How the scores work

In plain English

Say you open the Betts trade, see Dodgers 99 · Red Sox 1, and ask: how did they get that? Here is the whole method, with no jargon that isn't explained on the spot. The exact weights and the tests the model has to pass are further down.

It all runs on one question: how many wins did each player give the team that got him?

Baseball has a stat for that — WAR, Wins Above Replacement. It rolls everything a player does (hitting, fielding, baserunning, pitching) into one number: "this guy was worth about five more wins than a random Triple-A call-up would have been." A 0 is a scrub, a 2 is a solid starter, a 5 is an All-Star, 8 and up is MVP territory.

Two websites calculate WAR — Baseball-Reference and FanGraphs — and they disagree a little, because they measure defence and pitching differently. So we average them. That is what consensus WAR means on this site: the two versions split down the middle, so nobody can say we picked the one that flatters their guy.

But wins only count while he is on your team.

If the Dodgers trade for a player, he is great for three years, and then he leaves for the Yankees and wins an MVP, the Dodgers get no credit for the MVP year. We walk through his career one season at a time, starting at the trade, and stop the moment he shows up in another uniform. A year lost to injury or the minors doesn't count either way — it doesn't end the streak.

And we look at a few different lengths of time, because "did the trade work?" depends on when you ask.

• Right now — just the season he was traded in.
• Short term — that season and the next. Did he help right away?
• The years you actually bought — the big one. When you trade for a player you don't get him forever: you get him until he becomes a free agent, or until his contract runs out, whichever is longer. We work out how many years the club really owned and count those. If the club signed him to an extension before he had even played a game for them, those years count too — that was clearly part of the plan.
• Everything — his whole time with the club, however long it ran.

Then a check on how good he was when he actually played.

WAR adds up, so an ordinary player who stays six years can pile up more of it than a star who stayed two. To catch that we also look at a rate stat — how good he was per plate appearance or inning rather than in total. For hitters it is wRC+ (100 is an average hitter, 150 is elite, 70 is bad); for pitchers it is FIP− (100 is average, 70 is elite, 130 is bad — lower is better, so we flip it round to match). If he barely played, this part is skipped rather than guessed.

Then money.

Not strictly on-field, but you will see it in the score. For trades since 1985 we know roughly what a win cost on the free-agent market that year, so we can say "he produced $60M worth of wins and was paid $20M — that's $40M of surplus." A superstar on a huge contract can produce a pile of WAR and still come out even here, because the club paid full price for it.

So how does that become a number like 87?

We don't just add things up. We take every player who has ever been traded — about 17,000 of them — and line them all up. A player's score is where he lands in that line. An 87 means he did better for his new club than 87% of everyone who has ever been traded. A 50 is dead average. That is also why nobody scores exactly 100: no one is at the very front of the line.

For a player, four of those rankings are mixed together: the years-you-bought WAR (40%), the short-term WAR (20%), the rate stat (15%) and the money (25%).

For a club, it is the same idea but net: what they got minus what they gave up — and "what they gave up" means what those players went on to do for the other club. The mix is money (65%), the years-you-bought WAR (15%), the whole-tenure WAR (15%) and the short-term WAR (5%). Money is heavy there because it is the only piece that already accounts for both the talent and what was paid for it. Before 1985 there is no money data, so those trades use the WAR parts only.

One more thing: recent trades are marked provisional.

After one year you know about 40% of how a trade will turn out; after three, about 80%. Below that line the page tells you the number is still cooking.

So what stats did they actually look at?

The honest answer: WAR, sliced four ways, plus one rate stat and a price tag. Not batting average, not home runs, not saves, not Gold Gloves, not how clutch he was, not how many games he missed. WAR is meant to contain most of that, but it is one lens, and every part of the score is looking through it.