For a seventh consecutive year, it’s time to look back in order to look forward.
This is Foxfooty.com.au’s annual Pythagorean wins prediction piece, which reveals the teams who over- and underperformed in the previous season to predict who’ll bounce back or fall flat in the coming year.
The name makes it sound complicated, but all you need to know is it’s accurate. Of the 15 strongest predictions made by this formula since 2010, 13 were correct.
Watch every match of every round of the 2023 Toyota AFL Premiership Season LIVE on Kayo Sports. New to Kayo? Start your free trial now >
The core behind Pythagorean wins is figuring out who was lucky and who was unlucky in the previous seasons – determining how many games a team “should” have won – and predicting who’ll rise and fall this season as their luck reverts to the mean.
And if you read anything we wrote during last season, you probably know which team is about to show up.
Funnily enough, this time last year we wrote about how Collingwood was unlucky in 2021, and likely to bounce back in 2022 – we just didn’t think they’d become the greatest close game performers in VFL/AFL history – and how Port Adelaide was lucky in 2021 but likely to be less lucky in 2022.
The Magpies and the formerly-just-Magpies have certainly swapped spots this time.
WHAT ARE PYTHAGOREAN WINS AND HOW DO THEY WORK?
At the core is a formula, Pythagorean expectation, developed by baseball stats guru Bill James (this is why we used the word Moneyball in the headline – he’s actually relevant to it!). It estimates how many games a team ‘should have’ won based on its scoring. In baseball, this means runs scored and allowed.
This works because looking at a team’s attacking and defensive performances overall, rather than purely whether they won or lost matches, gives us a larger sample size and more information.
Let’s go to the extremes. Imagine two AFL clubs. Club A finished 9-1, winning all of its games by one point and losing the other by 100 points. Club B finished 8-2, all of its wins coming by 100 points, its two losses coming by one point.
Which team is better? The first team won one extra game but its percentage, and arguably its performance, is and was worse than the second team. We’d argue the second team was more impressive; Pythagorean expectation uses the same principle.
There is a core principle at work here: Good teams win games. Great teams win games by a lot.
Using research by footy statistician Tony Corke, we can adapt the formula used to analyse baseball teams to study the AFL.
It doesn’t mean we make a prediction on every team – you need a certain gap between a side’s actual wins and Pythagorean wins (at least 1.5, ideally 2+) for the statistical significance to be meaningful.
Since 2010, of the 17 teams that had a gap of 2.5 games or more between their actual and Pythagorean wins, the formula has accurately predicted whether they would rise or fall 14 times.
A prime example of this is Brisbane in 2018. They finished in the bottom four with just five wins, but with a percentage of 89.1% – the percentage of an 8.5-win team. They had been unlucky, and Pythagoras thus expected their luck to turn in 2019.
Safe to say it did, because the Lions hosted a qualifying final in 2019.
SO HOW DID PYTHAGORAS’ PREDICTIONS GO LAST YEAR?
Not amazingly. The formula didn’t have many strong tips last year, with just one team (Fremantle) having a two-win gap between its 2021 expected and actual wins, and just barely.
The Dockers were expected to decline but didn’t; the Bombers were expected to improve but didn’t. But Collingwood, the third-strongest tip, did climb the ladder as predicted. (Just, you know, way higher than predicted.)
The formula has much stronger predictions for the 2023 season.
Selwood’s retirement code switch! | 01:22
WHAT PYTHAGOREAN WINS PREDICTS FOR THE 2023 SEASON
Collingwood declining and Port Adelaide improving are two of the strongest predictions this stat will ever make; it’s as simple as that.
Last season 12 teams finished on the ladder basically where they should have, with their real win total within one win of their Pythagorean win total.
Fremantle slightly over-performed, while Richmond, GWS and Gold Coast somewhat under-performed – just enough to be significant, and we’ll analyse them below.
But the Magpies won a remarkable four more games than their percentage says they should have; the second-biggest gap since 2010. (The record-holder is still 2013’s North Melbourne, who won 10 games when they ‘should’ve’ won 14.6 – that was a season where they lost games by 1, 1, 2, 3 and 4 points, three of those against finalists.)
Meanwhile the Power were almost as unlucky as the Magpies were lucky, winning three fewer games than their percentage says they should have.
PYTHAGOREAN WINS LADDER FOR 2022
1. Geelong – 17.71 wins (Real win total: 18) [Differential: +0.29]
2. Melbourne – 16.22 (16) [-0.22]
3. Sydney Swans – 15.88 (16) [+0.12]
4. Richmond – 14.98 (13.5) [-1.48]
5. Brisbane Lions – 14.62 (15) [+0.38]
6. Fremantle – 14.25 (15.5) [+1.25]
7. Port Adelaide – 13.05 (10) [-3.05]
8. Western Bulldogs – 12.80 (12) [-0.80]
9. Carlton – 12.69 (12) [-0.69]
10. Collingwood – 11.90 (16) [+4.10]
11. Gold Coast Suns – 11.59 (10) [-1.59]
12. St Kilda – 10.85 (11) [+0.15]
13. Hawthorn – 8.73 (8) [+0.73]
14. Adelaide – 8.03 (8) [-0.03]
15. GWS Giants – 7.57 (6) [-1.57]
16. Essendon – 7.25 (7) [-0.25]
17. West Coast Eagles – 2.65 (2) [-0.65]
18. North Melbourne – 2.08 (2) [-0.08]
Positive differential (ie +0.29) = lucky; negative differential = unlucky; differential over ~1.5 is statistically significant
On the real ladder, Collingwood finished fourth with 16 wins and a percentage of 104.3%, while Port Adelaide finished 11th with 10 wins and a percentage of 110.3%.
We are not going to say the Power were a better team last year – the Magpies beat them when they played, got better as the season went along and were obviously very competitive in September – but across the course of the season, the two teams were comparable.
A lot of what separated them, in the end, was luck. Collingwood went 11-1 in games decided by two goals or less; Port Adelaide went 2-7.
And we know people are still going to argue against the idea that close games are luck. Sure, let’s give Craig McRae some credit. And as you might’ve heard a million times last year, Collingwood trained close game scenarios every now and then.
But explain this: if close game performance isn’t luck, why did the Power go 9-0 in close games across the 2020 and 2021 home and away seasons (but then go 2-7 last year)? And why did the Magpies go 2-8 with a draw across the 2020 and 2021 home and away seasons (but then go 11-1 last year)?
Did McRae sneak into Alberton and perform some kind of Space Jam-style heist of the magic inside the Port Adelaide playing group? Or were the two teams outliers in 2020/21, then not just reverting to the mean in 2022 but speeding past it in the other direction? All evidence points to the latter.
The point is this: most teams finish with an average record in close games. A team that over- or under-performs in close games one year would be expected to return to an average record the next year. And that return to average is what we’re predicting here – it’s one (particularly strong) factor that suggests Collingwood will win fewer games, and one factor that suggests Port Adelaide will win more games.
Anyway – let’s go a bit deeper into the five teams Pythagoras feels strongly about for 2023.
THE TWO VERY STRONG PREDICTIONS
Collingwood
2022 record: 16-6, 104.3%
2022 Pythagorean wins: 11.90 (gap of 4.10)
2023 prediction: Decline
The Magpies were 10th on the Pythagorean wins ladder list above; we do not think they were merely the 10th-best team last season. Clearly, in some of their close wins, they outperformed the opposition in crucial moments. (The Essendon game, with the easy-as-you-like D50 to F50 transition and Jamie Elliott goal, comes to mind.)
But, as we’ve discussed ad nauseum, they were fortunate to win so many close games in the home and away season. And we saw that in the finals, when in the most ironic results possible, they lost twice by a goal or less.
As a reminder: last year’s Magpies won games by 1, 4, 4, 4, 5, 5, 6, 7, 7, 11 and 11 points. (Between July 2 and August 5, they won six games by an average of 5.6 points. Absurd.)
Moore voted in as captain by teammates | 01:32
In contrast, they only won two games by 30-plus points. Remember that core principle from above? Good teams win games; great teams win games by a lot. (Geelong won 12 games by 30-plus points last year.)
They also won close games while completing a lot of comebacks – from 26 points down vs North Melbourne, 25 points down vs Carlton, 21 points vs Essendon and 10 points down vs Hawthorn (all fourth-quarter deficits), plus from three to four goals down early against Melbourne, Port Adelaide and Adelaide.
Here’s something you might notice from the teams listed in that paragraph – just one, Melbourne, played finals.
One could argue that, while completing those comebacks is very impressive, it would be more impressive to not trail by that much against those teams in the first place.
We’ll say this: it would be absolutely staggering if Collingwood won more than 16 games next season. But even if we take away a few of those close wins and say they’re a 12-win team, like Pythagoras suggests, they’re clearly still a finals contender.
We understand that all of our commentary around the Magpies being lucky sounds negative, Collingwood fans. It is just the mathematical reality. They were. And it was absolutely incredible to watch! No-one can ever take that away from you.
But there is essentially zero chance it will be repeated. Heck, there’s a legitimate chance the Magpies are a better team in 2023 but win fewer games, purely because their close-game luck turns.
Port Adelaide
2022 record: 10-12, 110.3%
2022 Pythagorean wins: 13.05 (gap of 3.05)
2023 prediction: Improve
The Power are a prime example of a team that wasn’t that different from year-to-year… but were a whole lot less lucky.
In 2021, they won 17 home and away games with a percentage of 126.3%, including an AFL-best record of 5-0 in close games (this was back in the old, pre-Collingwood days when that figure was impressive).
In 2022, they won 10 home and away games with a percentage of 110.3%, including an AFL-worst record of 2-7 in close games.
Right there you can see a big chunk of what changed for them – five close wins turned into five close losses. Clearly the Power were worse in 2022, taking a while to get going with an 0-5 start where their defence looked quite shambolic, but they weren’t seven wins worse.
They were broadly competitive last year. Take out the Round 2 loss to Hawthorn (which made no sense at the time, and even less as the season progressed), and the Power’s losses came against Geelong (twice), Melbourne (twice), Richmond (twice), Brisbane (by 11 points on the road), Fremantle (by 8 points on the road), Collingwood (by 6 points on the road), Adelaide (by 4 points after the siren) and Carlton (by 3 points on the road).
That’s a brutal list of fixtures – and only three of those losses were by more than 14 points.
Based on Pythagorean wins, the Power were a 15.65-win team in 2021, and a 13.05-win team in 2022. That sounds about right to us; worse, but not that much worse.
And so, if they can finally have a normal year in close games, they should be somewhere in the middle, right? 14 or 15 wins, meaning a finish between 4th and 6th, sounds like an entirely reasonable prediction.
At the very least we would expect the Power to win 11 or 12 games and contend for the top eight, which is the minimum expectation they face in 2023.
Powell-Pepper signs extension with Power | 00:31
THE THREE LESS-STRONG PREDICTIONS
All three of these teams are on the borderline of statistical significance; these predictions are much stronger when the difference between a team’s actual and Pythagorean wins is over two.
But they are still intriguing sides and worth discussing.
Gold Coast Suns
2022 record: 10-12, 102.8%
2022 Pythagorean wins: 11.59 (gap of 1.59)
2023 prediction: Improve
After a three-game winning streak surrounding their mid-season bye, last year’s Suns were a legitimate chance to make the eight, sitting 7-6 with five games left against teams that wouldn’t end up playing finals.
But a pair of heartbreaking losses to Port Adelaide (by two points on the road) and Collingwood (by five points at home) put them in serious trouble, and while they stunned Richmond after the siren a week later, when they lost to Essendon by 48 points in a dour Sunday twilight performance their year was effectively over.
The Suns went 2-3 in close games in the end, so this isn’t really a case of bad luck. They just faded late in the year, as they always seem to do.
Realistically their percentage was slightly boosted by a bottom four-level fixture (they played Hawthorn, North Melbourne and West Coast twice) so while we do think Gold Coast could continue to grow and finally play finals, they don’t have a particularly strong Pythagorean case.
GWS Giants
2022 record: 6-16, 84.6%
2022 Pythagorean wins: 7.57 (gap of 1.57)
2023 prediction: Improve
This is a case where Pythagoras doesn’t know about the Giants’ latest off-season teardown.
Tim Taranto, Jacob Hopper, Bobby Hill and Tanner Bruhn certainly helped GWS perform better than its six-win record suggests, but they won’t be there this season; situations like this are why it’s always worth studying more than just the numbers to predict a team’s fortunes.
While we like the Adam Kingsley hire, a seven-win or better year in his senior coaching debut would be a reasonable surprise.
Richmond
2022 record: 13-8-1, 121.6%
2022 Pythagorean wins: 14.98 (gap of 1.48)
2023 prediction: Improve
It’s kind of remarkable that Richmond didn’t win a final last year, given how good they looked over most of the season.
After a 2-4 start with odd losses to St Kilda and Adelaide, they lost just four more games, all of them close – by 6 points against Sydney, 3 points against Geelong, 2 points against Gold Coast and 4 points against North Melbourne.
The latter two, plus the draw against Fremantle, came in a three-week patch which ruined their chances of a home final – and in the end they lost yet another close game in September, by two points to Brisbane at the Gabba.
So we actually prefer the Tigers’ Pythagorean case over the Suns’ and Giants’. They went 2-5 with a draw in close games if you include the elimination final, and more to the point, they were very clearly a much better team after Round 6. They pass the eye test as a team that underperformed its final ladder position and win-loss record.
Oh, and then they added Jacob Hopper and Tim Taranto to try and fix a midfield that has long been their weakness (if they had one). We’d argue the defence is slightly shaky, which could cost them a few games, but expect most experts to be tipping a Richmond rise in 2023. (We’ll be one of them.)
