Deelightful Dee

What a win on Fri night, so pumped to have been there to experience it and cheer the team home! Just thought I'd update the expected scores data after the weekend. We equalled our best performance of the year from Round 3 vs Syd with 29 points more than expected (as per Champion Data's formula). This has only been bettered by the Crows when they kicked 33 points more than expected in Round 13 vs Eagles. Speaking of the Crows and coming up against them this week, they have the best differential between actual and expected score this season. This tends to be driven by them taking many very difficult shots and still converting them into goals. Given I haven't really watched them play so far this year, I'm not sure whether this is skill or luck (probably a combinated of both).

A stats based previous of our to 4 clash tonight: There are many similarities in how both teams play: strong contest & time in forward half are strengths. Lions are the best clearance differential team, but Dees are 2nd over the last 4 games. If Dees can keep up in clearances, a win is within reach!

Whilst talking about expected scores, @crow_data_sci from Twitter, recently posted expected score data covering the 2021 to 2023 season up to round 16. So I thought I'd take a look to determine a different / better perspective of accuracy, given that missing a goal from a hard shot vs an easy shot results in the same accuracy measure generally used in the AFL media. Looking at the actual (result) score less the expected score (xScore) shows that Bowey is our most accurate shot at goal, whilst JVR has started his career strongly in this regard. Fritter is the best of those with plenty of shots, whilst Trac, Gawn and, somewhat surprisingly to me, ANB are down the other end of the table.

In reference to your thoughts on ANB @binman, he's had a pretty solid year, averaging 9.1 ratings point per game, 15th best for the team. Not quite up to his best year of 10.7 (7th in team) in 2018 and 10.2 (10th) in 2021 He's only had 3 games that dipped quite a bit under that average, but overall, the data supports your hypothesis.

I like to use Player Ratings to try to capture player performance in one number as it takes into account where and how the ball was won and the result of the player's action (acknowledging that this still rarely tells the full story). As you can see below, usually when we win (results of games in top left), our average player rating is higher (bottom left). By looking at the individual player ratings and comparing them to their last 40 games' average, we can also see who has performed above and below what we might typically expect (right side graphs). Top right shows highest to lowest performance over/under expectations, whilst bottom right gives some perspective as to where the over/under performers have come from, i.e. our top/mid/bottom tier players. So for last round against St Kilda, I think it was clear that May had a spectacular game, but the top right shows that Jordan and Sparrow were also well above average which is probably down to the full time mid role they both played compared to what they previously have. On the other side of the equation, it was clear that Kossie struggled, but Salem and Grundy in particular actually were even lower than Kossie in terms of their usual output. From an overall team perspective, I also think it's interesting that we do tend to have a lull in performance during the middle part of the year (supporting @binman's loading theory). In 2021, it was for 10 games from rounds 10 to 19 where we lost 5 games and drew 1, before flicking the switch vs Gold Coast in Round 20 and progressively improving towards the oustanding performances in the prelim and GF. In 2022, the 6 games from rounds 11 to 17 were below par other than the round 15 Bris mauling. We then stablised for the rest of the season, but the drop off in the finals was clear. This year looks similar. The 5 games from rounds 10 to 16 were well below our season average, but we've seen an improvement in the last two weeks. If we can continue the upward trajectory, we could end up timing our run ideally for a shot at finals similar to 2021!

here you go @old55

Interested in @binman's question around whether our drop off in accuracy is indicative of a league-wide drop off due to weather and fatigue / loading, so I looked at all team's scoring compated to expectation (using Champion Data's calculation srouced from the Herald Sun) by round this season and calculated the average difference per round (see below data visualisation). League-wide, the average (grey line) looks to hover around the expected difference of 0 as you would expect, but it varies with some early rounds (3 and 4) being lower than average with the 'worst' round being round 11. In saying that, round 12 and 13 were actually better than expected and this was when we performed most poorly other than round 16. In terms of Melbourne's performance (red dots) in poor weather / rain, I can recall round 5 vs Ess, round 9 vs Port, round 15 vs Gee and round 16 vs GWS. The first two we scored above expectation, the last two well below. It's difficult to assess the fatigue / loading aspect for us, but there looks to be something in this if we assume we load around our bye in round 14. It will be interesting to see how the rest of the season unfolds from here as you wouldn't expect there to be loading from here, although fatigue / rain / wind / cold may be a factor. And sometimes, it could just be luck :)

Hi Demon faithful. First time poster, long time reader. Yes, RIchmond was the last time we won when expected to lose - see below for this year's actual vs expected results