WheeloRatings

Sorry, unfortunately my model flipped to Geelong by 1.2 points following team announcements! It will update again once subs are announced.

@Lord Travis the qualification of a minimum of 30 shots means only three Melbourne players qualify (Fritsch, Pickett and Petracca). Here is the accuracy of Melbourne players with a minimum of 10 shots this season: Tom McDonald - 72.73% (8 goals, 11 shots) Charlie Spargo - 71.43% (10 goals, 14 shots) Jacob van Rooyen - 69.57% (16 goals, 23 shots) Ben Brown - 69.23% (9 goals, 13 shots) Brodie Grundy - 60.00% (9 goals, 15 shots) Kade Chandler - 57.69% (15 goals, 26 shots) Jack Viney - 54.55% (6 goals, 11 shots) Alex Neal-Bullen - 52.63% (10 goals, 19 shots) Bayley Fritsch - 51.92% (27 goals, 52 shots) Kysaiah Pickett - 47.50% (19 goals, 40 shots) Max Gawn - 43.75% (7 goals, 16 shots) Christian Petracca - 28.21% (11 goals, 39 shots) Ed Langdon - 18.18% (2 goals, 11 shots)

The 'Current Rating' is weighted toward more recent results, so a team could be rated higher if they in better form. If you look at the following page of my site, you will see a 'Current Rating' section and an 'Overall Season Rating' section in the table. https://www.wheeloratings.com/afl_ratings.html The overall season rating treats all matches equally. Melbourne is rated second overall, behind Collingwood, on the overall season defensive rating but first on the current defensive rating. St Kilda drops from third overall defensively to fourth currently as their form was better at the start of the season. Teams also inherit a proportion of their rating from the end of the prior season, so early season ratings will be impacted by the prior year. This wouldn't have much of an effect this far into the season. Secondly, the ratings are not based on actual team scores. They are based on a weighted average of the team scores and what teams would have scored had they kicked at an expected accuracy, and weighted more towards the latter. So when Melbourne kick 8.18, their score is adjusted (increased) to reflect the below average accuracy and to better reflect their dominance. As background information, the model is optimised for the prediction of matches. Specifically, it is optimised to minimise the mean absolute error in predicted margins. Applying a weighted average in this manner performs better than using actual team scores or just using scoring shots. Lastly, the ratings also take into account estimated opponent strength and estimated venue advantage/disadvantage.

Players polling coaches votes in a given match (2006-): 5 players: 227 times 6 players: 1215 7 players: 1349 8 players: 488 9 players: 68 10 players: 3 Matches with 10 players polling coaches votes: 2007, Round 13, Collingwood v Hawthorn (Link) 2016, Round 17, Adelaide v Collingwood (Link) 2022, Round 22, Western Bulldogs v Greater Western Sydney (Link)

Yes, that is something I'm looking at and will try to incorporate it into the Stats Files update in the future.

Ruck Contests Hitouts To Adv. Darcy Cameron 60 13 1 Mason Cox 24 12 3

Votes Matches Polled 1 2 3 4 5 6 7 8 9 10 11 12 13 Christian Petracca 62 10 5 • 7 9 1 1 10 • 10 • 8 10 1 Clayton Oliver 40 9 2 1 9 8 • 2 7 7 2 2 Jack Viney 23 3 • • • • 6 • • 7 • • • 10 Jake Lever 16 3 9 2 • • • • • • • • 5 • Steven May 16 3 • • • • • • 1 • • 8 7 Max Gawn 15 5 5 • 3 • 3 • • 2 • 2 Brodie Grundy 12 2 • • 6 6 • • • • • • • • • Kysaiah Pickett 12 2 9 • • • 3 • • • • • • Ed Langdon 10 2 • • 6 • • • • • • • • 4 • Angus Brayshaw 7 2 • • • • • • 5 • • • 2 • • Kade Chandler 6 2 • • 1 • • • 5 • • • • • • Trent Rivers 6 1 • • • • • • • • 6 • • • • Christian Salem 4 1 • • • 4 Michael Hibberd 4 1 • • 4 • • • • • Lachie Hunter 3 2 • • 1 • • • • • • 2 • • Bayley Fritsch 2 1 • • • • • • • • • • • 2 Tom McDonald 2 1 • • 2 • • •

These figures are published in the Herald Sun for all players for all matches, but given the requirement to manually transcribe them I only post the highest for Melbourne each week. I can provide others on request.

I'm going to try to automate these posts as much as possible, so I should be able to quite easily incorporate the opponents' HO to advantage each week with no ongoing effort.

I can get the Pies hit outs to advantage. Would you like Melbourne's opponents' HO to advantage each week from now on? There are a few sources for expected scores - the figures I quoted are from the Herald Sun which are provided by Champion Data. They would be the most reliable as they capture the pressure on the player at the time of the kick. The AFLxScore and crow_data_sci twitter profiles both calculate Expected Scores based on the information available to them, which only use a proxy for pressure as the pressure information is not available to them. For example, you can use the fact that a player won a contested possession as a proxy for the amount of pressure they were under, but it won't accurately reflect the pressure they were under when they kicked the ball, only when they gained possession. A player could get an uncontested possession, then run into traffic and kick for goal under pressure. CD would rate this as a more difficult kick than AFLxScore.

Your post is missing the link to the expected scores.

Melbourne v Collingwood Pressure Q1: 176 - 180 Q2: 174 - 172 Q3: 201 - 206 Q4: 186 - 177 Tot: 183 - 186 Most Pressure Points (Weighed sum of pressure acts. Physical pressure acts are worth 3.75 points, closing acts are worth 2.25 points, chasing acts are 1.5 points and corralling are 1.2. https://www.championdata.com/glossary/afl/) Petracca: 74 Viney: 65 Neal-Bullen: 59 Pickett: 43 Sparrow: 40 Chandler: 40 Langdon: 40 van Rooyen: 37 Brayshaw: 33 Time in forward half 58% - 42% Here are my calculations (not official) for time in forward half for each quarter: Q1: 55% - 45% Q2: 50% - 50% Q3: 66% - 34% Q4: 63% - 37% Expected Scores 84 - 64

Melbourne v Collingwood https://www.wheeloratings.com/afl_match_stats.html?ID=20231308 Note: first number below is Melbourne, higher value is bold. Pressure Not available until tomorrow. Q1: ? - ? Q2: ? - ? Q3: ? - ? Q4: ? - ? Tot: ? - ? Most Pressure Points (Weighed sum of pressure acts. Physical pressure acts are worth 3.75 points, closing acts are worth 2.25 points, chasing acts are 1.5 points and corralling are 1.2. https://www.championdata.com/glossary/afl/) Not available until tomorrow. Time in forward half Not available until tomorrow. Here are my calculations (not official) for time in forward half for each quarter: Q1: 55% - 45% Q2: 50% - 50% Q3: 66% - 34% Q4: 63% - 37% Score sources Centre bounce 2.0.12 - 0.0.0 Ball up 0.4.4 - 1.1.7 Throw in 0.0.0 - 1.0.6 Turnover 6.14.50 - 7.7.49 Kick-in 0.0.0 - 0.0.0 Shots at goal Set position 7.10.52 - 3.4.22 General play 1.4.10 - 6.2.38 Expected Scores Not available until tomorrow. Centre Bounce Attendances CBAs CBA % 2023 % 2022 % Christian Petracca 17 81 64.5 74.6 Jack Viney 16 76 67.2 74.6 Angus Brayshaw 14 67 13.5 16.0 Tom Sparrow 12 57 47.4 32.2 Max Gawn 11 52 45.4 65.5 Brodie Grundy 10 48 58.6 83.7 James Jordon 2 10 19.5 0.2 Trent Rivers 2 10 4.5 0.0 Kysaiah Pickett 0 0 13.9 1.3 Jacob van Rooyen 0 0 5.8 Alex Neal-Bullen 0 0 3.9 3.5 Clayton Oliver 82.8 86.5 James Harmes 28.7 14.6 Tom McDonald 5.4 0.0 Harrison Petty 1.3 0.0 Josh Schache 0.0 13.8 Ruck Contests Ruck Contests RC % 2023 % 2022 % Brodie Grundy 42 50 51.5 77.4 Max Gawn 37 44 42.8 57.8 Jacob van Rooyen 5 6 9.0 Christian Petracca 0 0 0.1 0.1 Tom McDonald 8.9 7.0 Josh Schache 6.7 13.4 Ben Brown 3.8 3.6 Harrison Petty 3.1 0.0 Clayton Oliver 0.0 0.0 Hit outs Ruck Contests Hitouts To Adv. To Adv. % (2023) To Adv. % (2022) Brodie Grundy 42 26 5 33.2 30.2 Max Gawn 37 17 6 31.0 33.6 Jacob van Rooyen 5 3 0 16.0 Tom McDonald 25.0 33.3 Harrison Petty 22.2 Ben Brown 0.0 14.3 Josh Schache 33.3 https://www.wheeloratings.com/afl_stats.html

Votes Matches Polled 1 2 3 4 5 6 7 8 9 10 11 12 Christian Petracca 61 9 5 • 7 9 1 1 10 • 10 • 8 10 Clayton Oliver 40 9 2 1 9 8 • 2 7 7 2 2 Jake Lever 16 3 9 2 • • • • • • • • 5 Jack Viney 13 2 • • • • 6 • • 7 • • • Max Gawn 13 4 5 • 3 • 3 • • 2 • Brodie Grundy 12 2 • • 6 6 • • • • • • • • Kysaiah Pickett 12 2 9 • • • 3 • • • • • Ed Langdon 10 2 • • 6 • • • • • • • • 4 Steven May 9 2 • • • • • • 1 • • 8 Angus Brayshaw 7 2 • • • • • • 5 • • • 2 • Kade Chandler 6 2 • • 1 • • • 5 • • • • • Trent Rivers 6 1 • • • • • • • • 6 • • • Michael Hibberd 4 1 • • 4 • • • • Lachie Hunter 3 2 • • 1 • • • • • • 2 • Tom McDonald 2 1 • • 2 • • •

The aggregate of several Squiggle models is very similar, at least in terms of the gap between Melbourne (4th) and Western Bulldogs (5th). https://squiggle.com.au/ladder/

Here are the round-by-round hitouts to advantage for 2022. Average 1 2 3 4 5 6 7 8 9 10 11 12 13 15 16 17 18 19 20 21 22 23 24 25 Max Gawn 9.1 8 6 3 13 5 11 12 7 7 10 6 11 8 11 3 15 10 6 13 6 14 15 Luke Jackson 3.2 3 5 2 1 0 2 4 1 2 0 2 3 4 4 4 3 3 4 7 12 2 2 Sam Weideman 1.3 0 0 0 0 4 0 4 4 1 0 Tom McDonald 0.6 0 1 0 0 2 0 0 1 1 Ben Brown 0.1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1

I can provide the game-by-game hit outs to advantage for 2022 if you would like to incorporate them into your 2022 ratings.

The home advantage is probably the biggest factor overall actually, but mainly for matches played between teams from different states. That sounds like a reasonable theory. I haven't analysed the different paths through the finals in the simulation results, but one thing it won't take into consideration is 1st playing a battle-hardened 2nd/3rd in the prelim. The simulations just see it as 1st being the home team against a likely lower-rated opponent so would win more often than not. Brisbane and Port's premiership chances are definitely negatively affected by having to play the GF at the MCG, but Brisbane is affected more.

Firstly, a team's rating is the sum of the its attacking rating and its defensive rating. The attacking rating is based on whether teams score above or below average and the defensive metric is based on whether teams concede above or below average. However, instead of using a team's actual score, I calculate an "adjusted" score using a weighted average of their actual score and the score they would have kicked had they kicked at an expected accuracy. This is to account for the luck factor in goalkicking. The attacking and defensive ratings are updated following each match based on these "adjusted" scores compared to the expected scores. If a team's adjusted score is higher than expected, their attack rating increases (and vice versa). If their opponent's adjusted score is higher than expected, their defensive rating decreases (and vice versa). The expected scores take into account the teams pre-game ratings and any venue advantage. I have a page on my site which provides some more detail, noting that I have never actually got around to finalising it: https://www.wheeloratings.com/afl_methodology.html In addition, teams carry over ~65% of their rating from the previous season as there's a general regression to the mean from one season to the next.

Yes, absolutely, I'm more than happy for you to use it!

It's largely due to Melbourne being ranked #1 on team ratings by my model. Believe it or not, I actually have Melbourne favourites against Collingwood on King's Birthday, so a Collingwood v Melbourne final would be close to 50/50, or even Melbourne favourites. That's all based on the fact my model currently rates Melbourne 6.2 points higher than Collingwood, which I don't personally agree with, just to be clear. Collingwood's rating actually decreased from rounds 7 to 10 as their opponents were particular inaccurate (Adelaide 7.16, Sydney 6.12, GWS 7.13, and Carlton 7.15) and then didn't beat North by as much as expected. Even though I have Melbourne ranked fourth on the aggregate ladder, they're still more likely to finish second or third than fourth if you look at the ladder distribution. Melbourne's strong percentage definitely helps. I should note that my model doesn't look at the historical record of teams finishing fourth, it simply looks at the individual finals match up for the given simulation. FYI, here are the premiership probabilities for all teams finishing in a particular ladder position this year, with fourth being much higher than historical results with the current final 8 system: 1: 31% 2: 24% 3: 17% 4: 14% 5: 5% 6: 4% 7: 3% 8: 2%

I'll respond to your other post. @old55 is correct that the simulations use the Team Ratings as a basis, and my model has Melbourne currently ranked #1. I agree that it doesn't necessarily seem correct, but in my defence (or my model's defence), the aggregate of a number of models on https://squiggle.com.au/power-rankings/ also sees Melbourne ranked #1 and Collingwood #4. The model simulates the remainder of the season 50,000 times, including simulating the finals based on the outcomes for a particular simulated season. Basically, for a given simulation, I simulate the remainder of the home and away season and then simulate the final series based on the final ladder positions of that particular simulation. I do that 50,000 times and aggregate the results. Melbourne's premiership favouritism is largely due to them being ranked #1.

Yes definitely, I can add that into the standard stats that I include each week - it's not too much work. If I forget, just remind me! 😊

@Demon Dynasty @binman A large factor in the drop off in our goals / inside 50 is our accuracy. Shots / inside 50 has been quite consistent between rounds 5 and 12 but our accuracy has really dropped off. Round Inside 50s Shots Shots / Inside 50 Goals Goals / Inside 50 Accuracy 1 60 30 50.0 17 28.3 56.7 2 54 21 38.9 13 24.1 61.9 3 60 33 55.0 21 35.0 63.6 4 58 33 56.9 19 32.8 57.6 5 55 24 43.6 11 20.0 45.8 6 53 25 47.2 15 28.3 60.0 7 64 30 46.9 22 34.4 73.3 8 61 26 42.6 13 21.3 50.0 9 66 29 43.9 15 22.7 51.7 10 48 23 47.9 11 22.9 47.8 11 58 26 44.8 10 17.2 38.5 12 59 28 47.5 8 13.6 28.6

Yes I do have the hitouts to advantage by player. Hitouts, Melbourne v Carlton Ruck Contests Hitouts To Adv. To Adv. % (2023) To Adv. % (2022) Max Gawn 38 20 9 30.6 33.6 Brodie Grundy 34 11 7 34.6 30.2 Jacob van Rooyen 7 2 1 18.2 Tom McDonald 25.0 33.3 Harrison Petty 22.2 Ben Brown 0.0 14.3 Josh Schache 33.3 Hitouts to advantage by round Average 1 2 3 4 5 6 7 8 9 10 11 12 Brodie Grundy 7.5 4 5 10 13 12 5 9 7 8 4 6 7 Max Gawn 6.1 7 0 5 6 6 4 9 9 9 Jacob van Rooyen 0.4 1 1 1 0 0 0 0 0 0 1 Tom McDonald 0.3 1 0 1 0 0 0 Harrison Petty 0.2 0 0 1 1 0 0 0 0 Ben Brown 0.0 0 0 0 Christian Petracca 0.0 0 0 0 0 0 0 0 0 0 0 0 0 Josh Schache 0.0 0