WAR (Wins Above Replacement Player)

 For the DEFINITIVE WAR explanation, go to this Baseball-Reference page, or you can slog through mine.

Over the course of time, I'll mention Wins Above Replacement Player (WAR) often enough that I should dedicate an easy-to-reference page to understand not only what it is, but the version that I use, since different people have different equations. I will state at the outset that I have NO IDEA how WAR is calculated, and aren't particularly interested in finding out. Like any other statistic, WAR allows for comparisons between players, and the number of itself is irrelevant. The fact that a player has a career WAR value of 60 is irrelevant--the fact that only about 100 players (give or take) have achieved that level in baseball history is VERY IMPORTANT.

There are as many versions of WAR calculations as there are web sites devoted to baseball analytics, and in the end, the differences between them will be relatively minor. All work off the same basic premise, which is:

How many more runs than the average replacement level player did the player account for in his season/career

There is a pretty direct relationship between wins and runs (the more runs, the more wins--didn't see that coming, did you?), and so if what a player does adds to runs scored OR fewer runs allowed, that goes in the positive column, and if the player decreases runs scored or allows more runs to be scored, that goes in the negative column. In the end, all WAR does is put some context into what we can see with our own eyes--no WAR calculation (or any statistic, for that matter) exists to say that Adam Dunn's 2011 season was a good one (it was historically awful, probably one of the five worst seasons in baseball history, but that's a story for another day), and no reasonable statistic states that Albert Pujols has been a bum his entire career. All statistics like WAR and others do is allow some rankings and give some meat to what we already suspect.

The WAR statistics I use are from baseball-reference.com. They differ from FanGraphs (and likely other sites), but I'll discuss that further in the post. Baseball-reference WAR  number was developed by Sean Smith at BaseballProjection.com and consists of seven parts:
1. Rbat (runs batting)--Number of runs better or worse than average the player was as a hitter. This is calculated from an AVERAGE player vs. a replacement-level player, which is an important distinction--the average player is, well, average, whereas the replacement-level player is the quintessential 4A player--just good enough to grab the 24th or 25th spot on the roster.
2. Rbaser (runs from base running)--Number of runs better or worse than average the player was for all base running events. This isn't just stolen bases, but advancing on hits, reaching on errors and other base running events.
3. Rroe (runs reached on error)--Number of runs better or worse than average the player was at creating reached on errors. If you think about it, this is a marker of base running skill and speed, since slower runners will not reach on errors as often because they allow the fielder to correct for any mistake. I have no idea if this includes base runners reaching on throwing errors (i.e., the ball is thrown in the stands--should the hitter receive credit for that?).
4. Rdp (runs from grounding into a double play)--Number of runs better or worse than average the player was at avoiding grounding into double plays. I'll be the first to state I have no real idea what this means other than it likewise reflects the skill and speed to avoid double plays, both as a batter and as a baserunner, but I could be wrong.
6. Rpos (runs from position)--Number of runs above or below average due to positional differences. Positions like C, SS, and 2B get a bonus.Positions like 1B, DH, LF get a penalty. This is fairly important and will become easier to understand with an example.
7. Rrep (runs over replacement player)--the number of runs this player is better than a replacement player. Replacement is set for a .320 team winning percentage, which in a 162-game season is a 52-110 season, which is pretty darn bad.

The next two  charts show Hank Aaron's career, numbers and then his WAR values:

These are Aaron's values for the seven components of WAR:

Even though there are seven components, it's pretty apparent that only four really amount to anything--batting, fielding, position and replacement player. Granted, Aaron was a tremendous hitter, but he wasn't a bad fielder at all, as the numbers attest--his numbers are generally positive, implying that he fielded his position (primarily right field) well. The negative numbers shown in the Rpos column reflect that he was a corner outfielder, something that this WAR calculation doesn't value as highly as a middle infielder, center fielder or catcher. When comparing Aaron to other right fielders, this won't matter, since they will all have similar negative numbers. In a perverse way, Aaron's 23-year career will actually penalize him to a certain extent since he will have accumulated negative numbers for more seasons than a less-talented player with a shorter career, but that is generally outweighed by his excellent hitting until the very end of his career.

To understand the yearly WAR values, any season between 2.0-4.9 is a very solid year, 5.0-7.9 is all-star caliber, and 8.0+ is a year of historic proportions. Aaron's career total of 141.6 ranks 6th all-time, which is what is fully expected--Aaron had a long, successful career where he put up productive numbers. WAR simply confirms that when we look back at Aaron's career, it truly was as excellent as we thought.

Without getting too weighed down, WAR takes into account things like runs scored, park effect and other things that can complicate comparisons between eras. For example, the Colorado Rockies played in a high-run era in a high-run park and usually get deflated compared to what a cursory review of their numbers would suggest. Conversely, the Dodgers of the mid-60's played in a pitcher's era in a pitcher's park (but luckily had the luxury to not face Dodger pitching), and WAR calculations take those conditions into account and give a more accurate portrayal of what at first glance might be considered average offensive numbers.

To contrast a bit, the following are Chase Utley's career numbers, followed by his WAR values:

Chase Utley has had a solid career, is an excellent hitter, but here we can see how WAR looks at players. Since Utley is a second basemen, he gets a bump in his position component (if he was a shortstop or catcher, it would be higher). Utley fields his position well, but the boost given to middle infielders and catchers is what makes their WAR values comparable to offensive juggernauts like Albert Pujols and Prince Fielder. If the fielding position adjustment is removed, the differences become greater.

In the end, it's just another statistic, but it's very helpful when trying to make comparisons across eras, especially for players that we never saw play. Since all these numbers are era-dependent, the better fielders of today don't dwarf the fielders of the 1900s, because e 1900 fielders are compared to fielders of that era and the current players against current players. No one in their right mind should say "Hanley Ramirez ain't no Mickey Mantle"--they play different positions and had completely different responsibilities, but we can at least see how they compare using WAR and a couple other tricks. It works best within positions (was Honus Wagner really a good shortstop? Yes, the best of all-time, by a fairly healthy margin).

WAR also tells us a couple of things we might not agree with (and might be invalidated with new tools in the near future):
1. Defense is NOT as imporant as people might think. A player's defense is nowhere near as valuable as his offense, and for those players who were gifted defensively (Bill Mazeroski, Mark Belanger, etc.), they most certainly DID NOT make up for their batting with their defense. At best, they offset it, at worst, their offensive shortcomings outweighed their defense. Reviewing the Aaron and Utley value numbers shows the offense/defense ratio as about 4-6:1.
2. To help envision a replacement-level team, consider this lineup of players, made up of relatively contemporary players who managed to have a career with a cumulative WAR around 0, which is quite tricky to manage when you think about it:
C-Paul Bako (career WAR of -2.5)                           SP-Andy Hawkins (2.2)
1B-Lee Stevens (.9)                                                   SP-Jimmy Haynes (1.1)
2B-Tony Womack (1.2)                                             SP-Mike LaCoss (.5)
SS-Alex Cintron (-.4)                                                 SP-Jason Bere (.3)
3B-Geoff Blum (.1)                                                    SP-Blue Moon Odom (-.2)
LF-Craig Monroe (2.9)                                              CL-Jorge Julio (1.4)
RF-Roger Cedeno (.2)
DH-Johnny Gomes (2.1)

This is not to say that some of these players didn't have decent seasons. I haven't explained how WAR works for pitchers, but I'll save that for another post. In the end, it's just another number that allows us to rank and evaluate players.