Discussion Perfect SuperCoach

Spots

Rising Star Nominee
Joined
13 Mar 2016
Messages
587
Likes
926
AFL Club
West Coast
@steveedwards just read the article on HSun, hope at some point you can put it up on here for the masses without a subscription. Just wondering if your analysis stretched to the classic 'ghost ship' team to see what the best team could have been based on original team selections?

EDIT: Great article btw, very interesting analysis
 

chels

Rising Star Winner
Joined
8 Jan 2015
Messages
331
Likes
177
AFL Club
Sydney
Hola @steveedwards - how is progress on the doctorate? Only boring people finish "on time"; a lot of the fun (and later on in your career you might not get the space accorded) is in looking at what seem like good ideas but turn out to be blind alleys.

Looking forward to seeing the optimised team.
 

Spots

Rising Star Nominee
Joined
13 Mar 2016
Messages
587
Likes
926
AFL Club
West Coast
Hola @steveedwards - how is progress on the doctorate? Only boring people finish "on time"; a lot of the fun (and later on in your career you might not get the space accorded) is in looking at what seem like good ideas but turn out to be blind alleys.

Looking forward to seeing the optimised team.
FYI it is complete and on the Hun website. I dont want to steal Steve's thunder by posting it though...
 

Spots

Rising Star Nominee
Joined
13 Mar 2016
Messages
587
Likes
926
AFL Club
West Coast
Try going through the link on the Superfooty Twitter account.

I'm glad my trading strategy of using up all my trades early on wasn't far off optimal. The team itself I can't say the same.
yes thats a good tip. I've found often if I click on a link to a Hun article on twitter it opens up without the subscriber issue
 
Joined
21 Apr 2018
Messages
32
Likes
141
Hi everyone, thanks for your interest! To be honest I don't have a herald sun subscription so I can't actually access the article so if anyone else wants to post it please do. I really don't like that its behind a firewall - am thinking about going back to AFL Fantasy instead for that reason. Does anyone play that on here?

@chels thanks for asking - my thesis was accepted last Wednesday. Took me 4 years 2 months to submit so 8 months longer than "on time" but yeah definitely worth it.

@Spots I did run the analysis on the ghost ship. It came outside the top 100 so not as good as last year. One of these days it'll win though. I can post the results if you're interested.
 
Joined
21 Apr 2018
Messages
32
Likes
141
Also for those of you who like maths I actually published an academic paper on this stuff through the Australian Society of Operations Research (ASOR) - attached.

I swear my actual research is on more important topics than SuperCoach.

Edwards, S.J., (2019). Analysing Fantasy Sport Competitions with Mixed Integer Programming. Data and Decision Sciences in Action 2 – Proceedings of the 2019 National Conference of the Australian Society of Operations Research, eds. A. Ernst, S Dunstall, R Garcia-Flores, M. Grobler and D. Marlow. Springer, 2019, in press.
 

Attachments

Spots

Rising Star Nominee
Joined
13 Mar 2016
Messages
587
Likes
926
AFL Club
West Coast
Also for those of you who like maths I actually published an academic paper on this stuff through the Australian Society of Operations Research (ASOR) - attached.

I swear my actual research is on more important topics than SuperCoach.

Edwards, S.J., (2019). Analysing Fantasy Sport Competitions with Mixed Integer Programming. Data and Decision Sciences in Action 2 – Proceedings of the 2019 National Conference of the Australian Society of Operations Research, eds. A. Ernst, S Dunstall, R Garcia-Flores, M. Grobler and D. Marlow. Springer, 2019, in press.
Thanks Steve, I'd like to say i got to page 12 of your report before I got lost but it was somewhere on Page 6 :unsure:
 

chels

Rising Star Winner
Joined
8 Jan 2015
Messages
331
Likes
177
AFL Club
Sydney
Well done on submitting. I guess you have a "defence" coming up in a couple of months? That grand final is bloody intense but really satisfying at the same time. Just think: for a little while no one knows more about a subset of an esoteric subject than you! Cool, yeah.
 
Joined
25 Jul 2012
Messages
16,324
Likes
10,891
AFL Club
Collingwood
I realise it is a hindsight team but struggle to understand who would have picked Marshall from the start when he wasn't named for Round 1.

Even Stack , surely you would want at least 29 players named for Round 1 assuming you would start a donut at R3.

Maybe it shows we should disregard any rules we have learnt and just go for it.

Good read nonetheless.

EDIT

Trading Mundy IN when we were all trying to get rid of him stands out as well.
 

Connoisseur

Rising Star Winner
Joined
3 Jul 2017
Messages
3,776
Likes
8,112
AFL Club
Port Adelaide

Tried posting the article directly but it was more than the 10000 character limit. The link above should work for those that want to read it who are not Herald Sun subscribers.
Perfect SuperCoach team would have scored 8500 more points than the 2019 overall winner

SuperCoach players are winding down and enjoying a well-earned break after another huge fantasy season.
Now is a good time to embrace hindsight (and some complex maths) and take a look at what an Optimal SuperCoach team would have looked like for 2019. This team represents all the decisions a coach should have made to obtain the highest score that was theoretically possible.

But first, congratulations to Dwayne Delmenico, coach of this year’s winning team, Raxwell’s Crumbers. Very glad to hear that someone can go from 20,000th to the overall winner in the space of a year. Gives the rest of us battlers some hope for next year.

This year the optimal team scored 62,101, approximately 8500 points more than the winning team’s overall score of 53,679.

The table below provides an overview of the initial starting squad, including the percentage ownership of each player before the start of Round 1.
1569490549990.png

And here’s a summary of the decisions made each round such as who was named captain and who was traded in and out to maximise our score.

1569490644239.png
1569490663633.png
It is also worth pointing out that the team value of the final squad was a whopping $16,695,100 with $81,400 remaining in the budget. This is more than $3 million more than the value of the winning team.

Some observations...

SO MANY PODs
Thirteen of the 30 players in the starting squad had a percentage ownership of less than 10 per cent. They Includes Ricky Henderson, Ben Cunnington, Josh Battle, Rowan Marshall and Reilly O’Brien, who were all in less than 1 per cent of teams. This is a remarkably brave squad that would have likely been laughed out of most “rate my team” forums. It is also interesting to see George Hewett, a name not frequently mentioned in fantasy circles, make an appearance in the team for his captaincy-worthy performance of 154 in Round 7.

WEAK STARTING MIDFIELD AND FORWARD LINES
The initial midfield and forward lines are noticeably light on. Most of the most hyped pre-season premiums such as Patrick Cripps, Patrick Dangerfield, Isaac Heeney, Dustin Martin, Jack Macrae, Clayton Oliver and Nat Fyfe are all overlooked for players such as Lachie Neale, Marcus Bontempelli and Ben Cunnington. In fact, Heeney and Dusty never once make it into the team and Dangerfield only appears in Round 11. The relatively weak midfield and forwards lines freed up the budget to invest in a very strong backline and ruck division.

POWER RUCKS

Max Gawn and Brodie Grundy have made it back-to-back seasons as locks in the ruck. Despite this, probably the highlight of this year’s ruck arsenal was Reilly O’Brien. O’Brien started at $136,800 with an ownership of less 1 per cent before replacing Sam Jabobs as Adelaide’s No.1 ruck for the last 18 games of the season. In fact, O’Brien outscored Grundy, the highest total points-scorer of the season, in Rounds 7 and 12, forcing Grundy to the bench. In Round 13, when both Gawn and Grundy had the bye, O’Brien scored a massive 181 points, earning him the captaincy of the optimal team in that round.

BIG OMISSIONS IN DEFENCE
The most notable omissions this year comes in the defence. Three of the top five highest scoring defenders never feature in the side. Tom Stewart (second), Rory Laird (fourth) and James Sicily (fifth) are all overlooked. This comes despite the fact that the starting squad contained five players in defence all with an initial value of more than $400,000: Jake Lloyd, Lachie Whitfield, Shannon Hurn, Daniel Rich and Zac Williams.

CONSISTENT TRADING PATTERN EMERGING
This is the third year now that we have presented the optimal team of the year. Each time the same trading pattern has emerged. No trades are made until after the Round 3 price changes. The optimal team then trades hard, in particularly around the byes, using all three available trades each round. Then at the back end of the season the team coasts for the last five or six rounds without any trades in the bank. This is only really possible due to the depth of the squad late in the season. For example, in Round 23 the optimal team has the following players sitting on the pine: Nic Newman, Dane Rampe, Adam Treloar, Jarryd Lyons, Michael Walters, Todd Goldstein, James Worpel and Toby Greene. Pretty luxurious, really.

SIBLING SYNERGY
It’s also nice to see that the Curnow brothers are traded into the team on the same week, Round 12. Goes to show that you cannot underestimate the power of having a family oriented team.

SOME WHOPPING SCORES
The optimal team only scores over 3000 in the last round. However, three real teams scored more than the optimal team in the same round so it is clearly shows that these big scores are obtainable in reality.

A special thank you to Max, coach of Frad Bairhurst, for his help with this year’s analysis (Max actually took home the $1k for highest score of Round 19).

Steven is a PhD Candidate in the School of Mathematical Sciences at Monash University and works on large-scale optimisation problems
 

Ironhawk

Well-known member
Joined
15 Mar 2019
Messages
959
Likes
2,368
AFL Club
Hawthorn
Perfect SuperCoach team would have scored 8500 more points than the 2019 overall winner

SuperCoach players are winding down and enjoying a well-earned break after another huge fantasy season.
Now is a good time to embrace hindsight (and some complex maths) and take a look at what an Optimal SuperCoach team would have looked like for 2019. This team represents all the decisions a coach should have made to obtain the highest score that was theoretically possible.

But first, congratulations to Dwayne Delmenico, coach of this year’s winning team, Raxwell’s Crumbers. Very glad to hear that someone can go from 20,000th to the overall winner in the space of a year. Gives the rest of us battlers some hope for next year.

This year the optimal team scored 62,101, approximately 8500 points more than the winning team’s overall score of 53,679.

The table below provides an overview of the initial starting squad, including the percentage ownership of each player before the start of Round 1.
View attachment 11730

And here’s a summary of the decisions made each round such as who was named captain and who was traded in and out to maximise our score.

View attachment 11731
View attachment 11732
It is also worth pointing out that the team value of the final squad was a whopping $16,695,100 with $81,400 remaining in the budget. This is more than $3 million more than the value of the winning team.

Some observations...

SO MANY PODs
Thirteen of the 30 players in the starting squad had a percentage ownership of less than 10 per cent. They Includes Ricky Henderson, Ben Cunnington, Josh Battle, Rowan Marshall and Reilly O’Brien, who were all in less than 1 per cent of teams. This is a remarkably brave squad that would have likely been laughed out of most “rate my team” forums. It is also interesting to see George Hewett, a name not frequently mentioned in fantasy circles, make an appearance in the team for his captaincy-worthy performance of 154 in Round 7.

WEAK STARTING MIDFIELD AND FORWARD LINES
The initial midfield and forward lines are noticeably light on. Most of the most hyped pre-season premiums such as Patrick Cripps, Patrick Dangerfield, Isaac Heeney, Dustin Martin, Jack Macrae, Clayton Oliver and Nat Fyfe are all overlooked for players such as Lachie Neale, Marcus Bontempelli and Ben Cunnington. In fact, Heeney and Dusty never once make it into the team and Dangerfield only appears in Round 11. The relatively weak midfield and forwards lines freed up the budget to invest in a very strong backline and ruck division.

POWER RUCKS

Max Gawn and Brodie Grundy have made it back-to-back seasons as locks in the ruck. Despite this, probably the highlight of this year’s ruck arsenal was Reilly O’Brien. O’Brien started at $136,800 with an ownership of less 1 per cent before replacing Sam Jabobs as Adelaide’s No.1 ruck for the last 18 games of the season. In fact, O’Brien outscored Grundy, the highest total points-scorer of the season, in Rounds 7 and 12, forcing Grundy to the bench. In Round 13, when both Gawn and Grundy had the bye, O’Brien scored a massive 181 points, earning him the captaincy of the optimal team in that round.

BIG OMISSIONS IN DEFENCE
The most notable omissions this year comes in the defence. Three of the top five highest scoring defenders never feature in the side. Tom Stewart (second), Rory Laird (fourth) and James Sicily (fifth) are all overlooked. This comes despite the fact that the starting squad contained five players in defence all with an initial value of more than $400,000: Jake Lloyd, Lachie Whitfield, Shannon Hurn, Daniel Rich and Zac Williams.

CONSISTENT TRADING PATTERN EMERGING
This is the third year now that we have presented the optimal team of the year. Each time the same trading pattern has emerged. No trades are made until after the Round 3 price changes. The optimal team then trades hard, in particularly around the byes, using all three available trades each round. Then at the back end of the season the team coasts for the last five or six rounds without any trades in the bank. This is only really possible due to the depth of the squad late in the season. For example, in Round 23 the optimal team has the following players sitting on the pine: Nic Newman, Dane Rampe, Adam Treloar, Jarryd Lyons, Michael Walters, Todd Goldstein, James Worpel and Toby Greene. Pretty luxurious, really.

SIBLING SYNERGY
It’s also nice to see that the Curnow brothers are traded into the team on the same week, Round 12. Goes to show that you cannot underestimate the power of having a family oriented team.

SOME WHOPPING SCORES
The optimal team only scores over 3000 in the last round. However, three real teams scored more than the optimal team in the same round so it is clearly shows that these big scores are obtainable in reality.

A special thank you to Max, coach of Frad Bairhurst, for his help with this year’s analysis (Max actually took home the $1k for highest score of Round 19).

Steven is a PhD Candidate in the School of Mathematical Sciences at Monash University and works on large-scale optimisation problems
Thanks. Couldn't get the tables to post properly on mobile. Guessing you had to save these as an image?
 

Darkie

Moderator
Joined
12 Apr 2014
Messages
6,500
Likes
4,162
AFL Club
Collingwood
Fascinating stuff @stevedwards! A couple of questions for you:

- Do you play SC, and if so, has your performance improved over time having conducted this analysis?

- What are the key practical (SC) conclusions in your view? I have some in mind, but you’re the expert!
 
Joined
21 Apr 2018
Messages
32
Likes
141
@Darkie Yep I play SuperCoach. This year I tried to do a bot team but didn't get around to improving the projections so it more or less came down to following percentage ownership. After round 1 I was 100kth or so, then after 10 rounds was in the top 2k then finished around 14k I think. The main thing I learnt was that the less effort you put in the more likely you are to make stupid mistakes - had a lot of captain's loopholes that I simply forgot to update over the weekend.

I think the main practical outcomes are
  • Don't trade in the first 3 weeks
  • Trade hard early / around the byes and try to coast for the last few weeks. I have seen a lot of the winners have a similar strategy
  • You can make a lot of $$$ of midpricers if they go well (that's the hard part)
  • Trade out cash cows early. The optimal team seems to trade out players often when most of us are scrambling to trade them in.
  • Also avoid super premiums in your starting team (apart from Grundy and Gawn) - seems a lot of the time the optimal team starts with a stack of plays between 300k-500k and waits for the super premiums to fall but again that's because it knows which of the 300-500k players will increase so is still risky.
Not sure what else I am forgetting. You all probably know a lot more than me!
 

JT23

Rising Star Nominee
Joined
10 Jul 2012
Messages
174
Likes
19
AFL Club
Adelaide
Hi everyone, thanks for your interest! To be honest I don't have a herald sun subscription so I can't actually access the article so if anyone else wants to post it please do. I really don't like that its behind a firewall - am thinking about going back to AFL Fantasy instead for that reason. Does anyone play that on here?

@chels thanks for asking - my thesis was accepted last Wednesday. Took me 4 years 2 months to submit so 8 months longer than "on time" but yeah definitely worth it.

@Spots I did run the analysis on the ghost ship. It came outside the top 100 so not as good as last year. One of these days it'll win though. I can post the results if you're interested.
I love this analysis every year (even if it is impossible to predict like you can with the benefit of hindsight - it is still fascinating).

I have had about 3 years in a row now where my team has performed very poorly after several years of finishing in the top couple of thousand, so trying to learn what has changed or where my strategy has changed by looking at some of these analysis!

One of the only things I got right this year was starting Boak. I was tempted to pull the famous "Travis Cloke" trade out at the top of their cycle - but as usual had too many problems to try and fix (and probably not enough b@lls to be honest). All my mates were trying to get him in around Round 4,5,6 and I notice that this team got rid of him at round 7!

I would actually be super interested in the optimal Ghost Ship team if that was something that you could show easily?

Love your work mate and congratulations on the PhD!!
 
Joined
21 Apr 2018
Messages
32
Likes
141
I would actually be super interested in the optimal Ghost Ship team if that was something that you could show easily?
Love your work mate and congratulations on the PhD!!
Thanks mate,

I didn't actually look at the optimal ghost team because the results weren't super great but it is actually very interesting. Caleb Graham in the ruck!

Oh it actually looks like there's a mistake - I think you can only have 4 emergencies at once right? My team has 6 including 3 in the middle(!), which in itself is pretty amazing as it always take the lowest value. Check out Round 12! That's pretty beautiful
 

Attachments

Last edited:

JT23

Rising Star Nominee
Joined
10 Jul 2012
Messages
174
Likes
19
AFL Club
Adelaide
Thanks mate,

I didn't actually look at the optimal ghost team because the results weren't super great but it is actually very interesting. Caleb Graham in the ruck!

Oh it actually looks like there's a mistake - I think you can only have 4 emergencies at once right? My team has 6 including 3 in the middle(!), which in itself is pretty amazing as it always take the lowest value. Check out Round 12! That's pretty beautiful
Is there one too many midfielders in this team?

Its interesting to see how well this team goes - it actually just beat me, and I had 30 trades and changing captains etc (while it had none hahaha).
 
Top