Didn't AI predict us to win the premiership in 2023?

He's using A.I.?

I thought we'd have exclusive dibs on 'The Big German'!

Though he did end up as a part-time assistant coach at the Dorks in 2001...

...though, seriously, what could AI offer coaching - pre-season, player management, game-day preparation, in-game, recovery, injury assessment/management, etc., etc..

Those who understand it (definitely not me!), what do you think?

30 minutes ago, Timothy Reddan-A'Blew said:

...though, seriously, what could AI offer coaching - pre-season, player management, game-day preparation, in-game, recovery, injury assessment/management, etc., etc..

Those who understand it (definitely not me!), what do you think?

Football is a very data rich sport... especially with Champion Data game-day statistics available. I also used to work in the high performance/ strength & conditioning space at TAC Cup level in the mid 2010s and even back then we would have a plethora of data to look at in regards to game-day and training loads, recovery, etc.

I could see AI being used to analyse Champion Data statistics for trends that the successful teams are using in regards to defending, ball movement, attacking chains, etc. I could also see AI being used to track player loads/recovery data in much greater detail than any human could ever devote time to.

I hope Melbourne are up to their eyeballs in AI to be honest... it is going to be an absolute game-changer in professional sport in the next decade.

22 minutes ago, mfcrox said:

Football is a very data rich sport... especially with Champion Data game-day statistics available. I also used to work in the high performance/ strength & conditioning space at TAC Cup level in the mid 2010s and even back then we would have a plethora of data to look at in regards to game-day and training loads, recovery, etc.

I could see AI being used to analyse Champion Data statistics for trends that the successful teams are using in regards to defending, ball movement, attacking chains, etc. I could also see AI being used to track player loads/recovery data in much greater detail than any human could ever devote time to.

I hope Melbourne are up to their eyeballs in AI to be honest... it is going to be an absolute game-changer in professional sport in the next decade.

Thanks, mfcrox. That's a very positive view of AI, and in a context of something dear to all of us on here - MFC ascendancy and success.

Is it an area where significant differentiation can arise, between those those who understand and use it well and those who try it haphazardly? What is needed at a staffing/structural/management level to get it 'right'? Again, I know nothing about it, and wonder if there are traps to fall into, and/or advantages to be grasped.

I totally hear any skepticism and views here - our AFL game is a game of heart, grit, and 'gut feel,' and the idea of a laptop coaching a team feels a bit sterile. But it’s worth looking at AI not as a 'replacement' for the likes of Sam Mitchell's soul, but as a high-powered pair of glasses.

Here’s why even the most old-school fan might find it interesting:

• It’s about 'Seeing' the Unseen: We’ve all been frustrated when a player looks 'off' or lethargic. Currently, coaches guess why. AI can look at the biometric data and say, 'He’s not lazy; his central nervous system hasn't recovered from last week's 12km load.' It protects the players we love from injuries that are currently 'invisible' until they pop a hamstring.
  
  

• The 'Vibe' vs. The 'Fact': A coach might feel like we are losing the clearances because our rucks are beaten. The AI can instantly point out, 'Actually, the rucks are winning the tap, but our rovers are positioned 2Mtrs too wide for this specific ground.' It turns a 'vibe' into an actionable instruction in seconds.
  
  

• Removing Bias, Not Humanity: Humans are prone to 'recency bias' as it's called. We remember the last mistake a player made or has made (usually on a subliminal level) and judge them accordingly. AI doesn't have a 'dog house.' It looks at the cold, hard efficiency of every involvement/interaction. This may lead to ensuring the best players stay on the ground based on output, not reputation. And the like.
  
  

• We truly are on the cusp of a new era of data management. One that takes sterility (and guess work) out of a knowledge-base. And turns it into actionable takes beyond what we may have perceived prior. i.e taking the data analysis that is gathered (since the likes of GPS and other metrics) and actually doing some things profound with the data. That's where AI enters the fray.
  
  

• The Sam Mitchell Factor: Coaches like Mitchell aren’t trying to turn players into robots. They are trying to find the 1% tactical gaps and advantages that the human eye misses when watching 36 players moving at 25km/ph.
  
  

At the end of the day, a computer can’t give a three-quarter-time speech that makes hair stand up on your arms. It can’t teach 'courage.' But it can ensure that when a player goes to show that courage, they are in the best physical condition and the best tactical position to win the ball.

It’s not 'Moneyball' taking over the game; it’s just making sure teams like us, have the sharpest tools in the shed. I’d rather we be the ones wielding the tech than the ones trying to catch up to it in two years' time. This Sam Mitchell snippet is really about the AFL telecasting/approving to the wider audience that AI is in use and the likes of Mitchell are being projected as frontier leaders. I'd be disappointed if the Dees were not adopting AI use.

As an extension to thoughts...

1. Pre-Season & Player Management (The "Optimizer")

In the AFL, pre-season is about the delicate balance between "loading" and "breaking."

• Hyper-Personalized Loading: While coaches use GPS data now, AI can cross-reference that with sleep data, heart-rate variability (HRV), and even psychological wellness scores to predict a "soft-tissue window" before it happens. It moves from "the group is doing 8km today" to "Player X’s biometric signature says he’s at 85% risk of a hamstring strain if he sprints today."

• Drafting & "Moneyball" 2.0: AI can analyze thousands of hours of minor League (U18) footage to find players whose "spatial awareness" or "decision-making speed" under pressure matches current AFL stars, identifying "diamonds in the rough" that recruiters might overlook.

2. Game-Day: Preparation & In-Game (The "Super Assistant")

This is where Sam Mitchell and others are likely looking for that 1% edge.

• The "Opposition Oracle": Imagine an AI fed with every game the opposition has played for three years. On game day, it can alert the bench: "When the opposition is 2 goals down in the 4th quarter, they increase corridor usage by 40%." * In-Game Tactical Shifts: In the heat of the game, a coach's bias can take over. AI acts as an "unemotional observer," suggesting structural changes: "The opposition’s spare man is intercepting 70% of entries; recommend moving a defensive forward to negate." It’s about processing 22 moving parts simultaneously - something the human brain struggles to do perfectly for 120 minutes.

3. Injury & Recovery (The "Predictor")

I'm drawing on the DeepMind's 'AlphaFold Project' here: As it really shines as an example.

• Biomechanical Analysis: AI can analyze a player’s kicking or running gait in real-time. If a player starts favouring one side by even 2-3 mms - unnoticeable to the human eye - the AI flags it as a fatigue-induced mechanical failure, allowing for an early rotation.

• The "Digital Twin": Some elite teams are moving toward creating a "Digital Twin" of an athlete - like a virtual clone or model that can "play" the game first to see how much stress their specific joints and tendons will take. That is where the State-of-the-art will take us (in coaching and all manners of Industry).
  

Some final thoughts:

On the "Human Element": AI doesn't "whinge." AI frees up coaches to do more "human" work. If AI is doing the heavy lifting on the data, the Coach has more time for the "soft skills" - like mentoring, empathy, and the psychological "rev-up" that an algorithm can't replicate.

Think of it like a pilot with a sophisticated heads-up display. The pilot still flies the plane and makes the final call on the landing, but the AI ensures they have every scrap of data they need to make that call perfectly. For a club like Hawthorn and Sam Mitchell - and hopefully Melbourne - integrating this isn't just about 'using tech'; it's about making sure players are the best-informed athletes on the field.

I'd invite any expansion on this topic, that the OP has raised.

Go Dees!

Edited 15 hours ago15 hr by Stem Splitters
Formatting

At the training sessions, there are 4 to 5 analysts at the computers in the shed.

Plenty of data sources exist. They would be negligent if they are not adding AI scrutiny to the data.

Maybe he is just lonely and wants his old buddy Al around in the form of a chatbot to bounce ideas off?

Edited 13 hours ago13 hr by John Demonic

7 hours ago, John Demonic said:

Maybe he is just lonely and wants his old buddy Al around in the form of a chatbot to bounce ideas off?

Or maybe he wants to see what Zac Merrett looks like in all the different Hawthorn jumpers.

15 hours ago, mfcrox said:

Football is a very data rich sport... especially with Champion Data game-day statistics available. I also used to work in the high performance/ strength & conditioning space at TAC Cup level in the mid 2010s and even back then we would have a plethora of data to look at in regards to game-day and training loads, recovery, etc.

I could see AI being used to analyse Champion Data statistics for trends that the successful teams are using in regards to defending, ball movement, attacking chains, etc. I could also see AI being used to track player loads/recovery data in much greater detail than any human could ever devote time to.

I hope Melbourne are up to their eyeballs in AI to be honest... it is going to be an absolute game-changer in professional sport in the next decade.

Like you I hope the club is up to the eyeballs. Would be good to see such a statement from the club even in a modest acknowledgement as a recognition of innovation.

Also I am again impressed by the latent talent that resides in our supporters (Dland ) and disappointed that the club does not utilise this talent.

How can we get the club to engage.

13 hours ago, Stem Splitters said:

I totally hear any skepticism and views here - our AFL game is a game of heart, grit, and 'gut feel,' and the idea of a laptop coaching a team feels a bit sterile. But it’s worth looking at AI not as a 'replacement' for the likes of Sam Mitchell's soul, but as a high-powered pair of glasses.

Here’s why even the most old-school fan might find it interesting:

• It’s about 'Seeing' the Unseen: We’ve all been frustrated when a player looks 'off' or lethargic. Currently, coaches guess why. AI can look at the biometric data and say, 'He’s not lazy; his central nervous system hasn't recovered from last week's 12km load.' It protects the players we love from injuries that are currently 'invisible' until they pop a hamstring.
  
  

• The 'Vibe' vs. The 'Fact': A coach might feel like we are losing the clearances because our rucks are beaten. The AI can instantly point out, 'Actually, the rucks are winning the tap, but our rovers are positioned 2Mtrs too wide for this specific ground.' It turns a 'vibe' into an actionable instruction in seconds.
  
  

• Removing Bias, Not Humanity: Humans are prone to 'recency bias' as it's called. We remember the last mistake a player made or has made (usually on a subliminal level) and judge them accordingly. AI doesn't have a 'dog house.' It looks at the cold, hard efficiency of every involvement/interaction. This may lead to ensuring the best players stay on the ground based on output, not reputation. And the like.
  
  

• We truly are on the cusp of a new era of data management. One that takes sterility (and guess work) out of a knowledge-base. And turns it into actionable takes beyond what we may have perceived prior. i.e taking the data analysis that is gathered (since the likes of GPS and other metrics) and actually doing some things profound with the data. That's where AI enters the fray.
  
  

• The Sam Mitchell Factor: Coaches like Mitchell aren’t trying to turn players into robots. They are trying to find the 1% tactical gaps and advantages that the human eye misses when watching 36 players moving at 25km/ph.
  
  

At the end of the day, a computer can’t give a three-quarter-time speech that makes hair stand up on your arms. It can’t teach 'courage.' But it can ensure that when a player goes to show that courage, they are in the best physical condition and the best tactical position to win the ball.

It’s not 'Moneyball' taking over the game; it’s just making sure teams like us, have the sharpest tools in the shed. I’d rather we be the ones wielding the tech than the ones trying to catch up to it in two years' time. This Sam Mitchell snippet is really about the AFL telecasting/approving to the wider audience that AI is in use and the likes of Mitchell are being projected as frontier leaders. I'd be disappointed if the Dees were not adopting AI use.

As an extension to thoughts...

1. Pre-Season & Player Management (The "Optimizer")

In the AFL, pre-season is about the delicate balance between "loading" and "breaking."

• Hyper-Personalized Loading: While coaches use GPS data now, AI can cross-reference that with sleep data, heart-rate variability (HRV), and even psychological wellness scores to predict a "soft-tissue window" before it happens. It moves from "the group is doing 8km today" to "Player X’s biometric signature says he’s at 85% risk of a hamstring strain if he sprints today."

• Drafting & "Moneyball" 2.0: AI can analyze thousands of hours of minor League (U18) footage to find players whose "spatial awareness" or "decision-making speed" under pressure matches current AFL stars, identifying "diamonds in the rough" that recruiters might overlook.

2. Game-Day: Preparation & In-Game (The "Super Assistant")

This is where Sam Mitchell and others are likely looking for that 1% edge.

• The "Opposition Oracle": Imagine an AI fed with every game the opposition has played for three years. On game day, it can alert the bench: "When the opposition is 2 goals down in the 4th quarter, they increase corridor usage by 40%." * In-Game Tactical Shifts: In the heat of the game, a coach's bias can take over. AI acts as an "unemotional observer," suggesting structural changes: "The opposition’s spare man is intercepting 70% of entries; recommend moving a defensive forward to negate." It’s about processing 22 moving parts simultaneously - something the human brain struggles to do perfectly for 120 minutes.

3. Injury & Recovery (The "Predictor")

I'm drawing on the DeepMind's 'AlphaFold Project' here: As it really shines as an example.

• Biomechanical Analysis: AI can analyze a player’s kicking or running gait in real-time. If a player starts favouring one side by even 2-3 mms - unnoticeable to the human eye - the AI flags it as a fatigue-induced mechanical failure, allowing for an early rotation.

• The "Digital Twin": Some elite teams are moving toward creating a "Digital Twin" of an athlete - like a virtual clone or model that can "play" the game first to see how much stress their specific joints and tendons will take. That is where the State-of-the-art will take us (in coaching and all manners of Industry).
  

Some final thoughts:

On the "Human Element": AI doesn't "whinge." AI frees up coaches to do more "human" work. If AI is doing the heavy lifting on the data, the Coach has more time for the "soft skills" - like mentoring, empathy, and the psychological "rev-up" that an algorithm can't replicate.

Think of it like a pilot with a sophisticated heads-up display. The pilot still flies the plane and makes the final call on the landing, but the AI ensures they have every scrap of data they need to make that call perfectly. For a club like Hawthorn and Sam Mitchell - and hopefully Melbourne - integrating this isn't just about 'using tech'; it's about making sure players are the best-informed athletes on the field.

I'd invite any expansion on this topic, that the OP has raised.

Go Dees!

Same comment as to mfcrox. Hope you are are both not AI generated.

1 hour ago, dpositive said:

Same comment as to mfcrox. Hope you are are both not AI generated.

Are you kidding? Ever heard of Grammarly? It's an AI tool for writing purposes.

Edited 58 minutes ago58 min by Stem Splitters
Extended thought.

15 hours ago, Stem Splitters said:

I totally hear any skepticism and views here - our AFL game is a game of heart, grit, and 'gut feel,' and the idea of a laptop coaching a team feels a bit sterile. But it’s worth looking at AI not as a 'replacement' for the likes of Sam Mitchell's soul, but as a high-powered pair of glasses.

Here’s why even the most old-school fan might find it interesting:

• It’s about 'Seeing' the Unseen: We’ve all been frustrated when a player looks 'off' or lethargic. Currently, coaches guess why. AI can look at the biometric data and say, 'He’s not lazy; his central nervous system hasn't recovered from last week's 12km load.' It protects the players we love from injuries that are currently 'invisible' until they pop a hamstring.
  
  

• The 'Vibe' vs. The 'Fact': A coach might feel like we are losing the clearances because our rucks are beaten. The AI can instantly point out, 'Actually, the rucks are winning the tap, but our rovers are positioned 2Mtrs too wide for this specific ground.' It turns a 'vibe' into an actionable instruction in seconds.
  
  

• Removing Bias, Not Humanity: Humans are prone to 'recency bias' as it's called. We remember the last mistake a player made or has made (usually on a subliminal level) and judge them accordingly. AI doesn't have a 'dog house.' It looks at the cold, hard efficiency of every involvement/interaction. This may lead to ensuring the best players stay on the ground based on output, not reputation. And the like.
  
  

• We truly are on the cusp of a new era of data management. One that takes sterility (and guess work) out of a knowledge-base. And turns it into actionable takes beyond what we may have perceived prior. i.e taking the data analysis that is gathered (since the likes of GPS and other metrics) and actually doing some things profound with the data. That's where AI enters the fray.
  
  

• The Sam Mitchell Factor: Coaches like Mitchell aren’t trying to turn players into robots. They are trying to find the 1% tactical gaps and advantages that the human eye misses when watching 36 players moving at 25km/ph.
  
  

At the end of the day, a computer can’t give a three-quarter-time speech that makes hair stand up on your arms. It can’t teach 'courage.' But it can ensure that when a player goes to show that courage, they are in the best physical condition and the best tactical position to win the ball.

It’s not 'Moneyball' taking over the game; it’s just making sure teams like us, have the sharpest tools in the shed. I’d rather we be the ones wielding the tech than the ones trying to catch up to it in two years' time. This Sam Mitchell snippet is really about the AFL telecasting/approving to the wider audience that AI is in use and the likes of Mitchell are being projected as frontier leaders. I'd be disappointed if the Dees were not adopting AI use.

As an extension to thoughts...

1. Pre-Season & Player Management (The "Optimizer")

In the AFL, pre-season is about the delicate balance between "loading" and "breaking."

• Hyper-Personalized Loading: While coaches use GPS data now, AI can cross-reference that with sleep data, heart-rate variability (HRV), and even psychological wellness scores to predict a "soft-tissue window" before it happens. It moves from "the group is doing 8km today" to "Player X’s biometric signature says he’s at 85% risk of a hamstring strain if he sprints today."

• Drafting & "Moneyball" 2.0: AI can analyze thousands of hours of minor League (U18) footage to find players whose "spatial awareness" or "decision-making speed" under pressure matches current AFL stars, identifying "diamonds in the rough" that recruiters might overlook.

2. Game-Day: Preparation & In-Game (The "Super Assistant")

This is where Sam Mitchell and others are likely looking for that 1% edge.

• The "Opposition Oracle": Imagine an AI fed with every game the opposition has played for three years. On game day, it can alert the bench: "When the opposition is 2 goals down in the 4th quarter, they increase corridor usage by 40%." * In-Game Tactical Shifts: In the heat of the game, a coach's bias can take over. AI acts as an "unemotional observer," suggesting structural changes: "The opposition’s spare man is intercepting 70% of entries; recommend moving a defensive forward to negate." It’s about processing 22 moving parts simultaneously - something the human brain struggles to do perfectly for 120 minutes.

3. Injury & Recovery (The "Predictor")

I'm drawing on the DeepMind's 'AlphaFold Project' here: As it really shines as an example.

• Biomechanical Analysis: AI can analyze a player’s kicking or running gait in real-time. If a player starts favouring one side by even 2-3 mms - unnoticeable to the human eye - the AI flags it as a fatigue-induced mechanical failure, allowing for an early rotation.

• The "Digital Twin": Some elite teams are moving toward creating a "Digital Twin" of an athlete - like a virtual clone or model that can "play" the game first to see how much stress their specific joints and tendons will take. That is where the State-of-the-art will take us (in coaching and all manners of Industry).
  

Some final thoughts:

On the "Human Element": AI doesn't "whinge." AI frees up coaches to do more "human" work. If AI is doing the heavy lifting on the data, the Coach has more time for the "soft skills" - like mentoring, empathy, and the psychological "rev-up" that an algorithm can't replicate.

Think of it like a pilot with a sophisticated heads-up display. The pilot still flies the plane and makes the final call on the landing, but the AI ensures they have every scrap of data they need to make that call perfectly. For a club like Hawthorn and Sam Mitchell - and hopefully Melbourne - integrating this isn't just about 'using tech'; it's about making sure players are the best-informed athletes on the field.

I'd invite any expansion on this topic, that the OP has raised.

Go Dees!

Well done, @Stem Splitters ! This is one of the most informative and eloquent posts I've ever seen on Demonland - you'd better be careful. It almost read like a university submission.

I agree with everything that you've written. Based on how I've seen AI roll out in my work at a very rapid pace, I'm open to a very broad range of possibilities.

That said, I had a little chuckle as I wondered if you used an LLM to polish this post.

17 hours ago, Timothy Reddan-A'Blew said:

...though, seriously, what could AI offer coaching - pre-season, player management, game-day preparation, in-game, recovery, injury assessment/management, etc., etc..

Those who understand it (definitely not me!), what do you think?

The reality is that all that information is not likely to be public as almost every sports outfit globally will have IP that they protect. There will be mainstream sports science that will be public - but that is public for all.

So the advantage really wont be from a tactical / player front IMO. If you did, id be asking the sports science team some hard questions as they should really be at the cutting edge.

The conversation starters and break through the 'generational' gap maybe something - but if you rely on ChatGPT to have a discussion with your playing group...maybe coaching isnt the right job for you.

17 hours ago, mfcrox said:

Football is a very data rich sport... especially with Champion Data game-day statistics available. I also used to work in the high performance/ strength & conditioning space at TAC Cup level in the mid 2010s and even back then we would have a plethora of data to look at in regards to game-day and training loads, recovery, etc.

I could see AI being used to analyse Champion Data statistics for trends that the successful teams are using in regards to defending, ball movement, attacking chains, etc. I could also see AI being used to track player loads/recovery data in much greater detail than any human could ever devote time to.

I hope Melbourne are up to their eyeballs in AI to be honest... it is going to be an absolute game-changer in professional sport in the next decade.

AI will help analyse data better for sure, but the insights it generates will need to be learned. This is the common misconception about AI's ability to provide value. The other aspect is that Champion Data is not going to release its data set into the world of AI - in fact it would be taking very significant steps to ensure it is protected as they make money in selling data insights to the industry.

There is a world where MFC teaches ChatGPT or DeepSeek to analyse data and interpret it according to their philosophy but Hawks or Suns might teach it differetnly. it still comes down to who is the better teacher and that is where human involvement / interpretation is still required for now.

I asked AI what should MFC do to win the premiership this year. Response appended, but I am surprised it doesn't recommend infecting all opppo players with a lingering disease.

Melbourne’s best path to a 2026 flag is to double‑down on their contest/defence DNA but fix their ball movement and forward-half efficiency, while aggressively refreshing the list and game plan around that core identity.

1. Clarify list build for 2026

• Lock in the genuine A‑graders as the untouchable core, then be ruthless around the edges (fringe talls, depth mids, small forwards) so every best‑22 spot is either star power or clear system role.

• Use list spots freed by recent delistings to target two areas: a powerful marking forward who can play high and a two‑way small forward who applies frontal pressure and finishes chances.

• Prioritise draftees and VFL‑developed players who already show AFL‑standard decision‑making and speed, not just “upside,” to impact in 2026 rather than a long build.

2. Fix ball movement and inside‑50 connection

• Keep the strong contest base but move away from slow, predictable long bombs; aim for a more aggressive handball‑receive and overlap run style that was a strength when Melbourne played well.

• Systematically train forwards and mids on leading patterns, shallow entries and hit‑up kicks to 35–45m, rather than constant deep dumps that made scoring inefficient in recent seasons.

• Build repeatable mechanisms for exiting D50 (shorts, switch, corridor burst) so they arrive inside 50 with time and separation rather than under pressure and outnumbered.

3. Sharpen contest and defensive system

• Maintain a top‑tier stoppage and clearance profile, using ruck dominance to control field position and time in forward half.

• Drill a “layered” contest structure: one layer at the ball, one at the fall, one holding shape defensively so panic doesn’t collapse the zone and expose the backline.

• Emphasise fast defensive transition after turnover, with forwards responsible for immediate pressure and denying easy exits to avoid being sliced on rebound.

4. Game‑day strategy and flexibility

• Go into 2026 with clearly rehearsed Plan B and Plan C: e.g., a faster, corridor‑heavy mode when chasing, and a territory/lock‑in mode when in front.

• Be willing to change match‑ups and structure early (swing a defender forward, extra behind the ball, tag hot mids) instead of waiting until games are effectively gone.

• Use the VFL (Casey) as a tactical laboratory, running the same systems so call‑ups slot seamlessly into roles rather than being “just depth.”

5. Culture, cohesion and preparation

• Lean into the player‑led leadership and connection that drove the strong VFL second half of 2025, and replicate that environment at AFL level.

• Run a demanding but targeted pre‑season: high focus on repeat efforts, defensive running, and skill execution under fatigue to support a high‑tempo style.

• Set non‑negotiables around effort, defensive buy‑in and role discipline; selection and contracts should reward players who consistently execute the system rather than just talent.

If you want, I can sketch what a best‑22 and specific role tweaks for 2026 could look like based on their current list and VFL performers.

Edited 15 minutes ago15 min by sue

Has AI reached the stage where it can see patterns in vision (match vision in this case)? And, if so, where it can analyse and usefully report on them?

To some extent, this would bypass IP constraints?

That's impressive, @sue! Thanks!

I guess a lot of it is 'Well, derrr!' but it's so sound, footy-lingo-, grammar-, sense- and actual sport-wise.

The digital age left me on the platform a long time ago. That train seems now to be leaving Earth's atmosphere...