The best hitters in baseball fail roughly 65% of the time. A .350 average, one of the best marks in the sport, still means the hitter made an out on almost two thirds of his at-bats. That's not a flaw. That's the whole game.
Why do even great hitters fail so often?
Hitting a baseball is one of the hardest physical acts in sports, and the pitcher only has to win the physics for a fraction of a second. A big-league fastball reaches the plate in under half a second, so hitters have to decide whether to swing, guess where the ball will be, and commit before it's even halfway there. Ted Williams was the last man to hit .400 in a season, batting .406 for the Boston Red Sox in 1941. Even in that legendary year, he made an out nearly six times out of ten.
Compare that to basketball. Steph Curry, one of the best three-point shooters ever, converts around 42% of his attempts for his career, per Basketball-Reference. Baseball's margins are just tighter. A .300 hitter is very good. A .400 hitter is basically mythical. The gap between great and failing most of the time is razor thin, and that gap is where variance lives.
What is variance, and why does baseball have so much of it?
Variance is the swing between a player's true talent and what actually shows up in the box score on a given day, week, or month. Baseball has more of it than almost any other major sport because the sample size per game is so small. A hitter gets maybe four at-bats a night. A basketball player touches the ball dozens of times. A quarterback throws 30-plus passes. Baseball forces you to judge a guy on a handful of swings, which is exactly why one bad week doesn't mean a hitter lost his stroke, and one hot week doesn't mean he found some secret formula.
Take BABIP, batting average on balls in play. Sounds nerdy, but it's simple: of the balls a hitter puts in play (not counting strikeouts or homers), what percentage fall for hits? League average runs around .290 to .300 most seasons, according to FanGraphs. A hitter can smoke the ball right at a shortstop and get nothing. He can blooper one over second base and get a hit. Same contact quality, wildly different outcome. That's variance doing its job, and it's a big reason predicting single games is so hard even when you know the players well.
Does variance mean stats don't matter?
No. It means stats matter more over a longer window, not less. A single at-bat tells you almost nothing. A full season tells you a lot. That's why scouts and analysts lean on metrics like exit velocity and launch angle from Baseball Savant instead of just watching one game's box score. A hitter can go 0-for-4 with four scorched line drives that all found a glove. His underlying contact quality says he's still hitting the cover off the ball. The results just haven't caught up.
Same reason a .500 hitter over a 10-game stretch isn't actually a .500 hitter for the season. Ten games is about 35 to 40 at-bats, nowhere near enough to separate a hot streak from true talent. Over 600 at-bats, though, the noise cancels out and the real number shows up. Variance shrinks as the sample grows. It never disappears, but it stops running the show.
How should you think about variance when making predictions?
Treat any single game as a coin flip layered on top of a player's real talent, not a referendum on how good he is. A star hitting into bad luck for two weeks is still a star. A journeyman running hot for two weeks is usually still a journeyman. The players who separate themselves in season-long tracking understand this and don't overreact to short stretches, good or bad. That mindset is basically the whole game plan behind getting sharper at making sports predictions over time.
It also explains why hitting streaks and slumps get so much attention. A guy who's 2-for-25 gets written up like he's cooked. A guy on an 11-game hit streak gets treated like he's unstoppable. Both stories are usually more about variance clustering than anything mechanical changing in the swing. Joe DiMaggio's 56-game hit streak in 1941 is still untouched more than 80 years later because stringing together that much good luck on top of talent almost never happens, as detailed by Baseball-Reference.
What does this mean for how you watch and predict the game?
The fun isn't guessing whether a great hitter gets a hit tonight. It's understanding the odds well enough to know when a prediction is actually smart versus just a guess. Implied probability from the lines gives you a baseline for how likely an outcome really is. Your job as a predictor is figuring out where you know something the market doesn't, whether that's a hot matchup, a pitcher's tendencies, or a park factor that favors contact hitters. Variance means you'll be wrong plenty. That's baked into the sport at every level, from Ted Williams down to today's lineup.
If you want to test your own read on these matchups against real lines and real outcomes, Download GAGE and start putting your predictions up against the odds.
Is a .300 batting average actually good?
Yes. A .300 average is a strong season for a modern hitter, since league averages typically sit closer to .240 to .250 across all of MLB in recent years.
Why do batting averages vary so much year to year for the same player?
Small sample sizes and BABIP swings are the main drivers. A full season of at-bats still isn't enough to fully cancel out random variation in where batted balls land.
Does variance affect pitchers the same way it affects hitters?
Yes. Pitchers deal with it too, especially through their own BABIP against and strand rate, which is why ERA can swing wildly between seasons even when a pitcher's underlying stuff hasn't changed much.