What Is WAR in Baseball? Wins Above Replacement Explained

WAR, or Wins Above Replacement, estimates how many wins a player added to his team compared with a replacement-level player, the kind of fringe big leaguer or minor league call-up a team can get for almost nothing. A player with 5.0 WAR in a season was worth about five more wins than that freely available fill-in would have been in the same playing time.

That one number tries to fold hitting, baserunning, fielding, position and pitching into a single scale, which is why it turns up in MVP debates, contract talks and Hall of Fame arguments. It is also an estimate rather than a box score count, and the two most quoted versions, from Baseball-Reference and FanGraphs, regularly disagree. This guide explains where the number comes from, why the versions differ, what counts as a good season, and where WAR falls short.

What WAR measures

WAR works in two steps. First it estimates how many runs a player was worth: through hitting, running, fielding and the position he played, or through run prevention if he pitches. Then it converts those runs into wins. Baseball-Reference leans on the long-standing rule of thumb that about 10 runs equal one win, and FanGraphs calculates a runs-per-win figure each season that usually lands between 9 and 10.

A few things follow from that design:

• WAR is a counting stat, like home runs, not a rate stat like batting average. A player who misses half the season usually finishes with less WAR, even if he played brilliantly while healthy.
• Position players and pitchers sit on the same scale, so a 4-WAR shortstop and a 4-WAR starting pitcher are rated as roughly equal contributors.
• Both main versions adjust for ballpark and league scoring levels, so a hitter in a thin-air park does not get free credit.
• Season totals add up to career WAR, which is why the stat is so popular in Hall of Fame debates.

The MLB.com glossary describes WAR in the same spirit: how many more wins a player is worth than a replacement-level player at his position.

Replacement level: the zero point

The R in WAR is the part newcomers find strangest. WAR does not compare a player with the league average. It compares him with a replacement-level player: the best option a team could find without spending real resources, such as a Triple-A veteran, a waiver claim or a bench player on a minimum salary. FanGraphs describes this player as a freely available minor leaguer or a AAAA player, baseball slang for someone too good for Triple-A but not quite good enough to stick in the majors.

Baseball-Reference sets replacement level at a .294 winning percentage, which works out to a record of about 48-114 over a 162-game season. A roster made entirely of replacement players would still win some games, just not many.

That baseline is shared. In March 2013, Baseball-Reference and FanGraphs agreed on a unified replacement level of 1,000 WAR for every 2,430 major league games, which is one full season of 30 teams playing 162 games each. FanGraphs then splits the pool with 57 percent going to position players and 43 percent to pitchers, or 570 and 430 WAR.

The arithmetic holds together. Thirty replacement teams winning about 47.7 games each would collect about 30 x 47.7 = 1,431 wins. Every game produces exactly one win, so a season has 2,430 to hand out, and 2,430 - 1,431 leaves roughly 1,000 wins for real players to earn above replacement. If you want the background on why the season is 162 games long, see our guide to how many games are in an MLB season.

How WAR is built for position players

FanGraphs publishes its position player formula in plain terms:

WAR = (batting runs + baserunning runs + fielding runs + positional adjustment + league adjustment + replacement runs) / runs per win

Baseball-Reference uses the same structure with its own inputs, plus a separate small line for grounding into more or fewer double plays than expected. Each piece is measured in runs:

• Batting runs: offensive value relative to an average hitter, adjusted for park and league
• Baserunning runs: stolen bases, caught stealing, taking the extra base and similar events
• Fielding runs: defensive value relative to an average fielder at the same position
• Positional adjustment: credit or debit for how hard the position is to play
• League adjustment: a small correction so the American League and National League pieces balance
• Replacement runs: the gap between an average player and a replacement player, scaled to playing time

The first five pieces tell you how far above or below average a player was. The replacement runs then move the zero point down from average to replacement level.

The positional adjustment

A shortstop and a first baseman who hit exactly the same are not equally valuable, because far fewer people can play shortstop well. WAR handles that with a fixed adjustment for each position. These are the FanGraphs values, in runs per 162 defensive games:

• Catcher: +12.5
• Shortstop: +7.5
• Second base, third base and center field: +2.5 each
• Left field and right field: -7.5 each
• First base: -12.5
• Designated hitter: -17.5

A player receives a share of the figure based on how much he actually played there, so someone who plays half a season at shortstop earns about 7.5 / 2 = +3.75 runs from it. Baseball-Reference uses its own positional values built on the same logic.

A worked example: same bat, different gloves

Here is a simplified full season for a hypothetical player, using round numbers and 10 runs per win. Real calculations carry more decimals and adjustments, but the structure is the same.

Our made-up hitter plays every day at shortstop. He is worth +15 batting runs, +2 baserunning runs and +5 fielding runs. His positional adjustment is +7.5, we will set the league adjustment at 0 for simplicity, and we will give him 20 replacement runs, an illustrative round figure for a full-time player.

Add it up: 15 + 2 + 5 + 7.5 + 0 + 20 = 49.5 runs above replacement. Divide by 10 runs per win: 49.5 / 10 = 4.95, or about 5.0 WAR.

Now move the same bat to first base, where he is also 5 runs better than an average first baseman. The positional adjustment flips to -12.5: 15 + 2 + 5 - 12.5 + 0 + 20 = 29.5 runs, or about 3.0 WAR. Make him a full-time designated hitter, with no fielding runs at all: 15 + 2 + 0 - 17.5 + 0 + 20 = 19.5 runs, or about 2.0 WAR.

Same hitter, three very different values. For a sanity check, take a perfectly average hitter, runner and fielder who plays every day at second base: 0 + 0 + 0 + 2.5 + 0 + 20 = 22.5 runs, or 2.25 WAR. That sits right next to the roughly 2.0 WAR that Baseball-Reference describes for an average regular.

How WAR works for pitchers

Pitchers are where the two main versions split most clearly, because they start from different raw numbers.

Baseball-Reference begins with RA9, runs allowed per nine innings. It works like ERA but counts every run, earned or not; our guide to ERA in baseball explains that difference and walks through RA9 and FIP with examples. Baseball-Reference then adjusts for the quality of the defense behind the pitcher, the opponents he faced, the parks he pitched in and whether he started or relieved. The philosophy: start from what actually happened on the scoreboard, then strip out the help or harm from the fielders.

FanGraphs begins with FIP, fielding independent pitching, which uses only strikeouts, walks, hit batters and home runs. Its WAR version treats infield pop-ups as strikeouts, since they are nearly automatic outs. FanGraphs then adjusts for park and league, uses different replacement levels for starters and relievers, and gives relievers extra credit for working high-leverage spots, the late and close situations where a run matters most. The philosophy: grade the pitcher on the outcomes he controls most directly and set aside what happened once the ball was in play.

FanGraphs also publishes RA9-WAR, a runs-allowed version, for readers who want that view on the same site.

Why one season can get two grades

Take a hypothetical starter who throws 180 innings and allows 90 runs. His RA9 is 90 x 9 / 180 = 4.50. His strikeouts, walks and home runs, though, produce a 3.50 FIP, because plenty of ground balls found holes behind a weak defense.

That is a gap of one run per nine innings. Over 180 innings there are 180 / 9 = 20 nine-inning units, so the gap is 1 x 20 = 20 runs, or about 2 wins at 10 runs per win. Baseball-Reference hands some of that back through its defense adjustment, but it starts from 4.50. FanGraphs starts from 3.50. Expect the FanGraphs number to be noticeably higher for this pitcher, and the reverse for a pitcher whose runs allowed beat his FIP.

bWAR vs fWAR: why the numbers differ

You will see Baseball-Reference's figure written as bWAR or rWAR, and FanGraphs' as fWAR; MLB.com's glossary uses the same labels. Both share the 2013 replacement level and the same broad framework, so they usually land in the same neighborhood. The inputs are where they part ways:

• Pitching: runs allowed at Baseball-Reference versus FIP at FanGraphs, as shown above. This produces the biggest gaps.
• Fielding: each site uses its own defensive data. FanGraphs replaced the range part of UZR with Statcast's Outs Above Average based fielding runs in 2022, then moved its remaining fielding and baserunning pieces to Statcast metrics in its 2024 update, both retroactive to the 2016 season.
• Catcher framing: the skill of getting borderline pitches called strikes. FanGraphs counts it in WAR, while coverage from MLB.com and FanGraphs during the 2025 season noted that Baseball-Reference's WAR does not.
• Revisions: both sites update their methods and historical data, so a player's WAR from years ago can change after the fact.

Baseball Prospectus publishes a third system, WARP, with its own inputs again. The practical rule is simple: compare players within one version. Setting one player's bWAR next to another player's fWAR is measuring with two different rulers.

What counts as a good WAR season

There is no official scale. Two widely used rules of thumb, both for a full season, will get you most of the way.

Baseball-Reference's leaderboard notes use a scale developed by Sean Smith of BaseballProjection.com:

• 8 or more: MVP quality
• 5 or more: All-Star quality
• 2 or more: starter
• 0 to 2: reserve
• Below 0: replacement level

The FanGraphs library uses finer labels for position players and starting pitchers:

• 0 to 1: scrub
• 1 to 2: role player
• 2 to 3: solid starter
• 3 to 4: good player
• 4 to 5: All-Star
• 5 to 6: superstar
• 6 or more: MVP

The two scales differ most at the top, where Baseball-Reference's chart reserves MVP quality for 8 or more and FanGraphs starts it at 6. They agree on the middle: about 2 WAR is an average full-time regular. By either chart, our hypothetical 5.0 WAR shortstop had an excellent season, while the same bat as a 2.0 WAR designated hitter was ordinary.

Where WAR helps and where it falls short

The best way to use WAR is as a summary, not a verdict. If two players are 0.3 WAR apart, treat it as a tie and look at the pieces underneath. If they are 3 WAR apart, the number is telling you something real. And when a player's WAR seems surprising, check which component is driving it. A huge fielding number in one season, or a pitcher with a big gap between his ERA and FIP, is usually where the two versions disagree.

For the hitting side of the ledger, the batting runs inside WAR follow the same idea as the league-adjusted stats in our guide to OPS in baseball, where OPS+ and wRC+ use a scale on which 100 is average.

WAR quick reference

Keep this list handy the next time a broadcast graphic or an MVP argument quotes WAR:

• WAR means Wins Above Replacement: wins added over a freely available fill-in
• Replacement level: a .294 winning percentage, about 48-114, shared by Baseball-Reference and FanGraphs since 2013
• About 10 runs make one win
• Position players: batting + baserunning + fielding + position + league + replacement runs, divided by runs per win
• Pitchers: bWAR starts from runs allowed (RA9), fWAR starts from FIP
• About 2 WAR is an average regular, 5 is All-Star level, and 6 to 8 or more is MVP territory depending on the scale
• Compare players within one version, and treat small gaps as ties