The London Spitfire and Philadelphia Fusion will square off this weekend to determine the champion of the first season of the Overwatch League. Each team managed an identical 24-16 record in the regular season, but are they really the same? To (somewhat) answer that, we turn to the teams’ Pythagorean win-loss records.

Pythagorean win-loss record is a fairly simple concept in sports. Its core notion is that “points” scored and “points” allowed are the primary determining factors in how good a team is and what its win-loss record should be. “Points” can represent whatever a league uses to keep track of score, whether it be points, runs, goals, or, in the Overwatch League, maps.

Take the example of a baseball team that wins its first game 10-0 and then loses its next four games by a score of 5-4 each. That team is 1-4, but is it really a .200 team? Well, yes, it is, because individual wins and losses matter. But that team has scored 26 runs and given up just 20, so you would expect its record to be different. It just got “unlucky” by having all its runs bunched together in one game, rather than spread out over all five games.

That’s a major simplification, of course, but the notion that teams are especially good (or bad) in close contests has generally not held up to scrutiny. In other words, our fictional team isn’t 1-4 because it couldn’t come through in the clutch and win some of those 5-4 games. That makes for a nice narrative in the papers, but winning close games isn’t a demonstrably repeatable skill.

The formula for determining a team’s Pythagorean W-L percentage is fairly simple. It’s just:

(PF ^ 2) / (PF ^ 2 + PA ^ 2)

Where PF = Points For and PA = Points Allowed. For our fictional baseball team, it would be:

(26 ^ 2) / (26 ^ 2 + 20 ^ 2) = .628

.628 * 5 games played = 3.1 wins. So, we’d expect our team to have a record closer to 3-2 than 1-4.

Now, the step you’ve been waiting for: applying Pythagorean W-L record to the Overwatch League standings, using maps won and maps lost in place of points. That gives us the following:

As you can see, London has a Pythag of 27-13, better than its true 24-16 record, while Philadelphia has a Pythag of 23-17, which is worse. The W-PW column is just that: wins minus Pythagorean wins. A team with a negative score in that column was less “unlucky” and had a real record worse than its Pythag, while a team with a positive number was “lucky” and had a real record that was better than its Pythag.

Note that London had the biggest negative difference, finishing three wins lower than its Pythag W-L, while the Florida Mayhem, bad as they were, had the largest positive factor and “should” have been even worse, with just a 4-36 record. The Boston Uprising was the only team that had the “right” record, according to Pythagorean win-loss.

Does that mean London is the better team? Based on full-season records, I’d say yes, although that’s largely on the strength of the huge map differential London racked up in the first half of the season (+35) versus the second half (-2). Philadelphia was much more consistent between halves (+8 vs. +5). If I use SAMP to try and predict things, London has a 62% chance of winning each game, which makes for a 67.6% overall chance of winning two out of three.

As we’ve seen, however, both teams have bucked the odds in their playoff matchups so far. And, like SRS, a full-season measurement like Pythagorean W-L can’t take into effect how well a team is playing at any particular point in time. Given their second-half performance (9-11 for London and 11-9 for Philadelphia) and their similarly unlikely paths through the playoffs, I’d say they’re pretty much the same on that count.

That said, I’m going to go with the numbers and pick London to win it all. I think they’re finding their stride again after a rough second half and playing like the team we saw in the first half of the Overwatch League’s inaugural season. Whatever the result, is should be an enjoyable matchup!