Max Barry
@SquiggleAFL
Followers
6,392
Following
257
Media
700
Statuses
4,525
AFL Prediction & Analysis. Aggregates the internet's finest football computer models. Writes novels as @MaxBarry . Also likes netball.
Melbourne, Victoria
Joined May 2017
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Brazil
• 1211585 Tweets
Elon Musk
• 727945 Tweets
Meu Twitter
• 491510 Tweets
#BUS1stFANCON_KnockKnockKnock
• 199617 Tweets
Adeus Twitter
• 187935 Tweets
Wisconsin
• 116879 Tweets
Caitlin Clark
• 77858 Tweets
Bluesky
• 50055 Tweets
ブラジル
• 47187 Tweets
#GFRunTheWorldConcertD1
• 45255 Tweets
#プレイバックガチャ
• 39196 Tweets
Angel Reese
• 38472 Tweets
#INZM_30Mviews
• 24335 Tweets
michigan state
• 23117 Tweets
Djokovic
• 22689 Tweets
野菜の日
• 20124 Tweets
Popyrin
• 13432 Tweets
Stanford
• 13246 Tweets
台風一過
• 11532 Tweets
I coach kids' netball so I built a mutating genetic algorithm to generate efficient team rosters. No lie, it may be the greatest thing I've ever done.
87
14
305
Years since last negative free kick differential (incl. 2021*):
0 BRI, COL, FRE, GCS, HAW, PTA, RIC, STK
1 CAR, ESS, GEE, GWS
2 MEL, SYD
3
4 ADE
5
6
7
8 WBD
9 NTH
10
11
12
13
14
15
16
17
18
19
20 WCE
11
18
164
Melbourne and West Coast have each defeated one top-8 team this year.
2
8
142
I way underestimated how many people would be interested in my netball AutoRoster. Now I have more DMs than I can reply to. So first, fist bump to my fellow roster victims. Second, I'll put this thing on a public site for everyone in the near future.
I coach kids' netball so I built a mutating genetic algorithm to generate efficient team rosters. No lie, it may be the greatest thing I've ever done.
87
14
305
11
7
113
Ever since I saw a Reddit post predicting the new Tassie team will wind up with a corporate name like Southern Strength I can’t stop worrying.
13
2
104
The AFL did a great job under the circumstances, but 2020 is one heck of an unbalanced fixture. Sucks to be Collingwood, who had one of the hardest journeys and wind up in an away Elimination Final.
29
12
100
This week I've been teaching coloured dots to play footy. They have a lot to learn.
12
5
92
We will be simulating every game of season 2020 in real-time as if it were really being played. Preview below. More to come.
9
12
91
For the first time, Squiggle's 2019 prediction will factor in off-season player movements, incl. players returning from injury. Still a work in progress, but here's how it's shaping up...
54
14
87
What I learned from years of crunching football numbers so you don't have to:
9
16
89
2023 prediction as at December, accounting for form, fixture and list changes.
45
3
85
Who won the round? An algorithmic ranking of clubs that says who had the best and worst weekends, based on their movement in the ladder projections of the internet's finest AFL models. Now in beta.
7
14
81
There’s an alternate universe in which the Pies lost all those and data modellers are trying to convince people they’re actually pretty good.
The Pies have won 13 of their past 33 games by single digits 😎
@JoshGabelich
ranks those victories:
21
4
90
3
3
78
Think it's under-reported how horrendous St Kilda's fixture was this year. Double up games were against Richmond, GWS, Melbourne, Hawthorn, North; had home games against interstate sides only 4 times but had to travel away interstate 5 times + 1 to Kardinia + 1 to Tas.
11
19
77
I wrote an article: When You Should Bet.
7
17
68
My netball AutoRoster is now online - give it a go here!
13
13
70
This is currently the most one-sided tip in Squiggle history, by some margin.
5
2
70
As a Richmond supporter, this is a kind of low blow I never expected when I started this site.
2
0
69
The wrongest Squiggle tip since we started in 2017, beating out Gold Coast's 2018 upset over Sydney (91%).
Western Bulldogs by 59 with a 95% win chance. Western Bulldogs v West Coast @ Docklands
#AFLDogsEagles
1
0
0
3
5
66
One of the most interesting things about this account is how clearly it illustrates the difference between what umps consider important - getting each decision technically correct - and what fans consider important - not unfairly advantaging one side.
#AFLPowerDogs
: Umpired Terribly
#AFLFreoNorth
: Umpired Well
#AFLTigersPies
: Umpired Poorly
#AFLSwansSuns
: Umpired Well
#AFLGiantsCats
: Umpired Poorly
#AFLDonsHawks
: Umpired Well
#AFLLionsEagles
: Umpired Poorly
#AFLDeesSaints
: Umpired Superbly
#AFLBluesCrows
: Umpired Terribly
73
9
279
8
1
63
You fix 90% of what’s wrong with AFL draft bidding if you correctly account for the fact that 10 mediocre players aren’t worth 1 very good player.
4
2
64
I'd like to submit Exhibit W, Your Honour: West Coast crowds have gotten smaller and quieter, and they're tracking for their first negative free kick differential season in 20 years.
9
6
63
Feeling proud and humbled that this year it was only 22 minutes elapsed between the release of the AFL fixture and DMs asking when I’m updating my site.
4
0
62
Net list strength gain vs last game of 2022 due to trades, retirements, return from injury:
Heaps: CAR, NTH, BRI
Plenty: WCE, GCS, STK, RIC
A bit: WBD, ADE, COL
Not much: ESS, HAW, GWS, MEL, SYD, PTA, FRE
Same: GEE
3
2
59
Weird, this model I trained with xScores because they’re more reliable than actual results turns out to be better at predicting xScores than actual results.
4
2
58
Now with fine-grained precision for those matches where 50% doesn't quite cut it.
5
3
58
If the season is called off, I stand ready to painstakingly create a virtual match report for every single game.
4
1
57
What people need to understand about these 17-5 proposals is that it would require me to rewrite so much frickin code
0
2
54
I recommend an AFL project for all budding programmers because it's a great way to learn how reality will systematically violate every single fucking assumption you make.
3
2
56
I love that we had a sub rule, which everyone hated, and through a 6-year period of continual rule improvements, we have made it all the way back to the original sub rule.
4
1
53
Wrote a tipping algorithm for my kid's netball league even though I assumed it'd be way more unpredictable than AFL, but nope, turns out if I could bet on Under 13s community sport I'd be absolutely mint.
1
0
52
This seems to happen a lot:
1. Champion Data produce interesting numbers
2. Explanation is only available to select journos
3. Media piece on numbers omits explanation, is misleading, or wrong
4. Public is like wtf is with those numbers
8
7
50
Hey you, person vaguely interested in probability. I wrote a short piece on how 99% is way more terrifying than you expect. Read here:
4
8
48
It's simple. 8 teams open the season in the Opening Round. The other 10 open their seasons in Round 1, not the Opening Round. Richmond & Carlton open their seasons in the Opening Rounds and then play their traditional season opener in Round 1. Just open your eyes.
3
1
49
The top 4 computer models tracked on Squiggle are beating all 12 expert tipsters at The Age, all 30 at the Herald-Sun, and all 16 at the AFL. That's 4/15 models outperforming 58/58 humans.
7
8
50
Unlike the official AFL one, the Squiggle Ladder Predictor allows you to correct injustices via Fixture ➡️ Allow tipping of past games.
5
5
48
AFL: The first round of 2024 will be Round 0
Data nerds: *incoherent screaming*
1
2
47
Added Power Rankings to the site, based on the cumulative wisdom of the internet's finest models. It's pretty. It's interactive. It speaks truth to power.
6
6
47
The Tigers run away with it after half-time, winning by 22 points. A successful first game in Virtually Season 2020, with an entertaining contest and only a brief spate of database errors. Tune in tomorrow night for Sydney vs Essendon.
7
4
47
If the ump had paid Prestia that 50, there would have been enough dissent from Sydney to bring him within range of the MCG.
2
1
46
Fremantle have an 18% chance of making finals, which isn't too shabby. Roll a die, get a 6, that's the Dockers making it.
4
1
45
Monday Fun Facts - AFL
(1) The team with more scoring shots at qtr time wins 70% of the time
(2) The team in front at qtr time also wins 70% of the time
(3) If you can only be one, it's better to be the team with more scoring shots (by 0.1 percentage points, data 2000-2020).
5
0
43
I reckon in a week of disastrous climate news we don’t give footballers extra cars.
So it turns out the mark of the year winner gets $10K, a pie warmer and a year’s supply of pies. Its only small but surely the AFL can find a car for the mark and goal of the year like in the old days. The 17 changes Andrew Dillon can make in the next year
44
2
60
4
1
43
Have never blocked people on this site before, but starting now that blue tick replies get prioritized and 95% of them are dead set idiots.
3
0
41
The rare exactamundo!
2
2
40
In inner-city Melbourne, HostPlus has begun openly trolling Richmond fans.
1
3
41
The Hawks are virtual premiers! Mitch Lewis slots a goal with only a minute remaining to sinks Port Adelaide by 6 points. Hawthorn came from 5th, winning 4 finals and inflicting Port’s only defeat for the year in the match that mattered most.
12
10
40
I admire AFL Tables' decision to say nah, the first round is Round 1, instead of spending stupid amounts of time rewriting terrible hobbyist code like the rest of us
The real problem with Opening Round is that it's going to make using AFL Tables a little bit more annoying for the rest of time - because they're just referring to it as Round 1, and we're gonna have to double check which round 2024 events really happened in.
15
6
148
3
1
39
Let's not pretend that 100k fans get a seat at the MCG. Members of competing GF clubs are allocated 28-34k tickets. A 50k-seat stadium is fine if you stop giving tens of thousands of tickets to sponsors, corporates and neutrals.
If you want Perth & Adelaide to host a Grand Final, lobby your Govt. to build a stadium that holds 100 k. Until then, shut up.
35
37
624
7
2
37
St Kilda's Fixture: A History ⚜️
2014: rated 18th by Squiggle's model
2015: 18th
2016: 9th
2017: 18th
2018: 18th
2019: 10th
2020: 16th
2021: 18th
4
2
40
Pre-season ladder prediction by AFL Media’s
@_sjblack
has NINE teams in the exact right spot at the moment - half the ladder!
3
1
40
This week in DataViz Crimewatch: where 46 is neck-and-neck with 52, but 43 is miles behind 45.
Who will win the 2021 Coleman Medal? 🥇
Find out which contender has the best run to the finish line:
17
3
74
1
1
39
Things as likely as Geelong winning the minor premiership (about 70%):
- Pies top 3
- Giants top 4
- Tigers top 6
- Eagles / Crows make finals
- Brisbane 4th-8th
- Suns bottom 3
- Carlton bottom 2
- At least 3 of the above being wrong
0
2
37
Expected scores are informative, but expected score *ladders* crush the numbers down to binary wins and losses, destroying most of the info gained. I bet a simple percentage ladder is more predictive.
3
1
38
I thought my site being attacked by maniacs, but it turns out there's a class at La Trobe University with an assignment to build a footy website based on the Squiggle API.
2
1
38
Well that’s the end for Squiggle game tweets. So the commercial bots can stay, but the hobbyists will disappear.
Starting February 9, we will no longer support free access to the Twitter API, both v2 and v1.1. A paid basic tier will be available instead 🧵
11K
17K
20K
7
1
36
This was a successful year for AFL models. For the first time, the average of all models was (ever so slightly) more accurate than the average of all bookies' closing lines, both in terms of tipped margins and win confidence.
1
3
36
Once again the AFL ignores the analytics community’s proposal to rename Etihad Stadium as ID00018.
1
4
36
Screenshot for posterity: the day I made the AFL website. Article by
@ByMarcMcGowan
--> Huge thanks to everyone who helped put this virtual league together, or is following it along! It's been fun.
@Stattraction
@AFLalytics
@AflLadder
@fuzzybluerain
3
3
35
If Essendon v Richmond gets moved to Perth, they'll need oxygen for all the West Coast fans exhausted by booing both sides.
4
4
36
Glad to see Hawthorn have recovered from their virtual premiership hangover.
0
1
35
In 2021 I said, "Of the 26 experts and models I’ve tracked for three consecutive years,
@petryan
has the best record [of predicting the final ladder]... he’s been getting better, too." This year Peter's was the most accurate out of 45 experts & models.
2
1
34
I genuinely hope they enjoy their work at CD because they get smashed on a weekly basis for things they never said nor meant to imply.
“Expected scores” is a concept invented by Champion Data, an alternative reality in which there’s a statistical likelihood that a goal, behind or miss will be scored from a particular position. In this parallel universe, Richmond are equal with Geelong.
48
2
28
3
1
33
Here is my new article: "All I Want for Christmas is an AFL API"
5
9
33
Netball rostering just got even easier! Now you can set a few key positions and let AutoRoster figure out the rest.
7
7
33
Gee this is telling! Locked out of official AFL data, these guys painstakingly compile their own shot data by hand, prompting a small explosion of articles & insights on Reddit, social media - and the AFL’s own site!
#sharingiscaring
Hey imagine that,
@StatsInsider
provides freely available shot data and
@AFLcomau
writes an article using it? Why not use official data?
- Is the process 2 long?
- Is the data too hard to access?
Is this really the situation Aus biggest sport is now in?
5
7
49
3
5
32
An interesting Reddit thread on how the stats look once you adjust for AFLW's shorter game time:
5
3
33
Models are better tippers partly because they can effortlessly consider large numbers of recent results, an advantage that shrinks in finals and is almost gone by the Grand Final. So go with your heart.
5
1
32
West Coast need 37 free kicks on Saturday to avoid finishing with a negative free kick differential for the first time since 2001.
4
3
33
I mean not to nitpick but that's 6 out of 8 columns for info that can be easily inferred from the other 2 (name & pts), yet percentage, which does matter and can't be inferred, is left out.
1
0
32
@hasumpstuffedup
This pic actually shows the cameras are NOT synced, as it's impossible to draw a line from all three camera angles that intersects. The ball is to the left from the left angle, dead over from the right, and to the right from in front.
1
0
32
One thing I've learned while aggregating tips is that people have a lot of ways of writing "Western Bulldogs."
11
2
31
At first glance, the fixture is more balanced than you'd expect given Gather Round, but there are still a few wtfs.
5
2
32
A real-quick first pass at a 2020 forecast, accounting for fixture and off-season player movement.
15
5
29
Things I noticed while pressing "UPDATE SEASON" button:
- Melbourne should be hot hot favourites
- Richmond will be much better
- Good fixtures: ADE, SYD, GEE, RIC
- Not great fixtures: Docklands tenants, MEL
2
1
31
Decided to write an article on what I'd change about AFL rules. Somehow it's already 5 pages.
3
0
30
Here's a deeper look into the 2018 fixture in retrospect, and how St Kilda were shafted by both fate and design:
Think it's under-reported how horrendous St Kilda's fixture was this year. Double up games were against Richmond, GWS, Melbourne, Hawthorn, North; had home games against interstate sides only 4 times but had to travel away interstate 5 times + 1 to Kardinia + 1 to Tas.
11
19
77
4
11
31
The AFL app scoreworm is wrong again. Collingwood weren’t over 20 in front in the 4th qtr, and if they had been, it would have taken more than 3 goals to put the Hawks in front.
5
1
30
It’s official:
🏆 Overall champion:
@EloPredicts
(156 tips)
🏆 Bits champion:
@AflGlicko
(35.76 bits)
🏆 MAE champion:
@AflGlicko
(24.91)
Runners-up:
Overall:
@thecruncherau
(154 tips)
Bits:
@MatterOfStats
(33.24)
MAE:
@MatterOfStats
(25.09)
4
1
29
Isn't it physically impossible for the ball to be directly over the post when viewed from two (perpendicular) angles but not a third? Are these cameras correctly synced?
8
2
28
Collingwood can remain irrational longer than you can stay in the tipping comp.
0
0
28
First new model for 2024! It's:
@footy_charts
As seen on Twitter
Now competing in the Squiggle comp
1
1
27
Full results are in for 50 expert & media pundit preseason ladder predictions. Summary: Over the long term, Squiggle is going quite a lot better than any of them. I will be reposting this link once per day forever.
3
1
27
In March 2020, I built a calculator from
@MatterOfStats
's brilliant work on the predictability of AFL crowds. COVID immediately crushed the numbers, and now it only works as a peek into an alternate reality where we'd had no pandemic.
2
2
28