What “late and close” means in baseball

“Late and close” sounds like commentary, but it’s a formally defined statistical split, and the definition has two conditions that must both hold.

First, it must be late: the seventh inning or later.

Second, it must be close: the batting team is leading by one run, tied, or has the potential tying run at least on deck. That last clause is the part people miss. A team down by three with the bases loaded qualifies, because the man on deck represents the tying run. A team down by three with nobody on does not.

Put together: the game is in its final third, and the team at bat can realistically tie it with what’s already in front of them.

The tying-run-on-deck rule is what makes the split sensible rather than arbitrary.

A pure score-based rule — say, “within one run” — would exclude genuinely tense situations. Down 5–2 in the eighth with the bases loaded is a real moment: a grand slam wins it, and every pitch carries weight. The on-deck clause captures that.

It also correctly excludes situations that look close on the scoreboard but aren’t. Down three in the ninth with two out and nobody on is arithmetically a three-run game, but the tying run isn’t even due up. The rule says that isn’t late and close, and the rule is right.

In effect, the clause is a rough proxy for what leverage index measures continuously: how much the situation can still swing.

Late and close exists mainly as a filter on player statistics. A hitter’s late-and-close line shows how he performed specifically in these high-pressure spots; a reliever’s shows how he fared when it actually mattered.

It’s the closest thing the standard stat pages have to a “clutch” filter, and it gets cited constantly in broadcasts and arbitration cases.

It’s also used as a usage measure. Which relievers a manager brings into late-and-close spots reveals the bullpen hierarchy far better than the save column does — a pitcher who repeatedly gets those innings is trusted, whatever his title.

Here is the part that gets skipped when the split is quoted on air.

A full-time hitter accumulates roughly 100 to 150 late-and-close plate appearances in a season — about a sixth of his playing time. At that sample size, batting average is almost pure noise. A .310 late-and-close average and a .210 one can easily describe the same hitter in two different seasons.

Worse, decades of research have found that clutch hitting is not a stable, repeatable skill in any large sense. Players who excel in these spots one year mostly don’t the next. The differences that show up are dominated by randomness.

So the split is genuinely useful for one thing — showing what happened in the moments that mattered — and genuinely unreliable for another: predicting what a player will do next time. Treat a big late-and-close number as a description, not a forecast.

Knowing the definition is most useful if something tells you when a game meets it. That’s exactly what TwentySeven’s smart notifications do: pushes only for moments that genuinely matter — high-leverage spots, late-and-close games, walk-off opportunities, no-hitters in progress. No spam about routine runs in the third.

Pin one to your Lock Screen and the win-probability mini-graph updates on every pitch, so a glance tells you whether the late-and-close game you were promised is still worth watching.