End of Season Playoff Predictions

It is that time of the year in which most of the NWSL is focused on a push for the playoffs. Right now, there is only 6 points (which is possibly only 2 games), that separates Orlando in 5th (28 pts) with North Carolina in 11th (22 pts). Angel City is right in the thick of it with 23 points in 10th place. So how do our chances look?

Methodology

I used the Pythagorean Predicted Points Formula to predict how the table will look at the end of the season. This model was originally created by Bill James (who you’ll know if you’ve read Moneyball), and the baseball version is simply (Runs Scored^2) / (Runs Scored^2) + (Runs Allowed^2). There is no actual relationship to Pythagorus, but the formula reminded James of the Pythagorean Theorem, thus the name. Because soccer doesn’t just have wins and losses, like baseball, the soccer formula has been adapted to:

Where: GF is Goals For, GA is Goals Against, m is the number of matches, and a, b, c, and d are empirical coefficients.

For my purposes, I used Clive Beggs coefficients that he developed for the English Premier League (a = 2.78, b = 1.24, c = 1.24, d = 1.25). I applied this formula to the 2023 and 2024 seasons, and got an 88.27% correlation in 2023 and a 97.84% correlation in 2024. Especially because 2023 stands out as an exceptionally odd year, I think that this is strong enough to proceed, although it could be fun to tinker with these coefficients a little more.

The Predicted End of Season Table

So, obviously, this formula is just looking at the Goals Scored and Goals Conceded of each team and making it’s prediction just based on that. I did actually try this with xG as well, but actually found less of a correlation for the actual seasons results in 2023 and 2024. So factors such as a team’s best player going to Chelsea, for instance, are not taken into account. PredTotal (or the Predicted Point Total) is the column you’re probably going to want to look at, and Angel City is remaining in 10th place in this prediction. That seems about right, as the Pythagorean expectation at the current point in the season (PythagPTS) has Angel City overachieving by half a point and Houston underachieving by half a point. Houston have a slightly better goal differential, so I think that we’d expect them to remain above us.

In fact, the current playoff teams are all predicted to be the actual playoff teams. Louisville, strictly by Pythagorean expectations, should finish below Houston, but they’ve already banked 3 points more than expected, and that lead should carry them through. There are 21 points still available, but it just doesn’t seem like there is enough time for anyone to close the existing gaps.

Gotham is probably the most notable team. They are the only one expected to change their position, overtaking Orlando for 5th. They’ve been underachieving the most this year, with 5.3 points below expectation. In fact, solely going off the Pythagorean Predicted Points (PythagTotal), Gotham would finish second behind Kansas City. Again, this doesn’t sound crazy to me. Gotham has the second best defense and the second best Goal Differential. They’re in 6th place, but they have all the markers of a team that could go deep in the playoffs, if they put it all together, and that was before they signed Jaedyn Shaw. Of course, sometimes underachieving means being unlucky, and sometimes it actually means underachieving. But Gotham are definitely the team to watch for the stretch run, in my opinion.
The other thing that I find really interesting is that Kansas City is actually overachieving. They’re clearly a very, very good team. But maybe they’re not quite as good as they seem? Or maybe they’re just a well coached team that manages to get maximum points out of close games. In many ways, a team’s under- or overachieving is similar to a striker under- or overachieving their xG. This provides a level of expectation, but the really good teams will consistently do better than expected.

Conclusions

I fully expect at least one team to defy these predictions. This is a prediction based on a full season, but we all know that a team can get hot (or cold) for a stretch, and with the table as closely packed as it is, that can mean a big run up or down the table. The smaller the sample size, the harder it is to predict. But I think this prediction does provide a framework for how to think about the rest of the games. Especially from an Angel City perspective, this formula is largely based on a team that has Alyssa Thompson on it and it still thinks that we’ll miss the playoffs. And frankly, I’m okay with that. To me, finishing 8th, and going to Kansas City in the first round and getting torn to shreds, is not really any better than finishing 9th. I know that it’s a psychological barrier (just ask a Racing Louisville fan), and playoff experience would be good for a young team. But the focus needs to remain on the long term goal of building a championship caliber team, and a 34 point team this year is not dramatically closer to that point than a 33 point team, even if that’s difference of a playoff spot.

At the end of the season, I’ll revisit this and see how accurate it ended up being.

All data was collected from FB Ref.

Angel City’s next game is Saturday, September 13 at 9:30a Pacific, away to the North Carolina Courage, on CBS.