I find it very frustrating that any stat in baseball can be different just because the sites create it themselves (and decide HOW to compile it) rather than an entity like Elias doing it and giving it to them. Complicating this is that I don't want to just ignore it and use other stats since WAR is really really convenient and simpler than using multiple stats to determine overall value. Since it's unfair to go back and forth between them, I want to determine the best way to use WAR without going back and forth. I had heard there were others than fWAR and bWAR but just learned about WARP which is baseballprospectus' version. From what I have learned, I am leaning toward bWAR: 1) position players it uses DRS which is calculated within the current season, while fWAR uses UZR which is based on 3 years of data. This makes the data for players in yrs 1 and 2 much more varied with larger differences between fWAR and bWAR than veterans. 2) For pitchers, bWAR treats all outs and results the same while fWAR rewards strikeouts more than outs made by fielders. I'm certainly no expert so I welcome input from you. Maybe we can combine and average some combination of them and call it cWAR or Clutch WAR. Please post your ideas, opinions, and facts you can add toward explaining this to (or correcting) me.
For hitters I trend towards fWAR because I think BBR overvalues defense quite a bit. I do not give a single solitary f**k about pitching WAR. Hitters need a stat to quantify various factors aside from hitting. I can pretty well value how good a pitcher has been without some esoteric calculation.
I have always used fWAR just for consistency and because I started using it first. I like bWAR better for pitchers but I agree that choosing one to use for all of us would make it easier to discuss players. Unifying this stat is something that may be possible if the current publishing orgs could be brought together for that purpose. I do not like the idea of coming up with our own WAR stat as it would just further muddy up the water.
Jeter passed the eye test because he made the mundane plays look like spectacular plays. We have to get away from eye tests for determining value outside of audience reaction. Make WAR better. Don't obsess over it's faults.
Andrew averages them. That seems reasonable to me. I use BRef because I like their site more than fangraphs, which I can find bewildering at times. agreed on pitcher War being unnecessary.
fWAR uses OAA now instead of UZR for most of the defense part. rWAR (aka b WAR) uses quality of opponent in their calculations. I prefer fWAR for hitters. I prefer fWAR for pitchers for short term data with exceptions as I think xERA is needed to see if pitchers are doing something to minimize quality of contact (basically, I look at xERA, xFIP, and ERA and pick the value that fits 2 out of 3). Long term (think Hall of Fame, greater than 5 years of sample data), I prefer rWAR for pitchers...though really only need rWAR for Hall of Fame discussions, IMO. The quality of opponent part of rWAR creates too much noise IMO to be used in year or less than year data. Performance against Jose Abreu last season was absolutely horrid or great depending on whether he was playing hurt or not. Players ability fluccuates too much to useful for quality of opponent for samples of a season or less. Over a few years, the rWAR system is better for pitchers as the noise in their protocol goes away (I just don't it is useful for how a good a player is right now).
I don't really spend much time on baseballreference. Beyond FG and Statcast, DRA and DRC from Baseball Prospectus are good stats to monitor. Texas Leaguers and BrooksBaseball also have some good info on pitchers.
I don't have a sub to baseball prospectus and don't see the point in paying when there is so much free information available. So that leaves some combo of bWAR and/or fWAR. It's never going to be perfect but I think I will only use WAR for hitters. I will need to figure something else out for pitchers but WAR doesn't make sense. Both FG and BR use 1000 as the total WAR in any given year. Based on that and a winning percentage of .294 (48-114) being designated as a complete 0 replacement season, they then can determine how the various metrics impact the scores to determine individuals WAR. Because of the diffrrences in metrics used, BR has 590 WAR going to position players, while FG has 570. That means BR position player WAR will be 3.5% higher than FG regardless of any other factor. I wish they were the same but I guess if you start by having too many similarities you end up with the same WAR, LOL. So now I just need to decide if I like 1 better than the other, to average them, or to average them after adjusting for the 3.5%.
In the end I still depend on my own intuitive thinking as a final arbiter in players past and future values after all the math is done. Eyes wide open.
