6 hours ago, Deelightful Dee said:

Hit the nail on the head @At the break of Gawn

Until we can improve our consistency of method and execution in the turnover game, our results will be also inconsistent. This year, in every match we've won the scoring from turnover battle, we've won the game, other than against the Hawks where we only won it by 2 points but got outscored from stoppage by 30 (our 3rd worst result for the year behind -42 in R2 and -33 in R3). We lost the turnover battle by an average of 22 points in the first 5 rounds, won it by an average of 14 in the next 6 rounds (in which won our only 5 games for the year) and lost it by an average of 21 in the next 3. I know where I'd be focusing our opportunity for improvement over the bye period: win the turnover battle to give ourselves the best chance of winning games and in particular how we defend turnover from Opp D50 to Inside 50 which we are 12th over the last 5 weeks and 11th for the season according to Wheelo's premiership metrics tweet this week!

Here is the premiership metrics tweet mentioned:

13 hours ago, WheeloRatings said:

I have done some work replicating the player ratings and I have a reasonable model with pressure being the main factor I don't have access to. The breakdown of the source of ratings helps to understand the individual player ratings even if you still don't agree with the rating system.

The following are estimates only but they give you an idea of what's contributing to a player's rating. Gawn was rated very highly in winning the ball and through his ruck work, but he lost a lot of points through his disposal and in particular his two shots on goal. The rating system is based on how a player impacts both team's scoring chances. So easy missed shots at goal are heavily penalised and kicking difficult goals earn a lot of points.

On the flipside, while Georgiades kicked 7 goals, he missed 4 and pretty much kicked at the expected accuracy for his shots. He gained a lot of rating points through his possessions (marks/free kicks), he didn't gain many through his disposal. He also lost points through giving away free kicks and dropping a mark.

Most rating points from POSSESSIONS

15.3 Ratugolea
14.4 Gawn
11.9 Georgiades
10.9 May

Most rating points gained (and lost) from DISPOSALS

10.1 Langford
9.7 Pickett
7.9 Bergman
7.9 Rozee
---------------
-2.8 Petty
-2.9 Viney
-4.1 Oliver
-7.3 Gawn

Most rating points from HITOUTS

3.9 Gawn
0.8 Sweet

Most rating points lost from FREE KICKS

-2.6 Georgiades
-2.6 Visentini

Most rating points lost from NO PRESSURE ERRORS (incl. dropped marks)

-3.0 van Rooyen
-2.8 Georgiades
-1.9 Gawn

Hey @WheeloRatings. Thank you for this thoughtful response. It's appreciated and improved my understanding -> thinking about the game.

Your modelling is brilliant — and you are right. Not having access to 'pressure' data — and particularly the inputs used to assemble that data point — is a handicap. And that's my point: many football followers observe first-hand what a player does (or doesn’t do) during a game that appears to be neglected in what I would term 'complexity' inputs. That is, the layered decision-making, tactical compliance, off-ball movement, and game-shaping intent that don’t result in a stat — but change the shape of the game. Things like sacrificing a lead to create space, halting a chain with smart positioning, or neutralising a dangerous opponent without ever laying a hand on them.

For example:

1. Not All Actions Are Created Equal — Even When They Look Identical

Two disposals may appear the same on a stat sheet — but one may be:

• Delivered with tactical precision (e.g. breaking a zone)

• Made after scanning multiple options in a high-stakes moment

The Champion Data model attempts to weight difficulty (e.g. a pressured kick may be worth more), but it still relies on observable, event-based outcomes, not the complexity of the decision behind them.

2. Systemic Thinking Goes Unmeasured

Modern footy is defined by:

• Executing roles within highly structured team systems

• Making spatial decisions on the fly (e.g. when to fold back, when to block space)

• Adapting to team tactics in real time

Yet players who do these things well — often without touching the ball — receive little to no recognition in the ratings system.

3. Complexity Isn’t Always Visible in the Data

The system cannot assess:

• Whether the options ignored were better or worse

• The quality of positioning before the disposal

• The degree of difficulty in reading the play or shaping it without possession

In short, a player who makes high-complexity, low-volume decisions (e.g. a defender who sets up a zone intercept or a forward who self-sacrifices to create space) is penalised relative to players with high-volume, low-difficulty contributions.

4. Intelligence and Influence ≠ Quantity

Footy IQ — the ability to read the game, think a kick ahead, and shape space — is a core reason why players like Max Gawn, and notably, Sidey and Pendles, have sustained elite performance late into their careers. Yet these forms of intelligence are not adequately captured, because they are not consistently associated with discrete, countable events. In Max’s case, he’s even penalised in the data for an errant possession — despite that disposal coming at the end of a 100-metre sprint to drop into the hole, absorbing contact, and creating the space for a teammate’s clean exit. The nuance of why an error occurred, what it created, or how it shaped the contest simply isn’t factored in.

The result? A system that reduces football intelligence to raw outcomes — overlooking the structural, leadership, and game-shaping qualities that explain why these players remain indispensable, even when they’re no longer statistically dominant.

5. Summary

Champion Data’s ratings provide a useful baseline — but by failing to account for complexity in decision-making and game understanding, the system risks undervaluing the players who make teams functional. Until complexity is measured, true on-field influence will remain partially hidden behind the numbers.

5 hours ago, Queanbeyan Demon said:

Hey @WheeloRatings. Thank you for this thoughtful response. It's appreciated and improved my understanding -> thinking about the game.

Your modelling is brilliant — and you are right. Not having access to 'pressure' data — and particularly the inputs used to assemble that data point — is a handicap. And that's my point: many football followers observe first-hand what a player does (or doesn’t do) during a game that appears to be neglected in what I would term 'complexity' inputs. That is, the layered decision-making, tactical compliance, off-ball movement, and game-shaping intent that don’t result in a stat — but change the shape of the game. Things like sacrificing a lead to create space, halting a chain with smart positioning, or neutralising a dangerous opponent without ever laying a hand on them.

For example:

1. Not All Actions Are Created Equal — Even When They Look Identical

Two disposals may appear the same on a stat sheet — but one may be:

• Delivered with tactical precision (e.g. breaking a zone)

• Made after scanning multiple options in a high-stakes moment

The Champion Data model attempts to weight difficulty (e.g. a pressured kick may be worth more), but it still relies on observable, event-based outcomes, not the complexity of the decision behind them.

2. Systemic Thinking Goes Unmeasured

Modern footy is defined by:

• Executing roles within highly structured team systems

• Making spatial decisions on the fly (e.g. when to fold back, when to block space)

• Adapting to team tactics in real time

Yet players who do these things well — often without touching the ball — receive little to no recognition in the ratings system.

3. Complexity Isn’t Always Visible in the Data

The system cannot assess:

• Whether the options ignored were better or worse

• The quality of positioning before the disposal

• The degree of difficulty in reading the play or shaping it without possession

In short, a player who makes high-complexity, low-volume decisions (e.g. a defender who sets up a zone intercept or a forward who self-sacrifices to create space) is penalised relative to players with high-volume, low-difficulty contributions.

4. Intelligence and Influence ≠ Quantity

Footy IQ — the ability to read the game, think a kick ahead, and shape space — is a core reason why players like Max Gawn, and notably, Sidey and Pendles, have sustained elite performance late into their careers. Yet these forms of intelligence are not adequately captured, because they are not consistently associated with discrete, countable events. In Max’s case, he’s even penalised in the data for an errant possession — despite that disposal coming at the end of a 100-metre sprint to drop into the hole, absorbing contact, and creating the space for a teammate’s clean exit. The nuance of why an error occurred, what it created, or how it shaped the contest simply isn’t factored in.

The result? A system that reduces football intelligence to raw outcomes — overlooking the structural, leadership, and game-shaping qualities that explain why these players remain indispensable, even when they’re no longer statistically dominant.

5. Summary

Champion Data’s ratings provide a useful baseline — but by failing to account for complexity in decision-making and game understanding, the system risks undervaluing the players who make teams functional. Until complexity is measured, true on-field influence will remain partially hidden behind the numbers.

You raise some very good points and I'm not suggesting that it's a perfect system.

The system was initially developed 15 or so years ago and the game has changed in that time, and it's obviously limited to events which are captured. The events that are captured are largely focussed on the player who wins the ball and what they do with the ball, the nearest player applying pressure to the ball carrier, and certain other defensive actions (spoils/smothers).

It does a much better job than ranking points (now used for Supercoach) and fantasy points at assessing quality and influence over quantity.

For example, a kick is rated based on the location the kick is taken from, the pressure on the kicker, the location it ends up, and the possession "state" of the result (e.g. was it to a contest and therefore reliant on a teammate winning possession, was it a teammate's uncontested mark or gather, or did it go to an opponent or out of bounds).

Using these data points, the model assesses whether your team is more likely to score after the kick than before. It does a pretty good job, but it doesn't account for who you're kicking it to or where all the other players on the field are. E.g:

• Kicking it to a one-on-one marking contest involving Gawn is not the same as your average one-on-one marking contest.

• Kicking a backwards 45 degree kick may be worth negative points, but if that opens up the switch it may be preferable to kicking down the line.

Other specific limitations are:

• A key defender allowing his opponent to take a leading mark is not penalised for not competing, but is penalised for competing and giving away a free kick.

• A ruck is not penalised for not winning a hitout, but is penalised for winning the hitout if it's sharked by the opposition.

Some of the limitations with the system could be improved through better data capture and some adjustments to the system, but I'm not sure that a number of the issues you have raised could actually be rectified (or maybe they could and it's beyond my comprehension).

A model is never complete, but a stripped back representation of the reality. The ‘complexity’ argument is valid, as is the way Wheelo represents the current data sets. The trick would also to understand what’s is valued by the team in their system and how that could be measured in the data.

For example, maybe Collingwood prize the forward of the ball running that is often unrewarded - against us, 4 players sprinted to an imaginary line up/down the field the forward flank to provide 4 options during a d50 transition. Was damned impressive, but only one of those lads would get a stat.

We have probably put maximal weighting on contested ball and insides 50’s/ forward half time. I wonder what we need to value now?