Resource 2019 Squad Selection & Ratings

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Ratings updated from the Round 13 bye.

He may have had the past week off and been the focus of vitriol from certain members of the media still reeling from last year's Preliminary Final - but McGovern returns to the #1 spot.

Ryan moves back into B rating territory with his recent good form.

Almost all of the experienced players ratings are heading in the right direction with the exception of Redden who has been sliding down each week since the start of the season.
 

(Log in to remove this ad.)

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Squad Selection:

Round 14 (vs Essendon @ Perth Stadium).

R14-Squad-Selection.jpg

Returning players significantly boosting the squad as expected (Hurn and Yeo rated #3 & #10 respectively).

Cameron named for his debut.
 

flamingEMBERS

Norm Smith Medallist
Sep 8, 2011
8,951
7,169
AFL Club
West Coast
Indirectly they are as CD have a near-monopoly upon the data-gathering of statistics in AFL.

However, they are independent from CD's rating system. The ratings here are derived from my own system that utilises the following:

  • Positional Type: Defender (Playing Tall), Defender (Playing Small), Midfielder (Outside), Midfielder (Onball), Forward (Playing Tall), Forward (Playing Small), Ruckman.

  • Statistics: Goals, Metres Gained, Disposals, Marks, Tackles, Hit Outs, Goal Assists, Inside 50s, Clearances, Rebound 50s, Contested Possessions, Disposal Efficiency, Contested Marks, One-Percenters, Score Involvements, Interceptions, Tackles In 50, Kicks, Uncontested Possessions, Effective Disposals.
excuse my ignorance, but how do your ratings differ from CD Player Ratings?
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
excuse my ignorance, but how do your ratings differ from CD Player Ratings?
Information on how CD do their Player Ratings is here.

Players accrue or lose points every time they are involved in a passage of play. The score awarded to them each time they are in the play is determined by a complex algorithm formulated and refined over a number of years by Champion Data. Players accrue or lose points depending on whether they have a positive or negative impact on a passage of play.
A player's rating is determined by aggregating his points tally based on a rolling window of the previous two seasons. For example, after round six of the 2014 season, a player's rating will be based on matches from round seven of the 2012 season onwards. However, only a player's most recent 40 matches are used in the calculation of his rating. This creates a buffer for players missing matches through injury, suspension, omission or by not being involved in finals. A player's most recent 30 matches are given greater weight in determining his rating. Matches 31 through 40 are progressively reduced in weighting, from 100 per cent down to five per cent for the earliest game in the window.


Whereas in this system each player is ranked in their output against seven different algorithms that correlate to the positional types used here: Defender (Playing Tall), Defender (Playing Small), Midfielder (Outside), Midfielder (Onball), Forward (Playing Tall), Forward (Playing Small), Ruckman.

These involve the ranking of the player in respect to the rest of the competition for the following statistics over the past 23 weeks of matches, with the most recent 7 weeks given an additional 20% weighting:
  • Defender (Playing Tall): marks, contested marks, intercepts, one-percenters, rebound 50s
  • Defender (Playing Small): marks, intercepts, rebound 50s, kicks, metres gained, disposal efficiency
  • Midfielder (Outside): kicks, disposals, metres gained, inside 50s, disposal efficiency, score involvements, uncontested possessions, effective disposals
  • Midfielder (Onball): contested possessions, clearances, inside 50s, score involvements, tackles, metres gained
  • Forward (Playing Tall): goals, marks, contested marks, marks in 50, goal assists, score involvements, tackles in 50
  • Forward (Playing Small): goals, goal assists, tackles, tackles in 50, marks in 50
  • Ruckmen: hit outs, contested marks, contested possessions, clearances, goal assists, score involvements


Part of creating this system was due to my own frustration with the outcomes from CDs Ratings:
  • The aggregate window of 40 matches is too large; it for example results in D.Martin still being in the overall top 5 for their ratings despite playing within himself for the past 18 months now.
  • It rewards players for simply attending stoppages, thus overrating ruckmen and onball midfielders - 7 of the overall top 10 are either ruckmen or exclusively inside midfielders.
  • As a result other positions are underrated;
    • The highest ranked defender is #28 (McGovern)
    • The highest ranked forward is #16 (Franklin)
    • Outside midfielders are particularly hammered under this; Whitfield #65, S.Ross #73, Gaff #125...
 

flamingEMBERS

Norm Smith Medallist
Sep 8, 2011
8,951
7,169
AFL Club
West Coast
Information on how CD do their Player Ratings is here.

Players accrue or lose points every time they are involved in a passage of play. The score awarded to them each time they are in the play is determined by a complex algorithm formulated and refined over a number of years by Champion Data. Players accrue or lose points depending on whether they have a positive or negative impact on a passage of play.
A player's rating is determined by aggregating his points tally based on a rolling window of the previous two seasons. For example, after round six of the 2014 season, a player's rating will be based on matches from round seven of the 2012 season onwards. However, only a player's most recent 40 matches are used in the calculation of his rating. This creates a buffer for players missing matches through injury, suspension, omission or by not being involved in finals. A player's most recent 30 matches are given greater weight in determining his rating. Matches 31 through 40 are progressively reduced in weighting, from 100 per cent down to five per cent for the earliest game in the window.


Whereas in this system each player is ranked in their output against seven different algorithms that correlate to the positional types used here: Defender (Playing Tall), Defender (Playing Small), Midfielder (Outside), Midfielder (Onball), Forward (Playing Tall), Forward (Playing Small), Ruckman.

These involve the ranking of the player in respect to the rest of the competition for the following statistics over the past 23 weeks of matches, with the most recent 7 weeks given an additional 20% weighting:
  • Defender (Playing Tall): marks, contested marks, intercepts, one-percenters, rebound 50s
  • Defender (Playing Small): marks, intercepts, rebound 50s, kicks, metres gained, disposal efficiency
  • Midfielder (Outside): kicks, disposals, metres gained, inside 50s, disposal efficiency, score involvements, uncontested possessions, effective disposals
  • Midfielder (Onball): contested possessions, clearances, inside 50s, score involvements, tackles, metres gained
  • Forward (Playing Tall): goals, marks, contested marks, marks in 50, goal assists, score involvements, tackles in 50
  • Forward (Playing Small): goals, goal assists, tackles, tackles in 50, marks in 50
  • Ruckmen: hit outs, contested marks, contested possessions, clearances, goal assists, score involvements


Part of creating this system was due to my own frustration with the outcomes from CDs Ratings:
  • The aggregate window of 40 matches is too large; it for example results in D.Martin still being in the overall top 5 for their ratings despite playing within himself for the past 18 months now.
  • It rewards players for simply attending stoppages, thus overrating ruckmen and onball midfielders - 7 of the overall top 10 are either ruckmen or exclusively inside midfielders.
  • As a result other positions are underrated;
    • The highest ranked defender is #28 (McGovern)
    • The highest ranked forward is #16 (Franklin)
    • Outside midfielders are particularly hammered under this; Whitfield #65, S.Ross #73, Gaff #125...
I had a look at your spreadsheet it's quite impressive.

Does this mean that a player accumulates points based one what they are doing? Say Gaff can accumulate points as an onballer and an outside player? How do you know when to differentiate what they're playing as?
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
I had a look at your spreadsheet it's quite impressive.

Does this mean that a player accumulates points based one what they are doing? Say Gaff can accumulate points as an onballer and an outside player? How do you know when to differentiate what they're playing as?
Cheers,

The player ratings here are generalised in that regard (correlating output vs location/role) as:
  • there is a complete lack of data in that area that is freely available; and
  • by what means would one describe arbitrary definitions for such a purpose? is an outside midfielder an onballer if they are within 3m of the stoppage? 5m? 10m? what happens if they run off the back of a stoppage? or if they get tagged?

In the background here every player ends up with a rating across each of the seven positional types - think of the operation of the algorithms as akin to a filter.

So I can't say for example the amount of time Gaff is utilised between inside and outside roles, but I can provide a value upon his impact in each of those roles in context to the rest of the competition.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Ratings updated from Round 14.

McGovern owing #1 so much that the competition invented a suspension to halt his dominance for another week.

Yeo moves above Shuey for the first time this year to become our highest rated midfielder.

Gaff returns to A+ for the first time since Round 3.


With Naitanui not having played a game during the ratings period, he is defaulting to a D. As soon as he notches up two matches the rating will correct for him (was A until Round 12 when his number of ranked matches expired).
 

BeornJ

All Australian
Sep 17, 2017
642
968
AFL Club
West Coast
So do you manually input this data? I hope not. I Hope you've got a way it is all automatically updated
 

Max zero

Premium Platinum
Jul 19, 2005
12,810
8,168
Perth
AFL Club
West Coast
Ratings updated from Round 14.

McGovern owing #1 so much that the competition invented a suspension to halt his dominance for another week.

Yeo moves above Shuey for the first time this year to become our highest rated midfielder.

Gaff returns to A+ for the first time since Round 3.


With Naitanui not having played a game during the ratings period, he is defaulting to a D. As soon as he notches up two matches the rating will correct for him (was A until Round 12 when his number of ranked matches expired).
Ironically with NN we would be A ranked across the board with 'Ds' all on the bench.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
So do you manually input this data? I hope not. I Hope you've got a way it is all automatically updated
Yeah the main database is automated. The spreadsheet data here is copy and paste from that.
 

(Log in to remove this ad.)

flamingEMBERS

Norm Smith Medallist
Sep 8, 2011
8,951
7,169
AFL Club
West Coast
Cheers,

The player ratings here are generalised in that regard (correlating output vs location/role) as:
  • there is a complete lack of data in that area that is freely available; and
  • by what means would one describe arbitrary definitions for such a purpose? is an outside midfielder an onballer if they are within 3m of the stoppage? 5m? 10m? what happens if they run off the back of a stoppage? or if they get tagged?

In the background here every player ends up with a rating across each of the seven positional types - think of the operation of the algorithms as akin to a filter.

So I can't say for example the amount of time Gaff is utilised between inside and outside roles, but I can provide a value upon his impact in each of those roles in context to the rest of the competition.
Yep, that makes sense now.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Squad Selection:

Round 15 (vs Hawthorn @ MCG).

R15-Squad-Selection.jpg

Naitanui makes his long-awaited return in addition to Rioli.

Absences of McGovern and Barrass hamper the defence.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Ratings updated from Round 15.

Hurn and McGovern trade places for top spot once again.

Shuey on the back of his heroic last quarter against the Hawks breaks back into the top A++ bracket.


Once Naitanui plays another game his rating will correct (currently on track to re-enter as A rating).
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Squad Selection:

Round 16 (vs Fremantle @ Perth Stadium).

R16-Squad-Selection.jpg

McGovern and Barrass return in what is arguably the strongest selection of the season thus far.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Ratings updated from the Round 16 Derby.

The top 3 ranked players overall are now Eagles: McGovern, Shuey and Hurn.

The club has also moved into 1st position of having the highest rated squad of 22.

Ryan and Cameron surging up the rankings places the forward line into an A++ rating.
 

BeornJ

All Australian
Sep 17, 2017
642
968
AFL Club
West Coast
Ratings updated from the Round 16 Derby.

The top 3 ranked players overall are now Eagles: McGovern, Shuey and Hurn.

The club has also moved into 1st position of having the highest rated squad of 22.

Ryan and Cameron surging up the rankings places the forward line into an A++ rating.
Im assuming Cameron is moving up rapidly due to good goal, tackle and goal assist numbers.

Can I ask the difference between score involvement and goal assist?

On a related note, does the data differentiate between a player 'giving' a goal to another player (eg. handball or kick to them alone in the goal square. I'm thinking of Ryan deciding to kick to Cripps directly in
front of goal instead of having a shot ) and simply being involved in a scoring chain?
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Im assuming Cameron is moving up rapidly due to good goal, tackle and goal assist numbers.

Can I ask the difference between score involvement and goal assist?

On a related note, does the data differentiate between a player 'giving' a goal to another player (eg. handball or kick to them alone in the goal square. I'm thinking of Ryan deciding to kick to Cripps directly in
front of goal instead of having a shot ) and simply being involved in a scoring chain?
Yes, correct on both counts.

Cameron is 14th overall for goals, 74th for goal assists and 34th for tackles inside forward 50 (each per game).


The definition of the two queried statistics are generally accepted as follows:

Score Involvement: When a player is involved in an unbroken passage of play which leads to a score.

Goal Assist: Creating a goal by getting the ball to a teammate either via a disposal, knock-on, ground kick or hitout, or by winning a free kick before the advantage is paid to the goal scorer.


The former describes an involvement in a play that culminated in a score. That could be varied as a chain initiator, link-up role, assister or scorer.

The latter describes only plays that directly create a goal for a teammate.
 

Dylan82

Premium Platinum
Aug 14, 2004
1,464
3,792
Brisbane
AFL Club
West Coast
Other Teams
Cardiff City
Squad Selection:

Round 17 (vs Collingwood @ Perth Stadium).

R17-Squad-Selection.jpg

Winners across every line gives the selection an A grade rating for the first time this season.
 
Top Bottom