Run differential in baseball

Run differential is exactly what it sounds like: runs scored minus runs allowed. A team that has scored 620 and allowed 550 has a run differential of +70. One that has scored 500 and allowed 590 sits at −90.

You can compute it for a single game, a week, a month, or a season. For one game it’s just the margin of victory. Across a season it becomes something more useful: a measure of how thoroughly a team has outplayed its opponents, independent of how those runs happened to be distributed.

It is the rawest possible summary of team quality, and it takes no math beyond subtraction.

Here is the counterintuitive part: a team’s run differential predicts its future record better than its current record does.

The reason is that wins are lumpy. Winning 3–2 counts exactly as much as winning 12–1, so a record throws away the information about how much a team won by. Run differential keeps it. And since the ability to outscore opponents persists while the ability to win exactly the close ones mostly doesn’t, the number that keeps the margins carries more signal.

A useful rule of thumb: roughly ten runs of differential is worth about one win over a season. A team at +100 is a true talent 10-games-above-.500 club, whatever its record currently says.

This is the same insight Pythagorean expectation formalizes — Pythagorean just converts the differential into an expected winning percentage instead of a rough win count.

A raw differential needs a little framing before it means much.

• Games played matters. +40 in April is far more impressive than +40 in September. Per-game differential — total differential divided by games — makes stretches comparable.
• A few blowouts can distort it. One 18–1 win inflates the number without saying much about the team's typical night. Check whether a big differential comes from consistent margins or a handful of routs.
• Run environment matters. In a high-scoring era, a given differential is worth slightly less, because runs are cheaper. This is exactly what Pythagenpat’s variable exponent corrects for.
• Compare to the division. A +50 team in a strong division is in a different position than a +50 team feasting on weak opponents — which is where strength of schedule comes in.

The most interesting teams are the ones where these two numbers point in opposite directions.

A team above .500 with a negative differential has been winning the close ones and losing badly the rest of the time. It is a regression candidate — the record flatters it.

A team below .500 with a positive differential has been losing the tight games while winning comfortably elsewhere. Historically this profile improves, because losing one-run games is not a durable trait.

Neither pattern is a guarantee. Bullpen quality genuinely bends close-game outcomes, and a team built around a dominant back end can sustain a gap. But when a contender's record and differential disagree by a wide margin, the differential is usually the better guide to what happens next.

TwentySeven shows run differential where it’s actually useful — inside each game as the margin moves, and across the season alongside standings, power rankings, and Pythagorean over- and under-performers.

Because the app is built around win probability, you get both halves of the picture: the differential telling you how good a team has been, and the win-probability curve telling you how much each individual game hung in the balance.