can look at?

I’m not entirely surprised by this.  A team probably has more options to make a short-term replacement for a starting pitcher than for a position player, where they’re generally stuck with one or two players who are “backing up” that position.  Part of that value may simply be that you’d replace 7 innings per start from your stud pitcher with 5 innings from the spot starter and 2 innings from your relievers.

Really?  I find it very counter-intuitive: If David Ortiz gets injured, the Red Sox call up someone from Pawtucket and slot him into the batting order, with little direct impact on the other eight spots in the order.  On the other hand, those two innings from the bullpen wear out your first options, so they’re not available for tomorrow’s game, and so you end up either overtaxing your best arms or giving extra innings to your worst arms. So part of the value of a workhorse is supposedly saving your bullpen for other games, when you might need it more.