Pennant Impact of Big WAR Years

When rating careers, most folks will favor a high peak over a steady rate of accrual. They say that a fixed sum of value — like, 50 WAR over 10 years — tends to have more pennant impact if it’s unevenly distributed (say, five years of 7 WAR and five years of 3 WAR), rather than doled out as 5 WAR each year.

That view has intuitive backing. Whereas WAR is gauged against a replacement-level player, the average player is a more relevant value if you’re trying to build a contending team. If you swap Mike Trout’s 2013 value for that of Jay Bruce and Ben Zobrist, you gain WAR (10.2 to 9.2), but you lose Wins Above Average (5.8 to 7.0) — and you have one less lineup spot from which to build back that WAA.

I can’t quite refute that position, but I do have a soft spot for steady, Lou Whitaker types. (You noticed?) So I wonder how far that intuitive logic is borne out empirically, by actual pennants and championships.

 

I wish I could do a competent study of that question, but there are too many complications — the Yankees’ historical dominance, the decline of high-WAR seasons in the divisional era, the extra rounds of playoffs, etc. Any suggestions on how to approach the issue, without first earning a master’s degree in statistics, would be most welcome.

In the meantime, here are some simple arithmetic observations about big-WAR seasons, in terms of reaching and winning the World Series. First, please note:

  1. This post covers position players for the 109 years when a World Series was played (1903, 1905-93, and 1995-2013).
  2. All WAR figures herein are pro rated to a 162-game schedule, multiplying raw WAR by 162 and dividing by that season’s schedule. (Neither individual games played, nor team games lost to weather, etc., were considered.) For the 154-game schedule, WAR values were bumped up 5%, so the “7-WAR” threshold is 6.6 WAR.

 

7-WAR Seasons

Before we dive into the data pool, some recent history: In the wild-card era, 7-WAR players have figured less in the World Series than ever before. Just 14 of the last 38 WS teams had a 7-WAR player, or 37% — far below the prior norm of 53% — and those teams went 6-8. The “have-nots” went 13-11 overall, 6-4 against “haves.” Since 1995, none of the 17 teams with two or more 7-WAR players won the World Series (0-2).

Now, the plunge. In the 109 years when a World Series was played, position players had 651 seasons of 7+ WAR. Such years were more prevalent before the split to divisions and playoffs:

  • From 1903-68, there were 10.1 seven-WAR years per 30 team-years.
  • From 1969-2013, there were 7.3 seven-WAR years per 30 team-years (7.2 in the past 10 seasons).

Of those 651 seven-WAR seasons, 143 were for teams that reached the World Series, or 22%. (The four players traded mid-season were counted for the acquiring team.) They went 67-76 in those Series, counting each 7-WAR player separately even if a team had two or more such players. Counting each team just once, they went 50-60.

From another angle: About half of all World Series teams had at least one 7-WAR player. The “haves” went 50-60, while the “have-nots” went 59-49; have-nots facing haves went 32-22. That’s probably just a fluke of a small sample and the arbitrariness of a 7-WAR threshold — plus, we’re not even looking at pitching — but it still surprised me.

A quick rundown of the World Series fates of teams with any 7-WAR player(s):

  • Any 7-WAR player(s): 19% reached the Series (110/566), 50-60 record.
  • One 7-WAR player: 16% reached the Series (80/486), 34-46 record.
  • Two 7-WAR players: 36% reached the Series (27/75), 15-12 record.
  • Three 7-WAR players: 3/5 reached the Series, 1-2 record.

As with all things Fall Classical, a breakdown by Yankees and others is apt:

  • For the Yankees, 49% of 7-WAR years were for Series teams (31 of 63), with a WS record of 22-9 for those players, 15-7 for those teams.
  • For all other teams, 19% of 7-WAR years were for Series teams (112/588), with a WS record of 45-67 for those players, 35-53 for those teams.
Teams with One 7-WAR Player

Of all teams with exactly one 7-WAR player, 16% reached the Series (80 of 486), going 34-46. The rates have changed since the playoffs began:

  • Through 1968, 20% of teams with one 7-WAR player reached the Series (52/258), going 21-31.
  • Since playoffs began, 12% reached the Series (28/228), going 13-15.
  • In the two-division era (1969-93), 14% reached the Series (16/115), going 7-9.
  • In the wild-card era, just 11% reached the Series (12/113), going 6-6.

But the change is a bit less if you filter the Yankees. For all other teams with one 7-WAR player, the pennant rate went from 17% pre-playoffs to 12% since. And those that did make the Fall Classic went just 14-27 pre-playoffs, 13-15 since.

Multiple 7-WAR Players

Of the teams with exactly two 7-WAR players, 27 of 75 reached the Series, going 15-12. Again, there’s a pinstriped divide: Yankees, 7/10 made the Series, 5-2 record; others, 20/65, 10-10.

Five teams had three 7-WAR players; three reached the Series, but just one went all the way:

  • 1927 Yankees (Babe Ruth 13.0 WAR, Lou Gehrig 12.4, Earle Combs 7.5) — won WS, 4-0.
  • 1929 Yankees (Ruth 8.5, Tony Lazzeri 8.2, Gehrig 8.1) — finished 2nd, 18 games behind the A’s.
  • 1953 Dodgers (Duke Snider 9.8, Roy Campanella 7.5, Jackie Robinson 7.3) — lost WS, 2-4.
  • 1961 Tigers (Norm Cash 9.2, Al Kaline 8.4, Rocky Colavito 7.7) — 101 wins but finished 2nd, 8 games behind the Yanks.
  • 2004 Cardinals (Scott Rolen 9.1, Albert Pujols 8.4, Jim Edmonds 7.1) — lost WS, 0-4.

In all, 30 of 80 teams with multiple 7-WAR players reached the Series, going 16-14. (Yanks 12, 6-2; others 68, 10-12.) And again, there’s a big break in the rates around the wild card:

  • Through 1993, 28 of 63 teams with two or more 7-WAR players reached the Series, going 16-12. (That includes 3-4 in WS among 14 such teams in the two-division era.)
  • But since 1995, just two of 17 such teams made the Series, and both lost (the 2002 Giants and ’04 Cards); 7 of the other 15 lost in the playoffs.

By the way, just once did teams with multiple 7-WAR players meet in the World Series: The 1941 Yankees defeated the Dodgers in five games, four of them close. (Joe DiMaggio 9.6 WAR, Charlie Keller 7.0; Pete Reiser 7.8, Dolf Camilli 7.0.)

The last team with multiple 7-WAR players to win it all was the 1976 Big Red Machine (Joe Morgan 9.6 WAR, Pete Rose 7.0). Since then, 21 teams had multiple 7’s, but just three reached the Series, and they all lost: the 1980 Royals (George Brett 9.4 WAR, Willie Wilson 8.5), the 2002 Giants (Barry Bonds 11.8, Jeff Kent 7.0), and the 2004 Cardinals (see above).

Sixteen Series matched a multi-7-WAR team against one with no such players; the big stars went 6-10:

  • 1906 Cubs lost to the White Sox, 2-4
  • 1913 Athletics beat the Giants, 4-1
  • 1914 Athletics lost to the Braves, 4-0
  • 1926 Yankees lost to the Cardinals, 3-4
  • 1928 Yankees beat the Cardinals, 4-0
  • 1930 Athletics beat the Cardinals, 4-2
  • 1932 Yankees beat the Cubs, 4-0
  • 1942 Yankees lost to the Cardinals, 1-4
  • 1949 Dodgers lost to the Yankees, 1-4
  • 1953 Dodgers (three 7’s) lost to the Yankees, 2-4
  • 1970 Reds lost to the Orioles, 1-4
  • 1971 Pirates beat the Orioles, 4-3
  • 1972 Reds lost to the Athletics, 3-4
  • 1973 Athletics beat the Mets, 4-3
  • 2002 Giants lost to the Angels, 3-4
  • 2004 Cardinals (three 7’s) lost to the Red Sox, 0-4

Meanwhile, teams with multiple 7-WAR players went 9-3 against WS opponents with exactly one such player. Go figure.

A count of World Series results for all permutations of how many 7-WAR players:

  • Three vs. zero: 0-2
  • Three vs. one: 1-0
  • Two vs. zero: 6-8
  • Two vs. one: 8-3
  • Two vs. two: 1-1
  • One vs. zero: 16-22
  • One vs. one: 15-15
  • Zero vs. zero: 27-27
7-WAR Teammates

The most frequent teammates with 7+ WAR:

  • 7 years — Ruth & Gehrig (1926-32; 4 pennants, 3 WS titles)
  • 5 years — Aaron & Mathews (1957, ’59-61, ’63; 1 pennant, 1 WS title)
  • 3 years — Baker & Collins (1912-14; 2 pennants, 1 title)

Here are the top 21 teammate WAR totals within this 7-WAR pool. Nine pairs reached the Series, going 4-5. (Quick breakdown: Ruth and Gehrig, 5 years, 2-1 in WS; other pairs, 16 years, 2-4 in WS.)

  • 25.4 WAR — Ruth & Gehrig, 1927 (won WS 4-0)
  • 20.9 WAR — Ruth & Gehrig, 1930 (3rd place, 16 GB)
  • 20.5 WAR — Ruth & Gehrig, 1928 (won WS 4-0)
  • 20.4 WAR — Nap Lajoie & Terry Turner, 1906 Indians (3rd place, 5 GB)
  • 20.1 WAR — Ruth & Gehrig, 1931 (2nd place, 13.5 GB)
  • 19.3 WAR — Ruth & Gehrig, 1926 (lost WS 3-4)
  • 18.9 WAR — Home Run Baker & Eddie Collins, 1912 A’s (3rd place, 15 GB)
  • 18.9 WAR — Ty Cobb & Bobby Veach, 1917 Tigers (4th place, 21.5 GB)
  • 18.9 WAR — Ken Griffey, Jr. & Alex Rodriguez, 1996 Mariners (2nd place, 4.5 GB, missed playoffs)
  • 18.8 WAR — Barry Bonds & Jeff Kent, 2002 Giants (lost WS 3-4)
  • 17.9 WAR — Joe Morgan & Johnny Bench, 1972 Reds (lost WS 3-4)
  • 17.9 WAR — George Brett & Willie Wilson, 1980 Royals (lost WS 2-4)
  • 17.8 WAR — Baker & Collins, 1913 A’s (won WS 4-1)
  • 17.8 WAR — Lou Boudreau & Joe Gordon, 1948 Indians (won WS 4-2)
  • 17.6 WAR — Norm Cash & Al Kaline, 1961 Tigers (2nd place, 8 GB)
  • 17.5 WAR — Lajoie & Bill Bradley, 1903 Indians (3rd place, 15 GB)
  • 17.5 WAR — George Sisler & Ken Williams, 1922 Browns (2nd place, 1 GB)
  • 17.5 WAR — Hank Aaron & Eddie Mathews, 1959 Braves (2nd place, 2 GB)
  • 17.5 WAR — Reggie Jackson & Sal Bando, 1969 A’s (2nd place, 9 GB)
  • 17.5 WAR — Joe Morgan & Pete Rose, 1973 Reds (lost NLCS 2-3)
  • 17.5 WAR — Scott Rolen & Albert Pujols, 2004 Cardinals (lost WS 0-4)

 

8-WAR Seasons

If we raise the bar above 7 WAR by whole numbers, each of the next two steps drains about half the pool: 326 player-years of 8+ WAR, and 158 with 9+ WAR.

At the 8-WAR level, 76 of 326 were for Series teams (23%), with a 42-34 record for those Series players, 35-30 for those teams.

Of the teams with any 8-WAR player(s), 22% reached the Series (65/299), going 35-30. The usual caste system: Yanks 17/32, 12-5; others 48/267 (18%), 23-25.

  • 11 of 26 teams with two or more 8-WAR players reached the Series, going 7-4. But no such team has won it all since 1937. Through 1937, seven of 16 such teams reached the Series, all winning (1927-28/’32/’37 Yanks, 1913/’29 A’s, 1935 Tigers). There were no such teams from 1938-58. Since 1959, four of 10 such teams reached the Series, each losing (1961 & ’72 Reds, 1980 Royals, 2004 Cards).
  • Just the ’29 Yanks had three 8-WAR players, and they ran a distant 2nd.
  • 20% of teams with exactly one 8-WAR player reached the Series (54/273), going 28-26. (Yanks 13/25, 8-5; others 41/248, 17%, 20-21.)

Eleven World Series matched teams that each had an 8-WAR player:

  • 1909 Pirates beat the Tigers, 4-3 (Honus Wagner 9.6 WAR; Ty Cobb 10.3)
  • 1929 A’s beat the Cubs, 4-1 (Al Simmons and Jimmie Foxx 8.3; Rogers Hornsby 10.9)
  • 1936 Yankees beat the Giants, 4-2 (Gehrig 9.6; Mel Ott 8.2)
  • 1946 Cardinals beat the Red Sox, 4-3 (Stan Musial 9.0; Ted Williams 11.5).
  • 1955 Dodgers beat the Yankees, 4-3 (Duke Snider 9.0; Mickey Mantle 10.0)
  • 1956 Yankees beat the Dodgers, 4-3 (Mantle 11.9; Snider 8.0)
  • 1957 Braves beat the Yankees, 4-3 (Aaron 8.4; Mantle 11.9)
  • 1961 Yankees beat the Reds, 4-1 (Mantle 10.5; Frank Robinson 8.1 and Vada Pinson 8.0)
  • 1980 Phillies beat the Royals, 4-2 (Mike Schmidt 8.8; Brett 9.4 and Wilson 8.5)
  • 1985 Royals beat the Cardinals, 4-3 (Brett 8.2; Willie McGee 8.1)
  • 1989 A’s beat the Giants, 4-0 (Rickey Henderson 8.6; Will Clark 8.6)

 

9-WAR Seasons

At the 9-WAR level, 41 of 158 were for Series teams (26%), with a 22-19 record for those players, 20-19 for those teams.

Of the teams with any 9-WAR player(s), 26% reached the Series (39/151), going 20-19. For the Yankees, 12/22 reached the Series and went 8-4; for all others, 21% made the Series (27/129) and went 12-15.

  • 2 of 7 teams with two 9-WAR players made the Series, both sweeps by the 1927-28 Yanks. (The other five were the 1906 Indians, 1912 A’s, 1930-31 Yanks and 1996 Mariners.)
  • 37 of 144 teams with exactly one 9-WAR player reached the Series, going 18-19. (Yanks 10/18, 6-4; others 27/126, 12-15.)

Only the 1909, ’46 and ’55 Series had two teams with a 9-WAR player (see above).

 

10-WAR Seasons

Finally, at the 10-WAR level, 31% were for Series teams (23 of 74), with an 11-12 record for those players, 10-12 for those teams. Only the 1927 and ’29 Yankees had two 10-WAR players (guess who); the ’27 squad went 110-44 and then swept the Series.

Of the teams with any 10-WAR player(s), 31% reached the Series (22 of 72), going 10-12. For the Yankees, 9/14, 5-4; all others, 13/58 reached the Series, 5-8.

Pre-playoffs, 19 of 59 teams with any 10-WAR players reached the Series, going 9-10. Since 1969, there were only 12 ten-WAR years, those teams going 1-2 in the Series. Since 1962, only Joe Morgan (’75) has won the World Series in a 10-WAR year.

No World Series ever had opposing 10-WAR players.

Most 10-WAR seasons in this pool:

  • 10 — Babe Ruth, 3-2 in WS
  • 8 — Rogers Hornsby, 0-1 in WS
  • 6 — Willie Mays, 1-1 in WS
  • 5 — Ted Williams, 0-1 in WS
  • 5 — Ty Cobb, 0-1 in WS
  • 4 — Mickey Mantle, 2-2 in WS
  • 3 — Lou Gehrig, 1-0 in WS
  • 3 — Barry Bonds, 0-1 in WS

Again, the pinstripe breakdown: Ruth, Mantle & Gehrig combined, 17 seasons, 6-4 in WS. The five others listed above, 27 seasons, 1-5 in WS.

Besides those three Yankees, just five others won the World Series in a 10-WAR year: Eddie Collins (1910), Tris Speaker (1912), Lou Boudreau (1948), Willie Mays (1954) and Joe Morgan (1975).

__________

With all this, I can’t draw any conclusions. This anecdotal overview leaves me still not blown away by the pennant impact of big-WAR seasons — particularly outside of Yankeeland, and within the wild-card era — but maybe I just lack a good frame of reference.

On a parting note, for whatever it’s worth…

Greatest Teams and Their Best Players

… a look at the top 10 seasons by winning percentage, for 1901-60 and then for 1961-2013, listing all 7-WAR players, and the next-best under 7 WAR. (This includes non-WS years. All WAR figures pro rated as before.)

Pre-Expansion
  • 1906 Cubs (116-36, .763) — Frank Chance, 7.7; Harry Steinfeldt, 7.4 // Joe Tinker, 4.3
  • 1902 Pirates (103-36, .741) — Honus Wagner, 8.3 // Tommy Leach, 6.8
  • 1909 Pirates (110-42, .724) — Honus Wagner, 9.6 // Fred Clarke, 5.5
  • 1954 Indians (111-43, .721) — Bobby Avila, 7.3 // Larry Doby, 6.0
  • 1927 Yankees, 110-44 (.714) — Babe Ruth, 13.0; Lou Gehrig, 12.4; Earle Combs, 7.3 // Tony Lazzeri, 6.6
  • 1907 Cubs, 107-45 (.704) — None // Johnny Evers, 5.6
  • 1931 Athletics, 107-45 (.704) — Al Simmons, 7.9 // Max Bishop, 6.1
  • 1939 Yankees, 106-45 (.702) — Joe DiMaggio, 8.5 // Joe Gordon, 6.6
  • 1932 Yankees, 107-47 (.695) — Babe Ruth, 8.7; Lou Gehrig, 8.3 // Tony Lazzeri, 5.5
  • 1904 Giants, 106-47 (.693) — None // Bill Dahlen, 5.9
Expansion Era
  • 2001 Mariners, 116-46 (.716) — Bret Boone, 8.8; Ichiro Suzuki, 7.7 // Mike Cameron, 5.9
  • 1998 Yankees, 114-48 (.704) — Derek Jeter, 7.5 // Paul O’Neill, 5.8
  • 1995 Indians, 100-44 (.694) — Albert Belle, 7.8 // Jim Thome, 6.6
  • 1961 Yankees, 109-53 (.673) — Mickey Mantle, 10.5 // Roger Maris, 6.9
  • 1969 Orioles, 109-53 (.673) — Frank Robinson, 7.5; Paul Blair, 7.1 // Boog Powell, 5.9
  • 1970 Orioles, 108-54 (.667) — None // Paul Blair, 5.8
  • 1975 Reds, 108-54 (.667) — Joe Morgan, 11.0 // Johnny Bench, 6.6
  • 1986 Mets, 108-54 (.667) — None // Keith Hernandez, 5.5
  • 1998 Braves, 106-56 (.654) — Andruw Jones, 7.4; Chipper Jones, 7.0 // Andres Galarraga, 5.0
  • 1994 Expos, 74-40 (.649) — Moises Alou, 7.2; Marquis Grissom, 7.2 // Larry Walker, 6.7

In each set of 10 teams, there were 12 seven-WAR players and two teams with none. The average best player had 8.2 WAR in the first period, 7.9 in the second. The median best player for all 20 teams had 7.8 WAR.

Your thoughts? Anyone?

50 thoughts on “Pennant Impact of Big WAR Years

  1. 1
    Dr. Doom says:

    So, in the early days of baseball, a team had a 1/16 (7%) chance of going to the World Series. Throw in one 7-WAR player, and that quadruples. Yeah… sounds to me like it’s pretty significant.

    • 2
      Dr. Doom says:

      Doubles, excuse me. I was looking at the wrong subsection. Meanwhile the initial number triples for an 8-WAR player, and quadruples for a 10-WAR player. JA, I defifinitely think that it’s pretty ridiculously significant.

      • 4
        John Autin says:

        Good point, Dr. Doom. Perhaps I’ve been looking at too many things at once … something about a forest, and trees, and a bear with a full stomach.

    • 9
      John Autin says:

      Dr. Doom, FWIW — I pulled up the data for player-seasons of 6.0 to 6.9 WAR. Here’s a comparison of WS rates for 6.0-6.9 players against 7.0-8.4 players (using pro rated figures):

      Teams with any such player(s):
      — 6.0 to 6.9 WAR: 19% reached the Series (86/453), 35-51 record.
      — 7.0 to 8.4 WAR: 19% reached the Series (73/390), 31-42 record.

      Teams with exactly one such player:
      — 6.0 to 6.9 WAR: 18% reached the Series (71/390), 27-44 record.
      — 7.0 to 8.4 WAR: 17% reached the Series (58/350), 23-35 record.

      Two or three* such players:
      — 6.0 to 6.9 WAR: 24% reached the Series (15/63), 8-7 record.
      — 7.0 to 8.4 WAR: 38% reached the Series (15/40), 8-7 record.
      * Three teams had three players of 6.0-6.9 WAR, and all missed the WS. No team had more than two players of 7.0-8.4 WAR.

      One more thing: 28% of all WS teams had no player with even 6+ WAR (60/218). Those teams went 35-25 in the Series. In the wild-card era, 34% had no 6-WAR player (13/38), going 9-4. Nine of the last 19 WS champs had no 6-WAR player.

      All of this still doesn’t amount to a study. But it does point up two questions: Exactly where lies the WAR threshold of most significant pennant impact? And how much has that changed with the extra playoff rounds? For the overall data pool, I’d guess the first big pennant bump for an “MVP-caliber” year lies somewhere over 8 WAR.

      • 18
        no statistician but says:

        JA:

        Just a couple of comments:

        1) If I’m reading correctly, you aren’t considering pitching here, so part of the weirdness of some of the outcomes you cite may lie there.

        2) Another un-investigated issue: what was the competition like in a given year? Winning the pennant with no 6-WAR player but a bunch of 4+ players may be a more normal outcome when the other teams in the league aren’t sparkling with stars.

        • 19
          John Autin says:

          nsb — Absolutely right on both counts … and further reasons why I’m not able to construct a study.

          On the pitching front, the game has changed so much across eras — mainly in terms of reduced IP over time — that it’s hard to name an all-purpose WAR value that denotes star quality.

          As to gauging the competition, that would take a LOT of data.

  2. 3
    Mike L says:

    Nice piece. A couple of random thoughts which both relate to roster consistency. The great Yankee teams of the Reserve System and pre amateur draft era were able to assemble, pay for, and keep their great players. That included in investing in an extensive farm system. There’s an analogy to the old Montreal Canadians who used to have dibs on great players because of NHL territorial rights.

    With free agency, it’s increasingly expensive to acquire multiple players with the potential for high WAR without in some way compromising your ability to fill out the rest of your roster with consistent competence. And the 6 year binding system probably accentuates that, as you tend to get a tremendous dollar per WAR return from younger players, and then overpay substantially for older, free-agent eligible players on both an absolute and cost per WAR basis.

    I would make one more point that has very little to do with this post, but you got me thinking. I would bet that MLB would be better off in the long run without the six year binding period–maybe cut it down to three or four years at most. If anything, it creates more scarcity for older, more expensive players (and the possibility of more dead weight contracts.) More competition for the Albert dollars means lower prices (eventually). Here’s a radical thought–in a really free market, would you pay $22M per year for Elsbury, or $4M for Gardner?

    • 10
      fireworks says:

      Marvin Miller knew exactly what he was doing. He wanted the reserve clause challenged but he didn’t want full-fledged free agency. He wanted arbitration before free agency and limited free agency. The owners (with few exceptions, like Charlie Finley) didn’t have to the foresight to understand how Miller was getting exactly what he wanted and that arbitration was a good means to begin to increase compensation for a player whose rights are still controlled by his team as well as helping to raise everyone’s salary, while the limited free agency created scarcity which pit owners against each other for the services of veteran players, also inflating salaries.

      Whether MLB would be better off granting free agency earlier is a matter of interpretation: it would certainly help decrease the mega-contracts few players receive but it would bump up the salaries of the veterans who would have free agency whom under the current system only have arbitration. There would probably be a net loss of salary but I don’t think the loss is as great as most may think merely because the owners are demonstrably poor at operating in their best long-term interest because of the short-term desire to put asses in the seats and win. Excepting true rebuilds, teams always seem to take a risk on a player or three that doesn’t fit into a healthy long-term strategy.

      However I do think that earlier free agency may facilitate personnel changes greater than that we would find palatable. But then again that’s what they said about the idea of free agency at all.

      If you meant the overall financial health of the game, the game is doing as good as it has ever been in that regard.

      If you mean competitive balance, while a larger pool of free agents would drive down free agent salary it would still ultimately allow wealthier teams to be able to stock their teams with above average players more easily than less wealthy teams (and they’d still be able to overpay such players to acquire their services, except that they’d not need to overpay as much as they do now).

      If you mean the cost of attending a game for the average fan, well, if there was never free agency the prices would undoubtedly be lower than they are now but attendance at MLB games is better than in the so-called golden age. The biggest factor in whether the fans come out is not the price, but whether or not the team wins (except Florida, where nothing makes the fans turn out). I can’t speak as well for other places, and I know other places don’t fit as well, but as a New Yorker I always find it funny when New Yorkers complain about the prices for say Yankees tickets (or any of the other million entertainment options they have) because the discrepancy between the face value of the ticket and what fans are willing to pay for it is the market exploited by scalpers. I’d rather loudmouth Hank get my money than a scalper.

      If you meant better for the owners then I’d have to agree with you.

      • 13
        John Autin says:

        fireworks, I think you hit the main points on the effects of the current FA-and-arbitration system. I’d just add this:

        I think the artificial scarcity of free agents drives up both their cost per year and their contract length. That double-whammy skyrockets the risk involved in the premium FA market, and that long-term risk may do more to keep out the low- to mid-revenue teams than the straight annual value.

        If there were more free agents per year, I think there would be fewer contracts of 6+ years, creating a tighter correlation between pay and performance. That would make it easier for budget-conscious teams to manage cycles of contention.

        Consider the Twins and Johan Santana. He would have been a free agent after 2008, going into age 30. After the 2007 season, the Twins still had a good talent base; they had won their division 4 of the last 6 years, and would win 2 of the next 3, narrowly missing the other. They might have been willing to extend Johan for 3 years at the top of the pay scale, which would have given them the ace they sorely lacked in 2008-10. But knowing the market would give him at least 6 years at top dollar — and that Mauer & Morneau were due to “get paid” in the next couple years — they dealt Johan.

        Player salaries don’t really concern me. Teams are going to maximize their revenue, and players are going to command a certain share of that revenue. But I would rather see more of the money go to productive players, and less of it tied up in long-term contracts that often go sour. Salary-dump and walk-year trades are a blot on the game, IMHO.

        • 14
          Mike L says:

          The six year system creates myriad opportunities for distortions in the market. One of my least favorite is holding back talented young players to game the service time clock. Salary dumps and weird late contract trades are others. I agree with John A that loosening up the scarcity just a bit (I would make FA at 4 years of service time) would tend to better equalize the market over time. If we are concerned about less well funded teams losing talent earlier than you can attach a loss of draft pick without a QO, or you can use an “embedded put” where the old team can can bind the player to the next year by giving him the higher of the qualified offer amount or an arbitration award.

          • 17
            John Autin says:

            Re: suppressing service time, does anyone have a real handle on how much of this goes on? Of course we hear about the high-profile cases like Wil Myers, but is there any way to quantify it?

            Here’s some data I just generated which *could* be germane to the question; I just can’t tell how to read them:

            For the last 60 years, I looked at the number of qualifying seasons (a) in their first year, (b) within their first 2 years, and (c) age 24 or under. Here are the averages per 30 team-years, for the last six 10-year periods:

            10 Years .. Q/1 .. Q/1-2 .. Q/24
            1954-1963 … 7 … 19 … 26
            1964-1973 … 3 … 11 … 30
            1974-1983 … 2 … 11 … 27
            1984-1993 … 2 … 12 … 22
            1994-2003 … 2 … 10 … 16
            2004-2013 … 3 … 14 … 20

            Note that the first five decades above all had an expansion year, while the latest did not.

            For both “first year” and “first 2 years,” the 2004-13 figures are at least as high as anything in the prior 4 decades. But the numbers qualifying age 24 or under are generally less than the prior decades. Interpretations?

          • 20
            Mike L says:

            John A, I’m not sure how to quantify it, although the Giants also did it with Posey, if I recall. One way (very cumbersome) might be to look at first year players who attained a WAR per 162 games of at least 3 and check when they were called up. That level of performance that early would imply they should have been in the majors. I would think that the present economic system would most incentive suppressing the “can’t miss” because they would have a presumptively higher earnings potential.

        • 16
          fireworks says:

          After I posted I was thinking I forgot to add contract length being an important factor with more free agents–it would absolutely stunt length. But then I was thinking that if you had a highly competitive free agent market that makes it all the more easier for the wealthier teams to overpay for the elites because the degree of overpayment should be less, and they can just throw higher yearly values at players. Guys would get to free agency so much younger and would be more willing to take a shorter contract that doesn’t quite take them out of their peak but has a monster average yearly value.

          You’re right about the system being quite inefficient at allowing you to look at a group of peers at the end of their career and see their salaries reflect their value. I think the most efficient possible system of compensation has to first start with tying player salaries to a percentage of revenue, and only from there you can reform the arbitration system (so that it includes massive raises for players of the caliber of Trout *and* salary cuts for players whose game falls apart like a Dontrelle Willis).

          As for Mike L below an appropriate end to the service clock nonsense would be to only tie players to their team from the date of their signing.

          Of course we all know none of this will never happen because the owners lose every contentious labor battle with the players due to their inability to actually bargain as a collective. The players have no interest in helping the owners be smarter about personnel decisions and kind of like a microcosm of the American Dream each member of the union is hoping he is at least good enough long enough to suddenly be able to be paid several times more than he had been being paid for what is too often a mere fraction of prior performance.

  3. 5
    Doug says:

    To Mike L’s point @3, the difficulty today with assembling high-end talent has likely led to more competitiveness throghout the leagues and also likely a higher replacement level. Thus, even if a team can assemble a “powerhouse” team, there is less likelihood of the superstars reaching the 7 WAR level (as shown in your results).

  4. 6
    --bill says:

    I think Bill James did a study on this–maybe in Whatever Happened to the Hall of Fame?

    He created two pitchers with identical career stats, one modeled on Don Sutton (the low-variance pitcher) and one on Steve Carlton (the high-variance pitcher). His conclusion? A team that wins about 85 games a year would win more pennants with a high-variance type pitcher, although it would also have more below-.500 years.

    Which raises a question: how hard is it to put together a team that regularly wins about 85 games a year? That’s about 33 WAR; if you can get 11 players who average 3 WAR, then you have maybe room for 2 position players and a starter. Would adding Eddie Murray, Lou Whitaker and Don Sutton get you a string of pennants?

    • 7
      John Autin says:

      Thanks, –bill.

      FWIW, my interest in this subject no doubt stems from the teams of my youth — the Tigers of the 1960s-80s, and the 1980s Mets.

      Those Tigers had a few big years, but at heart they were that “85-win team” with low-variance players. From 1964-88, Detroit averaged 86 wins per 162 games, but had just three individual seasons at 7+ WAR by position players: 7.0 by Bill Freehan in 1968; 7.5 by Al Kaline in ’67; and 8.2 by Alan Trammell in ’87.

      Detroit had a winning record in 20 of those 25 years, but all but 3 of those fell in the range of 83 to 92 wins per 162 games.

      In that span, these are the Tigers teams that reached the playoffs or came oh-so-close, and their best WAR by position players:

      1967 (1 GB, lost the pennant on the final day)
      — 7.5 Kaline, 6.0 Freehan, 5.0 McAuliffe, 4.0 Cash

      1968 (won it all)
      — 7.0 Freehan, 5.9 Northrup, 5.6 McAuliffe, 5.4 Horton

      1972 (won AL East)
      — 4.2 Freehan, 2.9 Rodriguez, 2.6 Kaline, 1.8 Cash & McAuliffe

      1981 (2 GB, lost the division on the final weekend)
      — 5.7 Whitaker, 5.7 Trammell, 5.7 Kemp, 3.6 Gibson
      (pro rated to 162 games)

      1983 (2nd at 92-70, 6 GB)
      — 6.7 Whitaker, 6.2 Lemon, 6.0 Trammell, 4.7 Parrish

      1984 (won it all)
      — 6.6 Trammell, 6.2 Lemon, 5.1 Gibson, 4.3 Whitaker

      1987 (won AL East)
      — 8.2 Trammell, 4.9 Evans, 4.3 Gibson, 3.8 Lemon

      1988 (1 GB, blew a 4-game lead after Aug. 21)
      — 5.9 Trammell, 3.5 Whitaker, 2.5 Lemon, 2.5 Nokes

      I adopted the Mets upon moving to NYC in 1984. From 1984-90 combined, the Mets had by far the best record in baseball, but they never had a 7-WAR player. And their pitching was not as star-studded as some people think; those seven Mets teams featured just two 6-WAR pitchers, Gooden’s 12.1 in 1985 and Viola’s 6.4 in 1990.

      It’s quite possible that I am stubbornly refusing to accept the lesson of this history. The Tigers of 1978-88 combined for the 2nd-best record in MLB. They very well might have won more than one championship in that time if they’d had more high-variance players — say, Ryne Sandberg in place of Whitaker (they had virtually the same career WAR rates), and Dale Murphy instead of Lemon.

      On the other hand, Sandberg and Murphy never even got to the World Series. So, who knows?

      • 8
        Richard Chester says:

        This is an aside to JA: Did you ever have a chance to look at the list of Tiger post-season WPAs that I mailed to you?

  5. 11
    fireworks says:

    As I was looking at the post and we got to the first set of seemingly counterintuitive results I wondered about the difference between a team of above-average players with no/few real stars versus teams that aren’t as strong overall, or on the bench, but have stars. If a star player struggles come postseason time his team should be on average less able to handle his struggles than the good team that doesn’t have star players. I’m not saying the data bears that out, but it was my first thought. The greater the impact of a single player the worst off you should be if they are ineffective.

    The other thing I wasn’t surprised was that the Yankees “screw up” the results. Not just that the Yankees tend to be successful with the high WAR players while the rest of the league doesn’t get great results, but I wonder, especially from ’23 through ’62, how often they denied teams with high WAR players from even reaching the postseason, let alone winning the World Series. Of course I don’t think you can break down the data well in that regard because in terms of breaking down things into team and World Series success, there’s pretty much just the early, formative, somewhat corrupt era, the four decade Yankees era, and the next thing you know you have the divisional and wild card era to throw off the results.

    • 21
      John Autin says:

      fireworks, good point about the Yankees thwarting other teams that had high-WAR players. Some notes:

      – Three other teams with three 7-WAR players failed to win the WS:
      — 1953 Dodgers went 105-49, but lost to the Yanks in a 6-game Series.
      — 1961 Tigers won 101 games, but finished 8 games behind the Yanks.
      — 2004 Cardinals may owe their fate to the Yankees, indirectly. 🙂

      – In all, the Yanks defeated four of 12 other teams that lost the WS with multiple 7-WAR players: the 1941, ’49 and ’53 Dodgers, and the ’61 Reds.

      – Multiple-7-WAR AL teams that lost the pennant race to NYY:
      — 1921 Tigers (a bad team despite some great players, went 5-17 vs. Yanks);
      — 1922 Browns (2nd, 1 GB, went 8-14 vs. Yanks);
      — 1923 Indians (3rd, 16 GB);
      — 1937 Tigers (2nd, 13 GB, 9-13 vs. Yanks);
      — 1939 Red Sox (2nd, 17 GB);
      — 1947 Indians (4th, 17 GB, 7-15 vs. Yanks);
      — 1949 Red Sox (2nd, 1 GB, 9-13 vs. Yanks, lost pennant in Bronx on last weekend);
      — 2001 Mariners (lost to Yanks in ALCS);
      — 2009 Rays (3rd, 19 GB);
      — 2011 Red Sox (3rd, 7 GB, blew wild card);
      — 2011 Rays (2nd, 6 GB, won wild card, lost to Texas in DS).

  6. 12
    Hartvig says:

    If I understood it correctly Bill James’ Win Shares were based on the expected number of wins that a team should have- thus a team that should have won 100 games would have more Win Shares than a team that won 90 games and so on. It didn’t necessarily correspond exactly with the actual number of games won but it was usually pretty close.

    Has anyone ever looked at WAR in the same light? Meaning does a team with X WAR usually finish with Y wins?

    • 15
      John Autin says:

      Hartvig — Yes, team wins is an integral part of the WAR concept.

      Replacement level is currently defined such that a team totaling zero WAR would expect a .294 W%, or 47.6 wins out of 162 games.

      Using that figure, and the team WAR batter & pitcher WAR totals available on B-R, I compared 2013 team wins (actual and Pythagorean) to their WAR-based projection (WAR+47.6).

      Actual wins — Every 2013 team was within 8 wins of their WAR projection, plus or minus. Twenty-one of 30 teams was within plus/minus 5 wins.

      Pythagorean wins — Every 2013 team except one was within plus/minus 5 of their WAR projection. The Cardinals were plus-12 in Pythag over WAR, thanks to their famous clutch hitting.

      The correlation between team WAR and reaching the playoffs was very strong this year. Ten of the top 11 in team WAR made the playoffs, and the other (Texas) lost the play-in game before the wild-card game. Four of the top five in team WAR won their division; the other division winners ranked 7th and 9th.

      2013 batters’ WAR: http://www.baseball-reference.com/leagues/MLB/2013-value-batting.shtml
      2013 pitchers’ WAR: http://www.baseball-reference.com/leagues/MLB/2013-value-pitching.shtml

    • 27
      bells says:

      I’ve been casually compiling some data in the last few months in my spare time, and going back the last 10 years for all of MLB, the overall correlation between Wins and expected wins based on WAR is about .91, varying from .89 in some years to .95 in others. I don’t have access right now to the file for the full breakdown (it’s on a computer in the computer lab I teach in which is closed until January), but suffice to say a) it’s a really good predictor and b) there are always outliers.

  7. 22
    DaveR says:

    The lesson being: If you’re going to have a big WAR year, be a Yankee.

  8. 23
    BryanM says:

    JA. I think you are trying to tease out a very subtle point from very noisy data. Recognizing that the inferior team can frequently win a short series, maybe achieving some regular season standard ( say 95 wins) might show things in a clearer light. Some other thoughts that might answer related questions , if not the one you set out to answer.-
    What pattern of team WAR totals leads to actual wins? Of the last 40 teams with 50 team War or more , how many games did the teams with one or two stars win vs those with a more balanced attack ?
    If we think of players who have had at least one 7-WAR season and put them into 2 groups.
    A). Guys who were about as good as Lou Whitaker, but with a higher peak/ fewer good years. ( the Ernie Banks group)
    B). Guys who were much better than Lou ( the Mickey Mantle group)
    Then it may not surprise us to find that it’s better to have Mickey., but your real question is., given players of about equal career value, what’s better , peak , or consistency. My own guess, FWIW , is that under the reserve clause, peak was better , but in current economics , consistency is better , since you are more likely to get in June the WAR you bought in December

    Nice study

    • 24
      John Autin says:

      Bryan M., that’s very well said. “What pattern of team WAR totals leads to actual wins?” is an avenue worth pursuing. And that’s an interesting thought about free agency changing the respective worth of various value patterns.

    • 25
      John Autin says:

      Bryan M., I’ve started a spreadsheet based on your suggestion. We’ll see where it goes, but here’s a random early return: This year’s Cardinals were very unusual.

      First off, they won 8 more games than predicted by their WAR, presumably from their unprecedented RISP performance.

      Next, they had 62% of their WARpos concentrated in two players, Carpenter and Molina. That’s very high. I’m looking at 124 teams in the wild-card era with a Pythagorean record of 90+ wins, and that 62% is 3rd-highest.

      And, despite all that “clutch” hitting, they fell 4 wins shy of their Pythagorean record. One reason is, their hitting was just average in late-and-close situations (furthering the case that the RISP work was not a repeatable skill), while their pitching was below average in those spots.

      WAR usually correlates better to Pythag wins than actual wins. But a lot of their run-producing value came from that .330 RISP average, which doesn’t show up in WAR. I’ll guess that the 12-win gulf between their Pythag and their WAR-expected wins will be one of the highest in this pool.

      • 26
        BryanM says:

        JA , I think your decision to classify teams by pythag wins rather than actual ( if I understand you correctly) is very sound . The chain events- predicted runs – predicted wins should compare better to Actual runs – predicted wins , than to actual wins . The 2013 Cards got more wins out of Allen Craig than his WAR would suggest – counteracting the apparent concentration in Carpenter and Molina. Hopefully , the law of large numbers will overcome such anomalies and a pattern will emerge. Again a guess – top of the order WAR is better than. High OBP five hole WAR. ( apologies to Gene Tenace , who played on some winners)

  9. 28
    bstar says:

    I think it’s pretty easy to tear away at this whole “peak is better” thing.

    I’d like to comment on Bill James’ study described by bill @6 where he trades out a low-variance pitcher like Sutton for a high-variance one like Carlton.

    First, I think this is just one of those number-massaging exercises that James liked to do. Another example would be him creating a career for a fictitious player by using the first 10 years of one player’s career and then stapling on the last ten of another career. Stuff like that. I think it was meant more for fun than anything definitive.

    But let me comment on the conclusion that James makes, that more pennants *may* have been won by that 85-win team.

    -it’s a good thing he picked an 85-win team because if he used a 65-win, 70, 75, probably 80, or maybe a greater than 97-win team there would have been little to no effect on pennants won.

    So having one high-variance pitcher instead of a low-variance one only matters for pennant-winning if you’re in a specific range on the win curve (I don’t know, maybe 82-98 wins?). Otherwise it’s completely irrelevant.

    Here’s what I mean: Carlton’s transcendent 1972 season did not produce a pennant for his last-place Phillies. His almost-as-good 1980 season did (the Phils won the NL East by 1 game), but what % credit does Carlton deserve? I don’t know, but it’s not 100%! His third-best season, 1968, also failed to produce a pennant.

    -Every roster has high and low-variance players. Each contribute to the team’s success. I think that most GMs would want a mix of those two types of players, so one could argue that each have value. But suppose you’re a true-talent 90-win team in today’s game where five teams make the postseason in each league. You’re playing in an easy division where 88-90 wins usually takes home the prize.

    Isn’t having a roster of low-variance players preferable at this point on the win-curve? Simply making the playoffs and winning 90 games twice is better than winning 96 and 84 because that 84-win team probably doesn’t make it to the dance. I don’t know about you, but I would certainly rather have my team make the postseason twice as the 2 or 3 seed than be the #1 seed one year and then miss out the next.

    -If this study was in fact in the Politics of Glory, there was no such as WAR or even Win Shares. I’d say Bill James chose Carlton and Sutton because they both pitched from 1966-1988 and both had a similar number of wins. But Carlton has 84 WAR to Sutton’s 69! I would want Steve Carlton for 23 years also, and I might expect to win an extra pennant somewhere along the way.

    -if peak does have more value it is probably overstated in today’s game because winning a pennant/advancing to the Series now involves winning multiple rounds of playoffs. And, as JA’s study proves, teams with high-WAR players don’t necessarily perform better in the playoffs.

    I dearly hope there is more/actual evidence out there that high-variance players create more pennants, because it seems that most are treating this whole idea as unchallengeable and sacrosanct. It’s not. I heartily applaud JA for trying to find actual evidence of it.

    Can someone post a link to another study done on this subject? I’d love to read it. Until then, I’m going to continue treating peak value as nothing more than a tie-breaking edge for one player over another with similar career value.

    • 29
      Dr. Doom says:

      Hmmm… I think it’s pretty easy to tear down the whole “peak means nothing” thing.

      Let’s imagine two players. Both have 55 WAR.
      Player A plays 22 seasons, with exactly 2.5 WAR per year.
      Player B plays 10 seasons, with 10, 9, 8, 7, 6, 5, 4, 3, 2, and 1 WAR.

      Player A would literally never have been in the conversation as best player in the game. He almost certainly never would’ve been an All-Star.

      Player B would have been in the running for best player in the league at least 4 times, maybe 5. He would’ve played in 5-7 All-Star games.

      You’re SERIOUSLY suggesting there’s no difference between these two players?

      Plus, what if we move the level of comparison? Average is, basically, 2 WAR per year. Player A would have 11 WAA in his career. Player B would have 35 WAA. Again, you claim no difference?

      Finally, what if replacement level were just a LITTLE bit higher? B-Ref sets it at .3XX (I don’t know exactly what it is). What if they set it at .400? Then Player B, just like with WAA, would have much more than Player B? I just think there’s a great deal of reason to believe that peak means something, even without doing a Bill-Jamesesque study.

      • 30
        bstar says:

        So now you’re using WAA as your measure of peak?

        That’s interesting, Doom, because if you do that John Smoltz (38 WAA) has a higher peak than Juan Marichal (30 WAA). Didn’t you conclude Marichal’s peak was higher in the COG thread based on higher WAR in his top years?

        This proves that you can reach different conclusions about which player can have a higher peak based on how exactly you define what peak is. So “peak” is ultimately a somewhat nebulous concept.

        Can you find me two players with 55 WAR and a WAA difference of 24? That’s pretty unrealistic, but of course I would take player B. I did say I use peak for tie-breaking between two players of similar value.

        • 31
          Dr. Doom says:

          I’m not using it that way; I’m suggesting that it’s a possible way to do it. Adam Darowski does it that way, and a lot of people seem to like what he does.

          The thing is, I think it’s obvious (and you’re welcome to disagree with me) that peak matters. When we’re ranking players, I’d put a lot more weight on someone being in the conversation as the best player, even if only for a year or two, than a guy who was consistently average. And since what we’re doing in the COG essentially amounts to ranking players, I think peak matters a great deal.

          Essentially, I just think it’s a little too simplistic to just look at career WAR (or whatever) and say “BOOM! There it is!” without considering the context of that total value. Real world example: Tommy John’s career WAR is 9 wins higher than Sandy Koufax’s. I don’t think anyone seriously suggests that John is more deserving of the COG than Koufax. If you factor in peak, I would think Koufax does better than John. By straight WAA, Koufax tops John by 9. So yeah; I think peak matters.

          • 34
            bstar says:

            Doom, I hope you’re not painting me as THAT GUY who looks only at career WAR and nothing else (does someone like that exist?).

            But if I did, how off would I actually be? Sure, you would end up picking Chuck Finley over Sandy Koufax or something like that but you would certainly do a better job of pinpointing great careers than the BBWAA has done, no?

            It’s really interesting that so many people poo-poo on the idea of using only WAR, but I think you could do a lot worse. While only using WAA or only using a 7-year WAR peak *might* be a better way than only using WAR, I do think each of those methods can lead to some questionable conclusions as well.

          • 36
            Dr. Doom says:

            No, star, I don’t think you’re “that guy;” you’re WAY too thoughtful about baseball to be that guy. But “that guy” exists. I don’t think he hangs out here, but he does.

            And you could do MUCH worse than career WAR as your only criterion. Of course you could. I just think a career WAR that’s balanced with peak WAR is a BETTER day to go.

            Joe Posnanski and Tom Tango both wrote about peak value in the last couple of weeks. I’m too lazy to look the actual posts up right now. But the gist of Poz’s post was that the NFL elected Charlie Joiner based on career value, rather than peak. Now people look back at that election, and it looks dumb, because Joiner’s not the career leader in anything anymore, AND his peak was not nearly as good as some contemporary receivers. Likewise, Tango wrote about hockey (a sport I know roughly zero about) and how peak value is really all that matters in ranking players. He gave some germane example, but again, it went right over my head, even though I recognized the names. His point was that in other sports, being the best (or among the best) for a short period was preferable in other sports, but that baseball fans have a fascination with career totals, and that that may not be the best way to prioritize. I think those points are accurate; when we think of Mickey Mantle, we care a whole lot more about the best than the rest. Willie Mays is remarkable because he was just as good at his best, but did it longer. That’s my perspective on it. I also think, intuitively, that higher WAR seasons would lead to more pennants (or playoff appearances, as the case may be), but I don’t have the coding skills to run the simulation it would require, so I can’t prove it. But regardless, I think peak matters. You’re certainly entitled to your opinion, and you’re definitely right that most of the time career WAR is all you need, but that’s because MOST players have similar career arcs. For the exceptions (on both ends: both your Koufaxes and your Whitakers) I think peak is something to keep in mind.

          • 43
            bstar says:

            I agree with what you say, for the most part, Dr. Doom.

            I just think we disagree on how much to weigh peak. I think I’ve shown it’s very possible we’re overrating it, especially considering you can reach the opposite conclusion when comparing two players side by side if you simply change your definition of peak. Here’s another example of that:

            Sandberg: 38.1 WAA
            Whitaker: 42.4 WAA

            Sweet Lou gets the pennant-winning peak boost.

            Sandberg: 46.7 WAR7 (seven best seasons)
            Whitaker: 37.8 WAR7

            And, FWIW, there is nothing dumb about Charlie Joiner being in the NFL Hall of Fame. He and Art Monk unequivocally belong.

        • 32
          bstar says:

          Here’s two that might work: Sandy Koufax and David Wells.

          They both have around 53 WAR but Wells pitched 21 seasons while Koufax pitched 11. And, yes, of course Koufax has way higher WAR peaks, but the WAA difference isn’t nearly as big as your example: 22 for Wells and 31 for Koufax.

          Of course I would take Koufax, but I would campaign loudly for a spot in the Hall of Darn Good for Boomer.

          But my post way more about tearing the idea that peak value yields more pennants. I’m VERY OPEN to being wrong about this, Doom; I would just like to see some actual evidence that proves it.

      • 35
        BryanM says:

        It’s certainly possible that peak is better than consistency, but we still need a method to demonstrate it — to cite your example , as a GM , I would much, much rather have the 2.5 WAR guarantee than the 10-9-8….1,0,0 lottery guy, who, admittedly , has shown much greater potential, but has not shown greater value. Not that WAR is the be all and end all, but the recipe of putting pieces together to create winning baseball is exceedingly complex, and whether higher peak as opposed to more consistent players leads to more wins is, I beleive, undemonstrated.
        Of course, on other threads, we are talking about ranking players , as opposed to the impact of big WAR years on winning, and for that purpose, I think higher peak is a useful tiebreaker .

        • 49
          mosc says:

          Can I talk about how absurd this is for a second? Nothing personal Byran, I love your stuff, but this is really incorrect thinking.

          1) You don’t sign a guy to a 22 year contract. The 10/9/8/7/6/5/4/3/2/1 guy is out of the league after 10 years and then you sign somebody else to play the other 12. Maybe he gets 12 war in 12 years, that’d be a below average pick, but that’s still 67 war vs 55 war over those 22 years.

          2) low WAR does not fully indicate value. If a guy is a great defender but a terrible bat, or vice versa, you can use your bench to compliment those skills and get some additional value. Maybe it’s a bad defensive guy who slaughters left handed pitching and racks 2.5 WAR per year in 400 at bats and half a season’s innings in the fields. You can suppliment him with a good glove or a lefty bat and get a lot of WAR out of the position even without a star. On the reverse side, a guy who is not exceptional at anything but durability soaking up 600+ at bats to get you 2.5 WAR is helpful no doubt but is not nearly as valuable.

          3) WAR/AB matters! It matters a lot. A guy going 10/9/8/7/6/5/4/3/2/1 probably has something like 3x the war/AB of our 22 year 2.5 guy.

          4) Prospects have their own kind of logical thinking that is not always intuitive. You get 6 years of control, 4 of them cheaply. That puts extreme weight on the front end of their career value wise to the team. It also means you could care less what a prospect will do in his 30s when you’re deciding on current value to a team. Lets say both of those guys are at spring training for your team this year. You’d certainly be thrilled to take Mr 10WAR rookie and boot the 22 year vet down the road. You can release the peak guy after 6. You’ve gotten 45/55 WAR out of him and he’s gunna be no more than an average player the next 4 years. If you release the other guy, you’ve gotten virtually nothing above average out of the position for those 6 years.

          Even if you had to spend $100m over 22 years for each of those two guys, you’d still be insane to choose the 2.5WAR guy. You can spend ~$20m more and have a league average player fill in for the 12 year back end and net yourself 20-some WAR out of it at a considerably better value (WAR/$)

          • 50
            birtelcom says:

            Mosc, here’s a point-by-point reply as to why I think the concept of WAR supports Bryan’s suggestion that a guy who produces 55 WAR steadily over 22 years is likely equally as valuable as the guy who produces the same 55 WAR over ten years and then is done:
            (1) You argue that “you sign somebody else to play the other 12”, and then assume you can get a guy at above-zero WAR for those other 12 years. But that assumption is inconsistent with the concept of WAR. The very essence of the WAR concept is that the default alternative “other guy” is a 0 WAR producer. If that “other guy” is on average to be expected to produce 1 WAR a season, then WAR is not being correctly calculated in the first place. So when I’m discussing the contrast between the 55 over 22 years guy and the 55 over 10 years guy, I’m always assuming that my alternative for those last 12 years, after the 55 over 10 guy is done, is producing 0 WAR over those last 12 years. If you don’t use that assumption, you’re not really talking about WAR anymore.

            (2) Here again, mosc, you seem to be assuming that you can “replace” low or zero WAR years with a combination of guys that will get you more than the 2.5 WAR the 55 over 22 guy gives you over the full 600 or so PAs of a full season. But the problem is the same as in point (1). WAR is called Wins Over “Replacement” for a reason –if a team could generally piece together a replacement package more valuable than WAR is assuming, then we are off in our WAR calculations.

            (3) I don’t know why you assume WAR per AB (or WAR per PA) matters to this discussion. I agree that the 55 WAR over 10 year guy generates more WAR per AB or per PA than the 55 over 22 guy. But if one assumes that over the full 22 years a team has to replace all the ABs and PAs for which the 55 over 10 guy is not around, and you assume that those ABs or PAs are replaced at 0 WAR, then the total WAR per AB or per PA is the same for the team in the long run.

            (4) Your introduction of the prospect/salary/free agent issues certainly adds some realism to the discussion but now you are no longer comparing the 55 over 10 year guy and the 22 over 10 year guy in an apples-to-apples comparison. You are now assuming you are paying less per WAR for one than the other, so of course the one you are paying less per WAR for is more valuable — but that wasn’t the question.

            Lastly, you close by suggesting that the market is valuing WAR differently for the 55 over 10 guy than the 55 over 22 guy. That may or may not be correct, but if it is, that may simply be a flaw in the market’s valuation of some guys versus others rather than proof that one guy is actually more valuable in terms of winning than the other.

  10. 33
    John Autin says:

    INPUT NEEDED! I’m looking into a team-based WAR study along these lines: “Teams compiling this range of WAR averaged this much from their #1 WAR guy, this much from #2, etc. They averaged this much from their top 3, this much from their next 3, etc.” So far, I’m just thinking about position players, and the wild-card era.

    Here’s the thing — I’ll call it the Jeff Francoeur Problem: If you just rank every team’s players from most WAR to least, unproductive regulars will rank below some guys who barely played. I think that would cloud the issue. There are about 160 regulars in the era with -1.0 WAR or worse.

    I’m inclined to do each team’s player WAR rankings this way: Spots 1 through 8 are reserved for the 8 guys who had the most PAs, with those 8 ranked in WAR order. Spots 9 through whatever are a WAR ranking of their remaining guys with, say, 100+ PAs.

    Here’s one reason I favor that method: One characteristic of good teams is that they rarely stick too long with an unproductive regular. They have the smarts to recognize a problem, and either the depth, the trade resources, or the budget to address the problem.

    When a guy plays every day and slogs to a minus-2.3 WAR as Frenchy did in 2012, I think a WAR study should put that up with the main data, not buried at the bottom of a list surrounded by scrubs.

    What do you think?

    P.S. I’ve gathered the team WARpos data for the wild-card era, and the first thing that stands out is this: The 2001 Mariners absolutely lapped the field. Their 50.9 WARpos is 9.4 more than any other team in the period. There are 19 other teams between 35.2 and 41.5, but the M’s are off the chart. So, another question: When I make a study group of the 20 or so best teams of this era in WARpos, should I even include those Mariners, or just treat them as sui generis? I don’t know if they set the record for WARpos, but it’s more than the ’27 or ’98 Yankees, ’75-’76 Reds, and any other great team I can think of to check.

  11. 37
    Mike L says:

    Interesting discussion. I want to throw a few questions/points out there. With no data, one way or another, to support it. First, peak WAR as it relates to pennants has to be somewhat contextual, with players like Ernie Banks (or Carlton) being excellent examples. It’s implicit in argument that we hear made over and over (Cano)–why waste money on a guy who is going to take you from a 71 win team to an 78 win team if you don’t have other talent to surround him? The second–going back to Doom and Bstar’s argument over peak/career, I would argue that if I were a GM, I would sooner go for consistent value at a hard to fill position. So, if you show me a 3-4 WAR catcher or middle infielder who could stay at that level for a dozen years, I’m really interested. Jeter, for all the (justifiable) criticism he takes, managed at least 3 WAR for 14 consecutive years. Ivan Rodriguez had only one year below that (2.7) in 13 consecutive seasons. As to John A’s question about 2001 Mariners, I think it stays in, although it’s a bizarre team with a lot of career or near career seasons. Bret Boone (ahem) managed 8.8 WAR, starting a three year stretch, from 32-34, of 18.9 WAR, out of a 14 year career WAR of 22.6.

  12. 38
    John Autin says:

    I think terms like “high-peak” and “among the best in the game for a while” can also lead to oversimplification. I think there’s more gradation on both sides of that line than the discussion often lets in.

    Take Ryne Sandberg and Lou Whitaker. Their career totals and rates in both WAR and WAA are very close; Lou played a couple extra years (and pretty good ones), but if you just take their first 15 full seasons, they’re virtually the same.

    Ryno definitely had a higher peak, and was rightly considered among the best in the game for some years. But that doesn’t mean his peak was the same or close to those of others who were in that “best-in-game” discussion during Sandberg’s career. Sandberg’s best year was 8.5 WAR, which led the NL in 1984. And he had 3 other years over 7 WAR, ranked 2nd or 3rd in the NL. But none of his peak years cracked the top 20 WAR seasons during his career. He was never “the” best, by WAR; in ’84, Ripken topped him easily, and for Ryno’s other 3- to 4-year peak (1990-92 or 1989-92), he was topped by Bonds and Ripken (and Henderson for the 4-year span).

    Meanwhile, the simplification that occurs in the other direction is to treat a player who lacked a high peak, who was never “in the discussion,” as basically an average player or a little above that. But this is obviously not true of Whitaker. Run the WAR numbers for MLB second basemen year by year from 1978-92: Whitaker ranks 5th or better in 14 of 15 years (1st twice). You wonder how he only made 5 All-Star teams.

    These are unusual cases — Sandberg on the low end of the high-peak club, and Whitaker vice versa. Whatever “pennant advantage” there may be for a high-variance distribution of a player’s WAR, I very much doubt that such factor would make a significant distinction between these two players.

    So, the question I’d most like answered is, just where does that factor kick in, along the spectrum of high- and low-variance differences among players with similar career WAR stats? I think there might be a pennant advantage for, say, Sandy Koufax over Addie Joss — but what about Steve Carlton vs. Nolan Ryan? Or Gaylord Perry vs. Warren Spahn?

    • 39
      fireworks says:

      Hard to shake out how should consider peak performance vs consistent performance but we obviously have a natural bias toward peaks because they catch our attention more easily. It’s very similar to the way we have a natural bias toward exalting players that excel in one or two areas of the game versus those that are good in many facets. In that regard in terms of a player’s contribution there are also peak performance (in an area or two) versus consistent performance (in many areas).

      I have started and stopped on several different lines of thought trying to write my response, including the idea that peak vs. consistent for two players that ultimately have similar value is impossibly subjective. I think in the end it just has to come down to what you want to weigh as there will always be data to support either peak or consistency.

      In my mind one real-world con against peak is that great big seasons and the peak performers who have them have a naturally greater weight in one’s mind and thus impede to varying degrees appropriate analysis. That is to say that all of us as analysts, amateur or professional, sit here and try to find objectivity and deny our guts, our sentiments, our nostalgia. Well, not all of us. Some of us are elderly members of the BBWAA or talking heads on ESPN and MLB productions. In which case we are wont to appeal to gut and authority. But I digress.

      I tell people that you never really know precisely who a person is at that precise moment in their life. This is true even of the individual introspecting about himself. Ultimately our perception of whom an individual is is colored by, is filtered through whom we previously decided they were. The stronger that prior idea, the more concrete it is in our minds, the more difficult it is to assimilate new information into it.

      We’ve seen that sort of concept apply to why a player like Dwight Evans received no HoF support while his teammate Jim Rice did. Evans took a while to get his bat up to HoF standards and by the time he did he had already had his “good not great” player label attached. Rice, on the other hand, came out of the box swinging and got his label as a powerful middle-of-the-order bat early on.

      I’ve gotten away from peak vs. consistency in favor of the idea that once you get a label it’s hard to change it but my point is that peak performances create stronger labels and thus a greater bias and it’s a con when it comes to denying one’s gut.

      Basically, what I mean is that I have pretty much nothing to contribute to answering you question. LOL. Sorry JA.

      • 41
        John Autin says:

        Fireworks — re: “Ultimately our perception of whom an individual is is colored by, is filtered through whom we previously decided they were.” Thanks for crystallizing something I could not get into words.

    • 47
      tag says:

      John,

      I think any Ryno/Lou WAR comparison needs more context. Without getting into the “Ryno led his team to two division titles” argument, which is quite unfashionable and can get one mocked in saber circles, it is undeniable that Ryno was by far the best position player on his team(s), and was needed to be or the Cubs had no chance of winning. That level of “betterness” above teammates does make a major difference, in my view.

      In 1984, Bull Durham was the second-best everyday Cub and posted 3.7 WAR to Ryno’s 8.5. The Cubs got within a game of the WS (and earning the honor of being beaten by Lou and your Tigers), even with Larry Bowa and his -1 WAR at short and Keith Moreland and his 0.5 WAR in right. Ryno and some good pitching were the main reasons they were there.

      In 1989, Ryno again fronted a mediocre Cub crew to the NL East crown. Again his first baseman chipped in (Mark Grace to the tune of 3.9 WAR) and the young Greg Maddux and an out-of-his-mind Mike Bielecki did their part. But no other Cub regular had more than 2.3 WAR.

      The fact that Ryno during these times was bested in his WAR totals by such players as Ripken! Henderson!! and Bonds!!! in no way detracts from his greatness.

      Lou Whitaker had the luxury of playing with several quality players for a major part of his career, two of whom, it can be reasonably argued, were better than he was during their Tiger residency.

      Your post wondered about the impact of big WAR seasons. Ryne Sandberg is living proof that they can make a difference, and to me his having such seasons easily elevates him above such marvels of consistency as Lou Whitaker.

  13. 40
    Insert Name Here says:

    I only skimmed the comments so maybe someone else pointed this out already, but I think it’s problematic to find the % that made the WS with 7-, 8-, 9-, or 10-WAR players. This is problematic because for any season where multiple teams in the same league fit your criteria, only one of them can win the pennant, driving down the % of teams that make the WS. To remedy this, you could only look at teams who were the ONLY team in their league that season to meet your criteria, OR count teams that finished behind only teams of the same criteria as having “won the pennant” for the purposes of your analysis.

    • 42
      John Autin says:

      Thanks, INH — good suggestions there. I haven’t yet drawn the question tightly enough to study properly.

      • 44
        Insert Name Here says:

        Unfortunately, that’s the simple truth. As you pointed out at the beginning, this is a very tough concept to tackle. If you had the time or a lot of people to help, another possible study would be to study each season individually for some defined correlation between individual players’ WAR and teams’ success, and track how directly these correlate over time. The results would be very interesting and possibly very significant for the world of sabermetrics, but even if you evaluate since-1901 only, that’s 113 seasons to evaluate, which is a lot even if you divide them up among, say, 10 people.

        If only time weren’t so valuable….

      • 45
        Mike L says:

        John A, maybe you work backwards. The worst WS-bound record, if I’m not mistaken, in the 162 game era, is the 83-79 1973 Mets (who had Tom Seaver, with 11 WAR.) If you eliminate all records (teams and +7 WAR performers) below that, and create three ascending tranches of 8 wins, you might see a correlation within the tranches. And, if you start in 1962 as a starting point, you get out of most of the distorting 1923-1964 Yankee dominance.

      • 46
        Richard Chester says:

        Would you like a list of pennant winners with players of WAR of 7+?

Leave a Reply

Your email address will not be published. Required fields are marked *