Weighted WAR: Negative Seasons and Center Field Rankings

In the discussion of weighted WAR, we’ve covered the weighting system, schedule length, and timelining (part 1 and part 2), plus an aside to touch on positional classification. Now, we’ll finish up the discussion of WAR adjustments with one more topic: negative seasons, or more specifically, the reasons that I discard negative seasons from the rankings.

To be clear up front, I understand why negative WAR scores exist and agree that they should. It is in fact possible that a player can perform poorly enough to expect that a minor league journeyman would do better. Negative WAR values are relevant in any number of contexts – evaluating a team’s overall performance, projecting production moving forward, and so on. I just don’t think they add anything useful to weighted WAR’s evaluation of the player’s career.

Negative WAR seasons can be broken down into three hypothetical categories, which we’ll examine in turn. The first is the simplest: the WAR value is incorrect and the player’s overall contribution was actually positive. This could be due to measurement error in the categories WAR analyzes (say, a fielding score of -20 that should actually be -5), or due to a blind spot in the WAR system in question (such as pitch framing for catchers, or saving errant throws for first basemen). Either way, in this situation it is straightforward to see why a negative WAR value should not be held against the player.

For the second case, assume the WAR value is correct, in that the player’s performance is below hypothetical replacement level. However, in this case, assume that the player’s team does not actually have a better alternative available at his position; say the player puts up -0.5 WAR at shortstop, but their best utility infielder or AAA shortstop would be expected to post a -2.0. In that case, the player’s performance is actually better than the real replacement level available to the team, and the negative WAR value should not be held against him. (To be clear, I’m not campaigning for assessing replacement level on a team-by-team basis in a WAR system, just trying to examine a case in which a sub-replacement performance might not be detrimental.)

In the third case, assume that the WAR value is correctly negative and the team also has better options readily available. In that case, why is the team still using the player in a prominent role? Leaving aside the extreme case of player-manager Pete Rose writing himself into the lineup, most players are not in control of their own playing time. If the player isn’t producing and management keeps throwing him to the wolves anyway, I tend to think that the negative WAR value also should not be held against him in a big-picture evaluation of his career.

Since we’re in center field, consider the alarming case of Jim Wynn. Starting in 1965, the Toy Cannon posted the following OPS+ figures over an 11-year span: 144, 116, 138, 158, 167, 141, 72, 146, 107, 151, 134. That is good-to-spectacular production over an extended period, with one glaring exception. In 1971, Wynn’s BA/OBP/SLG line was .203/.302/.295, unacceptable even for the Astrodome in a low-scoring period. His WAR total fell off accordingly, dropping from 7.1 and 4.9 the previous two years down to -0.6. What happened? In the 1970-71 offseason, Wynn was stabbed in the abdomen by his wife, and took the entire season to recover, both physically and emotionally. This is a fairly predictable outcome, and could reasonably have been anticipated by his team.

Of course, not everybody is Jim Wynn; there’s not always a glaringly obvious and traumatic explanation for a sudden drop in production. But these things are often accompanied by an injury of some kind (see Cody Bellinger in 2021 for another CF example – injured shoulder during postseason celebration leading to a 44 OPS+ the following year). Regardless of the exact cause, it is ultimately the team’s responsibility to distribute playing time accurately, and if they insist on bringing a young player up too early, or sticking with an injured player or declining star for too long, I don’t think the player in question should be penalized in comparison to one whose team is better at assessing when to cut bait.

I’ll make a few additional system-specific arguments for discarding the negative years. First, the weighting system doesn’t combine well with negatives. Wynn’s mid-career negative season came at age 29. His MLB career lasted 15 years; 1971 ranks #14 among them (his final year, 1977, was worth -0.8 WAR). According to the general rules of the weighting system, 1971 should therefore be weighted at 35% strength. But if we’re including negative numbers, WAR would hold that Wynn had better seasons in 1962 (when he played for Tampa in the Florida State League) and 1979 (Coahuila in the Mexican League); he may not have been a major league quality player, but he didn’t cost MLB teams any wins in either of those seasons. To take the position to its logical extreme, Wynn was a more valuable player in 1951 (age 9) and in 2011 (age 69) than in 1971 (age 29), and those unplayed seasons should therefore rank above the negative years in the weighting system. If you do that, and weight all negative seasons at the minimum 5%, you might as well drop them entirely, as a player with 10 WAR of negatives (the most I’ve seen from a modern player who had at least one +3-WAR season and therefore qualified for the database) now loses half a point from his score, which is enough to drop the #50 CF all the way to… #53.

Second, note the “modern player” caveat in the previous point. Thanks to the timelining adjustment, there is actually a fourth category of player who can have a negative WAR score: a player with a non-negative score that gets timelined below 0. This is a player who actually helped his team win more games than the replacement level that would have been available at the time; even moreso than in the other cases, I’m not particularly interested in penalizing such a player, especially since the penalty is coming from an adjustment that is mathematically shaky at best.

Consider the case of 19^th-century star Paul Hines, who ranks #38 in center field. Hines was a consistently excellent player, but has three major down years in his career, which conveniently follow the patterns explored above: 1872, when he was 17 and played poorly in 11 games; 1877, an unexpected mid-career downturn; and 1890, at age 35 and at the tail end of his career. Hines’s basic WAR totals in these seasons were -0.3, 0, and -0.9, respectively. With length adjustment, those numbers would be -1.2, 0, and -1.0. But with timelining factored in as well, the totals become -1.7, -0.9, and -2.6, for a total of -5.2, the largest set of negatives among top-100 center fielders. I’m inclined to discard these years for the reasons laid out previously, but even if you both count them and don’t force the 5% weight on them arbitrarily, Hines loses merely half a point from his score and still ranks #38 in center field.

The type of player who benefits most from discarding negative seasons is not actually in the Hines model, the player who came up a year early and/or held on a year too long and/or had one significant mid-career down year. The type that gains the most is someone whose normal performance is average or a bit below, but has a high variance, tending to mix excellent years with awful ones. The quintessential case here is Jose Guillen. At his best, Guillen was legitimately terrific; from 2003-07, he had four seasons of at least 3 WAR. The issue was, well, all of his other seasons. Guillen’s 1997 rookie season grades out at -3.3 WAR (partly due to an alarmingly bad fielding score which doesn’t match the surrounding years); he didn’t get into positive territory until 2000, and then only barely. The year before his peak started, he posted -1.4 WAR across two teams in 2002; even in the middle of his excellent stretch, he put up a -0.9 in 2006. After 2007, teams gave him a few chances to recapture his peak form, and were punished each time, to the tune of -0.5, -2.0, and -0.4 WAR, at which point Guillen was finished. Guillen’s final totals were +16.6 WAR… and -10.3 WAR, for an overall 6.4 (rounding errors). Discarding negative WAR totals treats Guillen like a 16-WAR player when he definitely wasn’t one. Especially toward the end of his career, the memory of his peak might have reasonably caused teams to hope for a resurgence, and it never came. One could make the argument that negative WAR should count against a Guillen-type player, or anyone else for as long as a basic projection of performance would be above replacement.

I am sympathetic to this idea, but will offer two counterpoints. First, going that granular on every player with a negative season would require far more effort and skill than I would be able to invest. And second, the effects on players I’m interested in for the purpose of this exercise would be minimal. Even with the boost of ignoring his negative scores, Guillen is not among the top 100 right fielders. Neither is Chris Davis (8.6 removed negative WAR) at first base, nor Alfredo Griffin (8.0) at shortstop, nor Deron Johnson (7.6) at first, nor either Bobby Tolan (7.2) or Cito Gaston (6.9) in center. The same is true for the plethora of much older players mixed in among these totals. Only one player in the top 75 for the negative WAR adjustment makes the top 100 at his position (#79 3B Davy Force, who like Hines largely suffers from timelining, and who like Hines probably wouldn’t actually lose a spot in the positional rankings even with the negatives factored in.)

For the sake of completeness, here is the list of top-100 center fielders with at least two wins worth of negative adjusted WAR:

Player	Years	-aWAR
Paul Hines	1872-91	-5.2
Marquis Grissom	1989-2005	-4.1
Jim Piersall	1950-67	-3.8
Paul Blair	1964-80	-3.7
Willie McGee	1982-99	-3.5
Andre Dawson	1976-96	-3.3
Marlon Byrd	2002-16	-3.0
Dale Murphy	1976-93	-2.8
Dave Henderson	1981-94	-2.8
Amos Otis	1967-84	-2.3
Steve Finley	1989-2007	-2.2
Max Carey	1910-29	-2.2
Al Oliver	1968-85	-2.0

And now that we’ve quite literally dispensed with the bad, on to the good! Here are the active center fielders who are either in the top 100 or within a reasonable distance of that point:

Player	Rank	Years	WAR	aWAR	wWAR	2024 WAR	Rank Change
Mike Trout	5	2011-24	86.2	88.3	69.2	1.1	0
Andrew McCutchen	30	2009-24	49.3	49.7	39.8	0.8	0
George Springer	47	2014-24	67.4	39.9	32.6	1.1	+3
Kevin Kiermaier	51	2013-24	36.7	38.8	32.0	1.0	+3
Cody Bellinger	86	2017-24	24.5	27.9	25.1	2.2	+9
Byron Buxton	94	2015-24	24.7	27.4	23.7	3.6	+17
Brandon Nimmo	105	2016-24	23.2	25.3	22.1
Charlie Blackmon	124	2011-24	21.2	23.0	19.6

Not a lot of active players listed, but CF is a tough top-100 to break into. Nimmo’s score would put him in the top 100 at most other positions, including both corner outfield spots (he’s spent time at both of them but is not close to being reclassified). In center, he’s still a year away. And with Blackmon’s retirement, there are no other active players in the top 150 here.

Also, with how often he’s injured these days, it is increasingly easy to forget just how good Mike Trout was at his best. The answer: very, very good. Here’s how he stacks up among the top 25 overall center fielders (plus the usual bonus entries from 30-100):

Player	Rank	Years	WAR	aWAR	wWAR
Willie Mays	1	1951-73	156.3	154.5	96.1
Ty Cobb	2	1905-28	151.4	137.6	82.0
Mickey Mantle	3	1951-68	110.5	109.6	77.2
Tris Speaker	4	1907-28	135.1	122.4	75.5
Mike Trout	5	2011-24	86.2	88.3	69.2
Ken Griffey Jr	6	1989-2010	83.4	87.1	63.9
Joe DiMaggio	7	1936-51	79.2	74.5	56.3
Oscar Charleston	8	1920-41	48.8	70.3	55.9
Kenny Lofton	9	1991-2007	68.2	70.4	52.0
Andre Dawson	10	1976-96	65.0	70.3	51.9
Carlos Beltran	11	1998-2017	69.8	70.7	51.5
Duke Snider	12	1947-64	65.9	64.7	51.2
Turkey Stearnes	13	1923-40	48.8	75.0	51.1
Andruw Jones	14	1996-2012	62.6	63.9	49.7
Richie Ashburn	15	1948-62	64.4	63.0	48.2
Larry Doby	16	1942-59	57.8	64.3	47.3
Jim Edmonds	17	1993-2010	60.7	61.4	47.2
Jim Wynn	18	1963-77	55.6	56.2	44.9
Billy Hamilton	19	1888-1901	63.3	57.0	44.6
Chet Lemon	20	1975-90	55.8	57.9	43.7
Willie Davis	21	1960-79	60.7	59.0	42.4
Cesar Cedeno	22	1970-86	53.1	53.2	42.1
Kirby Puckett	23	1984-95	51.2	52.4	41.6
Vada Pinson	24	1958-75	54.2	53.0	41.5
Willard Brown	25	1937-48	24.1	49.8	41.4
Andrew McCutchen	30	2009-24	49.3	49.7	39.8
Steve Finley	40	1989-2007	44.0	46.6	35.6
Jim O’Rourke	50	1872-1904	52.4	49.9	32.2
Willie McGee	60	1982-99	34.2	37.5	30.2
Jacoby Ellsbury	70	2007-17	31.2	31.3	27.2
Bobby Thomson	80	1946-60	33.8	30.7	25.5
Dom DiMaggio	90	1940-53	33.7	29.2	24.2
Stan Javier	100	1984-2001	25.4	28.1	23.0

Most notable on the list above, aside from the glut of all-time greats (CF has five of the top 20 position players overall), is the increased presence of Negro League stars, with three who played exclusively in the NeL (or almost exclusively, in Willard Brown’s case), plus Doby, who debuted there. Throw in Cristobal Torriente, who ranks #32 despite only having part of his career considered, and you’ve got quite a respectable set of options.

Conveniently, that brings us to the topic of the next article. We’ve spent a great deal of digital ink exploring the adjustments made to WAR for MLB players, but the Negro Leagues present their own set of issues, which require their own versions of the adjustments we’ve explored so far. Next time, we’ll examine what can be done with the stats available for these leagues, through the prism of the position occupied by their greatest player: catcher.