The Roar
The Roar


Matching perception and predictions with reality

Autoplay in... 6 (Cancel)
Up Next No more videos! Playlist is empty -
Roar Guru
6th May, 2021

Followers of my writing know I work with a group called Following Football, which uses a simple method to rate teams in a wide range of sports – all football-adjacent such as Aussie Rules, American football, rugby and the like – and use those ratings to forecast game outcomes.

My leagues of interest are the NFL, CFL, American college football and, of course, the AFL.

Despite the simplicity of our methods (broadly, we adjust ratings by 1/8 the amount a team exceeds or fails to meet projections, which defines this as an ELO system like chess uses), our accuracy rivals any other mechanical forecasting method around in every sport we apply it to. In the AFL, the system has outperformed oddsmakers in 39 of 63 games this season, a 62 per cent success rate against the spread; it has correctly picked 44 games outright.

But part of what I do with the AFL during the season is track the predictions of others as well. Primarily we use it as a control for our own forecasts, but it also produces a record of what the expectations were going into a game to compare with what really happens. Sometimes it’s unrevealing, but sometimes, it’s a fascinating study in what we believe about a squad and how those beliefs change throughout a season.

This year, I’m starting with the oddsmakers’ point spread for a game and then adjusting it slightly to accommodate the forecasts of all the other sources I track – The Roar’s is one of about 16 I’m using this season, although it’s one of the few that doesn’t come up with its own expectation of the final margin. Those win-loss expectations create their own ladder, and here it is:

1. Western Bulldogs (8-0, +139 margin; 128 per cent)
Now’s a good time to mention that the percentage you’re seeing comes from applying that cumulative margin from the forecasted point spreads to the team’s actual point totals: it isn’t inherently more accurate that way or anything, but we all tend to think in percentages more than we do in cumulative margins, like we might in soccer. Excuse me: football. Sorry, I’m American.

At last, the Doggies are alone at the top of the prophecy ladder: they were consistently expected to be good and they have been consistently good. No real surprise here. The surprise is – or at least might be – who was tied with them until this round.

Tom Liberatore of the Bulldogs

(Photo by Daniel Pockett/Getty Images)

2. Port Adelaide (7-1, +131 margin, 126 per cent)
3. Geelong (7-1, +122 margin, 125 per cent)
Not the Power, although they were six expected wins to none until being two-point underdogs to Brisbane last week. No, it was Geelong which we expected to be 7-0 coming into this weekend, despite the stunning struggles they faced to start the season. It wasn’t until they were nine-point dogs to Richmond in Friday night’s upcoming clash that the top-of-the-ladder position fell from their mantle.


That perception of dominance isn’t far from reality: our own ratings expected them to win six of their previous seven, with only West Coast as a two-point favourite over them on our books in week six (the consensus that week, however, was Geelong by eleven. Turns out, West Coast was destroyed at GHMBA by 97; so much for our vaunted system’s accuracy).

4. Richmond (6-2, +97 margin, 119 per cent)
5. Collingwood (6-2, +49 margin, 109 per cent)
That’s right: the 17th-place club has been expected to win five of its first seven games (plus this weekend’s clash versus the 18th-place Kangaroos). That puts the wave of “what’s up with Collingwood?” articles in perspective: nobody’s writing extensive analyses describing “what’s wrong with North?”, because we’ve known what North was all season: a very young team under construction after what amounts to a complete collapse.

But here’s the list of game-by-game projections for Collingwood so far in 2021: lose to the Bulldogs by four (that’s right: just four points!), beat Carlton by four (that was their sole win so far), beat Brisbane by two (they lost by one – so far, not bad), defeat GWS by 22 (lost by 30 – first “uh-oh”), lose to the Eagles by 17 (lost by 27), defeat Essendon by six on Anzac day (lost by 24), and beat Gold Coast by 15 (they also lost that game by 24).

Brodie Grundy kicks

(Photo by Daniel Carson/AFL Photos via Getty Images)

The average winning margin predicted for Saturday? Just 21 points over the 0-7 Kangaroos, a team that’s lost by at least 30 every game this year. Stay tuned.

6. Melbourne (5-3, +99 margin, 120 per cent)
7. West Coast (5-3, +61 margin, 110 per cent)
8. Brisbane (5-3, +36 margin, 107 per cent)
No real surprises here: both the Eagles and the Lions are 4-3 with the expectation of winning this weekend, so this is spot on their ladder position. Melbourne is unbeaten, of course, but that wasn’t the way they were anticipated to start this season. Those three projections of loss came in games two, four, and six to St Kilda, Geelong, and Richmond, respectively – all teams that were projected to stay ahead of the Demons on the ladder this season as they were last. But this points up the over-the-top performance of the red and the blue in 2021 so far.


9. Sydney (4-4, – 24 margin, 96 per cent)
10. Adelaide (4-4, – 44 margin, 93 per cent)
Next are the two other surprise packages, teams that admittedly have faded a bit in recent weeks but started out surprising the collective consciousness with unbeaten starts (except where one beat the other in R2). Sydney surprised us in Rounds 1 and 4 to go 4-0, surprised us again by falling in Rounds 5 and 6, and surprised us a fifth time with their two-point upset of the Cats last weekend. That should make their 27-point underdog status against Melbourne worrisome for Demons fans. The Crows’ upset of Geelong in Round 1 was balanced out by upset losses to Fremantle in Round 5 and GWS last week.

James Rowe

(Photo by Mark Brake/Getty Images)

11. Fremantle (3-5, +16 margin, 103 per cent)
12. St Kilda (3-5, -33 margin, 95 per cent)
13. Gold Coast (3-5, -64 margin, 89 per cent)
14. Carlton (2-6, -51 margin, 92 per cent)
15. GWS Giants (2-6, -52 margin, 91 per cent)
16. Hawthorn (2-6, -94 margin, 85 per cent)
Nothing to write home about here. The Giants are probably the most notable of this group, having fallen from the most recent heights, but the consensus surrounding the charcoal and orange was that they were in for a rough winter considering how much talent they’d lost and, more recently, how much more is injured. It is interesting that the one game they were expected to win (besides the upcoming game Saturday) was back in the opener, when St Kilda exposed them. They were then consensus underdogs for six straight, including each of their three recent wins.

Josh Kelly of the Giants celebrates kicking a goal

(Photo by Cameron Spencer/Getty Images)

The last two teams pose an interesting dichotomy.

17. Essendon (0-8, -117 margin, 84 per cent)
18. North Melbourne (0-8, -271 margin, 62 per cent)
I don’t know anyone who would place these two teams on the same level, yet neither team has been expected to win any game they’ve been scheduled to play, including their games in Round 8 upcoming. North certainly merits those expectations: sitting 0-7 on the real-life ladder, a margin of minus-391 to their names and a sub-50 percentage indicating that they have if anything been over-estimated.

But Essendon was one of those dark-horse candidates to sneak into the eight this season. They were 6-10-1 last season, 13th place on the final ladder, and it wouldn’t have got you fitted for a straight jacket to have placed them in your tips to make finals in 2021. Yet no single game has merited tipping the Dons.


Sports opinion delivered daily 


Consider their season thus far: Hawthorn favoured by four (won by one), Port by 31 (won by 54), Saints favoured by 16 (Essendon routed them by 75), despite that rout, unbeaten Sydney favoured by 23 (they won by 53), Brisbane favoured by 18 (they won by 57), Magpies by six on Anzac day (Bombers won by 24), and Carlton by five last week (they beat the Dons by 16). This weekend, they are consensus 14-point underdogs to an erratic Giants team playing relatively well right now.

Could they win? Sure. Will they? As usual, we don’t seem to think so. Hawthorn and Carlton were each a coin flip, Port and Brisbane are demonstrably better than the Dons, Sydney was hot as a furnace in Round 4 and St Kilda and Collingwood both seemed marginally stronger than Essendon at the time of their games, as the Giants do this week.

If Essendon does win on Saturday afternoon, it sets up the definite possibility of breaking the expectation duck next week when they host Fremantle in Round 9. If it doesn’t happen then, it’s guaranteed to happen for one of these two teams in Round 10, because the Kangaroos and Bombers face off in Marvel Stadium in the anticlimax of the round.

Nat Fyfe of the Dockers

(Photo by Will Russell/AFL Photos via Getty Images)

It seems unlikely that the red sashes wouldn’t be favoured, but if North breaks out of the winless column this week or next (and both of their games are winnable, with Collingwood and Hawthorn on the other side)? It’s very possible that North opens their expectation register before the Dons.

But I don’t expect it.

A side note here – we also annually track the Once-Around Ladder, the hypothetical outcome if the AFL were to have a strict 17-game, everyone-plays-everyone-once fixture that takes only the first meeting of any double to see if the outcome is any different than the longer 22-game fixture produces. The answer is often “no”, but it’s an interesting endeavour.

We also do an 18-game variation which allows for duplicate derby games, including sometimes arbitrary intra-Melbourne match-ups depending on what the AFL’s schedule makers give us each season. So far that has not produced anything wilder than North Melbourne not falling off the wagon back in 2016. But when we start having duplicate games, we’ll share that ladder with you too – presumably, that won’t be worth sharing until the beginning of August or thereabouts.

Please don’t ask why we didn’t publish it last year until you’ve thought for a moment.