The (non-definitive) dos & don'ts in terms of what info to use for team evaluation, in no particular order 🧠
Do use ✅
- Models 📈
- Useful & predictive team stats 🔢
- Eye test 👀
- Discussion 👩💻
Don't use ❌
- Most media pundits 👨
- The ladder 🪜
More info for each below
We hear a lot of talk among AFL media about 'modern midfielders' being tall 🤔
Top 8 from the Brownlow:
Neale: 177 cm
Bont: 193 cm
Daicos: 184 cm
Butters: 181 cm
Gulden: 175 cm
Trac: 186 cm
Viney: 178 cm
Serong: 178 cm
#AFL
#BrownlowMedal
Just how much of an outlier was last night's result? The chart below highlights that game vs. all others in the past decade
The Dees were +32 inside 50s. Teams that win that metric by 30-35 typically win by 57 points 🤯
#AFLPiesDees
#AFL
Round 5
#AFL
team ratings quadrants
🙏🙂
😭🫤
If you had a laugh at the Fox Footy 'premiership window' chart, here's some quadrants you can take seriously
The Eagles get their first W of 2024 & move from despair to hope 📈
How much does winning impact which teams get the votes on Brownlow night? 🤔
Simply winning the game at all typically means the triumphant team gets 4 votes. If the team won by 3-4 goals, then usually they're walking away with at least 5 votes 🥇
#AFL
#BrownlowMedal
Round 3
#AFL
team ratings quadrants
🙏🙂
😭🫤
North & West Coast look like they're gonna be breaking my charts again this season by being so far below average, sigh
Somewhat relieved to see it's not just my model tipping the
@GoldCoastSUNS
this weekend
@SquiggleAFL
crew - why are your models tipping them? I will put the reasons for my model tip in the replies
#AFL
#AFLSunsPies
This season might end up being the end of an era for the
@GeelongCats
If it is, what a run it's been. Going to back 2004 (the season prior they missed finals), their dominance is unmatched
#AFL
Taken a lot of crap on this platform for my model not liking the Pies (relative to pundit consensus) & they entered the season ranked 6th
I’m honestly surprised they’re 0-3, but if ever there was a premier liable to have the bottom fall out, it was this team
#AFLSaintsPies
Round 2
#AFL
team ratings quadrants
The rating vs. average (X axis) shows how a strong a team is overall 📊
The trend (Y axis) benchmark will be rating vs. start of season & then after 5 games it will be vs. 5 games ago. The aim is to show the team's form trend 📈
More info ⬇️
What's more shocking: a sensible AFL policy we already loosely knew of, or the fact a lot of people are seemingly oblivious to the fact rich young men do coke? 🤔
New chart, who dis 👀
Recently I've been thinking of ways to leverage the model ratings into stuff that's simple & engaging 😍
If you like my content, lemme know what you think of this. Open to completely new chart suggestions as well 📊
.
@PAFC
close game results the past 3 seasons
2021: 5-0
2022: 2-7
2023 so far: 6-0
A great case study in how your past record in close games isn't indicative of your future record
#AFL
#AFLDonsPower
'Your model has team X too low/high, it must be broken 😂'
This is a common & unhelpful criticism of any model. It's also untrue that any decent analyst would rely solely on their ratings
Ahead of the 2024
#AFL
season, here's my brief guide for team evaluation in a 🧵
Interesting, though hardly surprising, that after all we've heard from commentary teams about how much easier it is to score from centre bounce with the 6-6-6 rule that the data does not support that conclusion at all
Feel like there's been a lot of different
#AFL
game start times this season?
In the 2023 H&A season there is 32 unique scheduled start times, up from 28 last season
In each season, there has been 6 start times that only featured a single game
As one of the many
#AFL
modellers on here, the most consistent criticisms I see is something along the lines of: 'how can you have X team so low [or high]?' 🤔
Genuinely asking - if the models always aligned with the ladder / pundit consensus, what would be the point of them?
Stats like 'X% of teams that went 0-2 failed to play finals' aren't as helpful as they seem❌
Captured in overall %s like that are terrible teams that have no chance. A good team going 0-2 doesn't have the same chance of playing finals as all teams that went 0-2, clearly
ANALYSIS | Footy fans beware. Your club can do irreparable damage to its premiership hopes in the coming fortnight, based on the league’s almost foolproof two-game rule.
How is it 2024 and people are still so staunchly anti-analytics in sport?
What’s more likely - that analytics is useful in almost all industries, but it can’t help sporting leagues, or the former players are wrong?
#NFL
This is what happens when you try to grade football players with an algorithm...
CJ Stroud's performance was “graded” a 77.8
And people treat this shit as gospel.
🤦🏼♂️
How did the
#AFL
ladder come to be?
1-4: pretty steady ➡️
5-8: no one saw that coming 📈
9-14: complete chaos 😵💫
15-18: got a bit spicy at the end 🌶️
#AFLFinals
There's been a big, big resurgence from the west of the town 📈
@GWSGIANTS
timeline:
Round 3: team-worst rating after a home loss to the Blues 🫠
Round 11: only ahead of the Eagles & North 😬
Round 17: risen to league average 😏
Semi finals: 3rd ranked team 😤
#AFL
#AFLFinals
The recent fall of
@melbournefc
from sharpshooters to wayward in front of goal is a good example of how accuracy is typically quite random
Splitting the 2022 into the 1st & 2nd half for each teams, we find goal accuracy was actually negatively correlated
#AFL
Clearly I don't buy into the close game thing (see my recent posts if that's not clear)
But is this not an odd thing to post when they lost the PF by only 16? Was that game just not close enough for whatever it is they've mastered in tight games to get them over the line?
🤔
The other day I asked a question on here: have North improved under Clarko? 🤔
To try answer this objectively, here's their 5 year rating trend*
2023 was better than 2022, but after losing to the winless Hawks by 45 & the 1 win Crows by 57, they're falling of the cliff again 📉
Round 5
#AFL
model tips 🤖
Dees by 12, with a 63% chance of victory 👿
Dogs by 18, 70% 🐶
Giants by 7, 56% 🟠
Blues by 19, 70% 🔵
Suns by 22, 74% 🌞
Power by 24, 75% ⚡️
Cats by 51, 92% 🐱
Tigers by 9, 58% 🐯
@MWL152
This is the funny thing about a lot of debates around stats and sport - whatever commentators and fans say, teams use analytics because they know it works
Debuting some new content this week 👀
How'd the model go this round? 🤔
Note these predictions differ slightly from what I posted pre-round. I decided to change my source data on a whim, which then turned into a slight tweak to the model as well (lol)
When it comes to the debate around close game results in the
#AFL
a common line of argument is that good teams perform better in close games than bad ones
That doesn't show up in the data though - there's zero correlation between performance in non-close games and close games
Brownlow Medal vote streaks 🔥
Danger's 2017 season will always be overshadowed by Dusty's achievements in the same year, but what a season it was 😤
From rounds 9-17 he polled: 3, 3, 2, bye, 2, 3, 3, 3, 3
Will this list be added to tonight?
#AFL
#BrownlowMedal
All I hear is talk of an even competition this season in the
#AFL
, & yet all I see are predictions of a GF featuring the Lions, Pies or Giants🤔
People are also being too quick to discount the Dees. Less than ideal offseason, but my model has them ranked 1st right now 🤷♂️
Round 1
#AFL
model tips
Blues by 12, with a 67% chance of victory 🔵
Pies by 1, 54% 🥧
Hawks by 9, 63% 🟤
Giants by 29, 83% 🟠
Cats by 18, 74% 🐱
Suns by 4, 57% ☀️
Dees by 6, 60% 👿
Power by 47, 92% ⚡️
Freo by 3, 57% ⚓️
Tips: 3
Bits: 0.19
MAE: 21.8
1. The Suns got a big boost destroying the Hawks last weekend
2. The model is in general low on the Pies. This has stemmed from a lack of dominant performances and the fact they haven't done well on raw scoring shots (though they're typically better on expected score)
Thought about the
#AFL
pre-season & the below is where I've landed in terms of what it means for the upcoming season 🤔
Match sim: meaningless
Pre-season: if a team wins decisively, that might be a good sign, & definitely isn't a bad one. The opposite is true for a decisive loss
Round 16
#AFL
elo model tips
Lions by 22, with a 77% chance of victory
Swans by 6, 60%
Crows by 47, 92%
Dogs by 34, 86%
Suns by 13, 69%
Power by 26, 80%
Blues by 18, 74%
Dees by 24, 79%
Saints by 46, 92%
Tips so far: 93
Pre-season
#AFL
team ratings 📊
These rankings are different from the post-GF rankings for 2 reasons:
- Ratings have been regressed back to the mean for a new season
- Model parameters, most notably HGA, have been updated to include the 2023 season
.
@kanecornes
Top 8 Physical Specimens in the League:
1-CHARLIE CURNOW
2- TOUK MILLER
3- KOZZIE PICKETT
4- NICK DAICOS
5- SHAI BOLTON
6- GARY ROHAN
7- CHANGKUOTH JIATH
8- ISAAC QUAYNOR
#AFLTrade
Round 4
#AFL
team ratings 📊
The Dees have established themselves at the top, while at the other end of the table North are at risk of being overtaken by the Eagles 😬
In regards to midfielders and height - does height typically benefit players in certain areas of the game? 🤔
This is not a definitive analysis, but at a glance it seems like the answer is no ❌
Only for contested marking does it seem like being tall is an advantage
#AFL
We hear a lot of talk among AFL media about 'modern midfielders' being tall 🤔
Top 8 from the Brownlow:
Neale: 177 cm
Bont: 193 cm
Daicos: 184 cm
Butters: 181 cm
Gulden: 175 cm
Trac: 186 cm
Viney: 178 cm
Serong: 178 cm
#AFL
#BrownlowMedal
Updated this chart for 2023 given the recent discourse
Kicking accurately & being efficient inside 50 are keys to winning, but they're not sustainable
Controlling territory & generating scoring shots is something you can take to the bank 🏦
#AFLFinals
#AFL
Genuine question: how does Crouch get cited for 'forceful front-on contact' with grading:
Careless conduct
Medium impact
High contact
And Butters doesn't even get cited? Maybe it was low impact, but that's still a fine. Laughably inconsistent, even by MRO standards
Round 2
#AFL
model tips
Pies by 3, with a 52% chance of victory 🥧
Crows by 21, 72% 🐦⬛
Freo by 16, 67% ⚓️
Dees by 16, 67% 👿
Swans by 38, 86% 🦢
Suns by 1, 50% ☀️
Power by 8, 58% ⚡️
Giants by 27, 78% 🟠
@MWL152
Right, so it's just self-reporting followed by a confirmatory test? Presumably so they don't avoid playing when the drug is actually out of their system
Story seems like nothing tbh