Simple question, but I can’t find the answer:
Are the wins above replacement totals for seasons prior to 1961/2 when they scheduled only 154 games based on extra wins in 154 games or 162 games? I see references to both 162 games and 150 games – I guess the latter being the assumed number of games a regular plays – or probably a nice round number near that,
The counting number in most references is based on the number of games the player’s team actually played.
I don’t think it would matter, would it? The WAR for the current players is based on how well they do against a replacement level player for that season, right?
So, a 154 game season player is tested against the 154 season player, and the 162 vs. the 162 replacement. So, I think it all washes out in the end.
IANASabremetrician.
A player in a shorter season is still at a very tiny disadvantage. The number of W above R you can be is mathematically limited by the number of games there are. If a team of replacement level players would play .275 ball (I think it’s about that) then in a 154 games season the team would go 42-112, and in a 162 game season, 44-118. That means that a team of averages players should have 35 WAR to split among them in a 154-game season, and 38 WAR in a 162-game season. Not a huge difference but it’s there.
To use an extreme example that illustrates how season length affects WAR, suppose they decided next year to only play 16 games, like a football team. Obviously it would not be possible for Mike Trout, who puts up 9 or 10 WAR a year, to have 9 or 10 WAR in a 16-game season.
So to a very small extent, players in the 154-game era were at a tiny disadvantage if you compare them to players in the 162-game era, as a result of season length. Really it is very small but it is there.
Thanks.
And yes I was thinking of comparing players from different seasons. So I’ll assume 7.7 WAR in a 154 game season would be roughly comparable to 8.1 WAR in a 162 game season.
Baseball reference has a lot of gory details on calculation of WAR. The following quote indicates that there is a total WAR of 1000 each season, which is split across the entire league. That applies to all seasons, so comparing across eras should be doable.
I honestly didn’t read the entire page, so I may have missed something.
Seems to me that’s as it should be. A slightly longer season is a slightly greater test of players’ endurance and drive, and a slightly greater risk of injury. A player who successfully maintains a given level of per-game performance over a longer season is statistically better than a player that does equally well, per game, in fewer games. Maybe the shorter-season guy (or, for that matter, the late callup or signing) could have kept it up all year, but he didn’t.
I am not personally a fan of the way BBRef calculates WAR because it isn’t actually directly correlated with, you know, winning.
For instance, here are two teams in the same division in 2015 and their actual records:
Toronto: 93-69
NY Yankees: 87-75
And here are their total team WAR:
Toronto: 50.5
New York: 41.7
So the two teams were in the real world separated by six wins, but according to WAR, they were separated by nine wins. That makes no sense unless they’re in different leagues, but they’re not; I picked them on purpose. WAR is either overvaluing Toronto, or underrating New York. Ultimately, if you’re allocating wins to players, that has to actually match up with baseball games that player’s team actually won.
Look at their Pythag. Toronto ‘should’ have been 14 games better than New York!
If you’re allocating only actual team wins, aren’t you debiting the individual records of guys who scored real runs, for the failures elsewhere of their teammates?
It’s unclear to me why you’d expect a perfect correlation for a summary statistic and dismiss something that is strongly correlated but not perfect.
RickJay, as I recall, is of the mindset that you should give awards and honors for what actually happened, not what “should have” happened. So in other words if you’re going to call it win shares, it should be the player’s actual contributions as reflected in real wins, not contributions to a hypothetical team’s wins, because how could a player add 9 wins if the team didn’t actually win the nine games?
I don’t agree with that, but I remember it having come up before.
But value equates to winning. A baseball team’s purpose is to win, not score runs or have a good Pythagorean projection. The total of a baseball team’s players’ “value” must, logically, add up to the team’s “value,” and the value is wins. If it’s not wins, what the hell could it be?
WAR does not, or should not, penalize a player for having a crappy teammate. Manny Machado last year had about 7 WAR, which is outstanding, even though he played on a mediocre team. Jose Bautista, who played on an excellent team, had fewer WAR. That is as it should be. Machado’s team was worse, but Machado himself was excellent, and made his team better than it would have been without him.
[QUOTE=Snarky_Kong]
It’s unclear to me why you’d expect a perfect correlation for a summary statistic and dismiss something that is strongly correlated but not perfect.
[/QUOTE]
I’m not dismissing WAR. As a wrap-up statistic it is very good. Players who finish the year with 9.0 WAR are always excellent players, and players who finish the year at -1.2 are always bad players. The list of career WAR leaders is a pretty solid list of the greatest players of all time. The correlation between WAR and success on the field is very good. It is not, however, perfect, and I was just pointing that out.
True. But WAR is a metric of individual player performance-value. Not of teams. No single player can win even one whole game by himself; his contribution must be in creating, and preventing, runs, by various means. And apart from HRs and Ks, even that’s mostly a shared endeavor. For the most part, we can only try to properly credit players for their parts in creating and preventing somewhat-abstract fragments of runs.
This is also true, in itself. The total, precise performance-value of a team, relative to its league, is indeed its actual win total. (Well, without getting into differences of schedule, but never mind that now.) That player WARs do not consistently add up to actual team wins above any particular hypothetical “replacement team” performance… just goes to show how much we don’t have a handle on in measuring the factors of baseball. WAR is an imprecise measurement for many reasons; for example, there is no greater credit given to players who create runs against better pitchers (except to the extent that there is an averaged adjustment for parks and leagues that contain the better pitchers).
Brute-forcing a metric of individual value to conform to a statistic of team value will always punish or reward some players for their teammates.