Sports

Baseball stats explained: what WAR and OPS mean

By Leandro Bruzaferro · · 5 min read
AI-generated illustration of a vintage baseball stadium scoreboard displaying batting statistics columns in warm afternoon light
SharePostLinkedInFacebook

Baseball coverage now assumes a vocabulary that did not exist a generation ago. A player is described as a five-win contributor, or as having an on-base plus slugging figure that means nothing to most people watching. The statistics are not as difficult as they sound. What makes them confusing is that nobody explains what problem each one was invented to solve.

Why batting average stopped being enough

Batting average measures how often a batter gets a hit per official at-bat. It ruled the sport for a century and has two flaws that eventually became impossible to ignore.

It treats every hit identically. A single and a home run count the same, which discards the difference between advancing one base and scoring immediately.

And it ignores walks entirely. A player who reaches base by drawing a walk has done something valuable, avoiding an out and occupying a base, yet the plate appearance simply vanishes from the calculation. Two players with identical averages can have wildly different value if one walks frequently and the other never does.

Everything that followed is an attempt to fix one or both of those problems.

OPS and what it captures

OPS is on-base percentage added to slugging percentage, and each half addresses one of the flaws above.

On-base percentage asks how often a batter avoids making an out, counting walks and hit-by-pitches alongside hits. It answers the question that actually matters to an offence, since a team’s innings ends when it accumulates outs.

Slugging percentage weights hits by the number of bases gained, so a double counts twice a single and a home run four times. It measures power.

Adding them is mathematically crude, and statisticians will tell you so, because the two components are not on the same scale and on-base percentage is worth more to run scoring than slugging is. Refined versions exist that weight them properly.

OPS survives anyway because it is simple, requires no proprietary data, and correlates well with runs scored. It answers “did this player reach base, and did he hit for power” in one number, which for most conversations is enough.

The main caveat is that OPS ignores context: park, era and league scoring environment all move it. This is why adjusted versions, scaled so that league average equals 100, are more useful for comparing players across different seasons or ballparks.

WAR as a single-number attempt

Wins Above Replacement is more ambitious. It tries to express a player’s total contribution, offence, defence, baserunning and position, as the number of additional wins he provided compared with a freely available substitute.

The “replacement” concept is the clever part and the part most often misunderstood. The baseline is not an average player. It is the level of performance a team could obtain at no cost, from the minor leagues or the waiver wire. Comparing to average would mean roughly half of all players had negative value, which is not a useful way to think about a roster.

Building WAR requires converting everything to a common currency: runs. Offensive contribution is measured in runs above what a replacement would produce. Defensive contribution is measured in runs saved or allowed relative to positional expectation. Baserunning adds or subtracts runs. A positional adjustment accounts for the fact that shortstop is harder to fill than first base. The total is then converted from runs into wins.

Pitching WAR follows the same logic on a different route, expressing run prevention relative to replacement, adjusted for park and for the strength of opposition faced.

Why fWAR and bWAR disagree

Two organisations publish WAR, and their figures for the same player differ, sometimes by a full win. Neither is wrong. They made different defensible choices at two points.

Decision point FanGraphs (fWAR) Baseball Reference (bWAR)
Pitcher evaluation Based on outcomes a pitcher controls directly, chiefly strikeouts, walks and home runs Based on runs actually allowed, adjusted for context
Underlying question What should have happened, given how he pitched What did happen while he pitched
Effect on defence-dependent pitchers Credit goes to the fielders Credit stays with the pitcher
Defensive metric for fielders One family of fielding models A different family of fielding models
Where the gap is largest Pitchers with unusual outcomes on balls in play The same pitchers, in the opposite direction

The pitching row is the main source of divergence. A pitcher who allowed few runs while striking out very few batters will rate well by runs allowed and poorly by controllable outcomes, because one system credits him and the other credits the defence behind him.

The fielding row explains most of the rest. Defensive measurement remains the hardest problem in the sport, the available models disagree with each other, and any WAR figure inherits that disagreement.

Reading a stat line critically

Three habits make these numbers useful rather than decorative.

Treat WAR as a range, not a point. A player at five wins in one system and four in another is somewhere in that band. Arguing about the decimal misunderstands the precision available.

Check the sample. Defensive metrics in particular need a great deal of data before they mean much, so partial-season WAR is far less reliable than the confident single number suggests.

Match the statistic to the question. For how well a batter performed at the plate, OPS or an adjusted version is direct and honest. For total value including defence and position, WAR is the right tool despite its uncertainty. Using WAR to argue about hitting alone imports a large amount of noise for no benefit.

None of this is a reason to distrust the statistics. It is a reason to use them the way their creators intended: as good estimates of things that were previously guessed at entirely.

Sources