Predicting Uber-Premium Midfielders

Written by Father Dougal on February 25 2020

Hi Everybody!

Back when I was first making the spreadsheet for doing midfield TLAs (True Levels of Ability) I decided to do a quick check on Uber Premiums. To me, that means players with an average of 125 or higher. After all, there are not that many, and they are important. There were only seven recent mids who have ever done that. I decided not to go back to the early days of Superoach when the scoring system was different. All Saints already did a great post on this, but I wanted to check for only their very first time ever scoring 125+ just to see if there was any way to predict that. With only seven players it wouldn’t take long to do a quick check. Looking at them there turned out to be some things that got my attention.

Four of the seven were 23 when they first posted an uber score. One was 24, one 25, and one 26. So, 23 is the big year to watch for. Clearly older players can make the jump, but 4 of 7 is significant. I did a quick search of my sheet for players who are 23 this season and came back with just one name. 

While there is more than one  23 year old mid, none were on my sheet, so probably not relevant. There are a lot of 24 year olds, and a number at 22, but for some reason there’s a gap at 23. Ok, again makes it easy, I grabbed the stats for their two seasons prior to their big one when they went boom, and then got highs and lows and averages, and compared that to the one 23 year old candidate this season. 

All seven 23-26 years old

1 Before 2 Before
Ave 114.8 100.1
Range 106-123 91-106

So, our candidate had an average of 116.5 last season and 95.1 the season before. That’s right in the middle of both groups. Well…..things just got all serious. 

So, I started looking in more detail. Which kept getting more involved as a went on, because this seemed to warrant real research and not just a jump to a conclusion I happened to like. 

Below is the table with the data I put together for the Magnificent Seven.

Age Boom!

1 Before

2 Before 3 Before 4 Before

5 Before

Last 10 Games
Macrae 23 127.1 106.9 94.1 99.9 100.1 61.4 114 22
Ablett 23 132.2 114.2 91.4 22
Pendlebury 23 129.2 110.5 106.5 103.5 88.7 56.8 112.1 22
Rockliff 23 132 109.7 97.3 113.5 85.9 121.1 21
T Mitchell 24 129.1 118.9 104 104.5 83.8 86.2 122.5 22
Dangerfield 25 131.8 119.9 105.6 112.9 118.9 80.3 123.2 21
Swan 26 126.9 123.2 102 100.9 99.4 78.1 22

1 Before

2 Before 3 Before 4 Before 5 Before
23 only Ave 110.3 97.3 105.6 91.6 59.1
Range 106-114 91-106
24+ Ave 120.7 103.9 106.1 100.7 81.5
Range 119-123 102-106
All Ave 114.8 100.1 105.9 96.1


Range 106-123 91-106 100-113 84-119 56-86

Last 10s are missing for Ablett and Swan because the data does not seem to be available. We can use their full season averages as guides if needed. 

So, some other interesting bits:

Every one of the Magnificent Seven had played 21 or 22 matches the year before they went boom. 

Every player who went boom when over 23 had an average higher previous season average than every player who went boom at 23. Season before of 106-114 for the 23 year olds and 119-123 for the 24 and overs. 

The last 10 for the old guys we have data for and presumably Swan, were all over 120, but since their season averages were much higher that may not mean much.  The last 10 for the 23 year olds we have data for, all had a higher last ten average than full season average. 

The averages for the six players who played 3 and 4 seasons before going boom are all good in those seasons. Minimum of 100 three before and 83 four before. We have no data for Gary because he was not in the AFL 3 seasons before he went boom!

1 Before 2 Before 3 Before 4 Before Last 10 Games
The  Candidate 23 116.5 95.1 60 69.6 126.7 22

So, how’s he look now? Played 22 games, so good there. Last 10 over season average by a lot, so good there. Actually, he has a higher last 10 than any of the Magnificent Seven! His 3 and 4 before averages are way low. So, not a perfect fit. But how close and how much do we care. 

I don’t see his “1 before” season being a bit over the averages of the other 23 year olds as being a problem.  His last 10 being so high I can’t see as being bad. His having such low averages 3 and 4 seasons before does seem relevant. How relevant? From a pure math point of view it drops the confidence in his going boom. The more seasons back the less I care, so his last two being right in line seems much more important. If they weren’t good we would not be looking at the 3 and 4 before. But we can look at his specific circumstances to see if there was a reason for the low scores. 

By now I suspect it is obvious The Candidate is Josh Dunkley. In 2016 and 2017 he played 13 and 7 matches, and he played as a forward, mostly. The Bulldogs were busy going for the title in 2016, and as a rookie he played rounds 1 & 2, and did little. Then he had a quick trial rounds 10-14 and was blah again. During their big, end of season run, Dunkly played the last seven matches for an average of 83.  In 2017 the Bulldogs were busy going for a spoon, and Dunkley was playing until he suffered a fully dislocated shoulder and had a mid season reconstruction. In the three matches before he hurt his shoulder, his average was 75. So, interesting, somewhat less bad, but still. 

About then I realized that since there are six 24 year old candidates this season that didn’t go boom last season, that means there are six players who I could look at as 23 year olds last season and see if they would have been predicted to go boom. Also, I can go back extra for Lachie Neale. 

No-Boom 1 Before 2 Before 3 Before 4 Before Last 10 Games
Cripps 23 117.1 119.4 97.7 107.6 96.3 128.7 22
Kelly 23 117.4 113.8 114.3 87.1 70.6 119.6 15
Bontempelli 23 114.7 104 105.1 107.7 103.2 93.8 19
Merrett 23 105.6 100.4 109.2 111.5 88.5 113.1 22
M Crouch 23 103.7 101.6 110.6 92.8 68.3 108.4 18
Neale 23 109 112.6 104.3 87.2 86.9 114.1 22

First, Neale in 2017. He fits the profile well. He played 22 matches. His last season average was within the 106-114 range, and his 2 before average is within the 91-106 range for 23 year olds. His three before is low though, but his four before is ok. His last 10 was higher than his season average and in the range there too. But, not only did he not go boom, his average dropped to 109! So, right away we see that fitting the profile is not enough. If I had this data around back then, I would have put Neale in my starting team. And then been sad. 

The Bont, Merrett and Crouch all had previous season averages below 106 and so were instantly in doubt. Bont was at least just below, and so maybe not too bad, but Merrett and Crouch were way below and right out. Crouch also didn’t play 21-22 games, and so out of the running. Bont also didn’t play 22, and had a last 10 way below his season average, and so out of the running too. Merrett was good with every other metric, but being so far off the previous season total he’d have to be out. As it turned out, Merritt and Crouch both did about the same as their previous seasons. Bont did 10 points better. 

Kelly is odd, in that his two before average was way high, and I have no idea if that is bad, but it does not fit. More importantly he only played 15 games, and that puts him well off the Magnificent Seven, so out, and as it turns out he did slightly better but no boom. 

Cripps fit perfectly. All four previous averages in the ranges, and he played all 22 with a last 10 of 128.7.  Hard to fit the profile much better. And yet, his average dropped by two points. 

It appears I have reached the obvious conclusion that going at 125+ is hard. Maybe if Cripps wasn’t the obvious player to be tagged in his team he might have made it. On the other hand, He did have a darn good season. 

So, predictions for this season? Data below.  

2020 1 Before 2 Before 3 Before 4 Before Last 10 Games
Oliver 22 109.4 114.7 111.5 70.3 111.9 22
Dunkley 23 116.5 95.1 60 69.6 126.7 22
Cripps 24 117.1 119.4 97.7 107.6 114.6 20
Kelly 24 117.4 113.8 114.3 87.1 122.2 14
Bontempelli 24 114.7 104 105.1 107.7 117.5 22
Merrett 24 105.6 100.4 109.2 111.5 97.9 22
M Crouch 24 103.7 101.6 110.6 92.8 104.1 19
Neale 26 121.3 111.9 109 112.6 118.2 22

Predictions will be in the Midfield TLA posts. They had to wait for this, since I’m gonna be referring back to this stuff some. 

But a hint? Dunkley is the only player other than Grundy that I think is a genuine must have this season. He will be in my team. 


Thanks for reading!


Leave a comment / Scroll to bottom

9 thoughts on “Predicting Uber-Premium Midfielders”

  1. Predicting premo jumps is hard enough – uber premo jumps more so!

    Solid effort though FD, I do like Dunkley a lot, but I can’t quite justify getting rid of Macrae for him, which would really be my only option (for a variety of reasons). Can easily see him smashing it out of the blocks though, especially as the dogs are looking the goods this year.


  2. Love your work Father, but I’m sticking with Macrae and Bont.. currently have Mitchell too.. all have the same bye.. so adding Dunks will screw me .. I’ll have to target him after the bye ..


  3. What a read FD, love this!

    Josh Dunkley was my first picked midfielder and is an absolute lock! I couldn’t be any more confident on him. Genuinely think his going to push the #1 mid spot. Can’t believe his only in 12% of teams.


  4. Awesome stuff FD, really interesting read.

    I’m in the Macrae camp, and I don’t usually like having players from the same team in the same area of the field, but may have to make an exception with Dunkley.

    I’m also of the belief that Cripps will be massive this year.


  5. Great article, I don’t like having two premo mids from the same team in my starting line up but its looking like I will be starting both JMAC and Dunkley.


  6. Great write up Father! Gee it’s hard to build a case to leave out MacRae/Bont/Dunkley besides the fact that they are all in the same team. I’m actually picking Bont to go nuts this year. I reckon the captaincy suits him and the doggies will be a much improved team.



Leave a Reply

Your email address will not be published. Required fields are marked *