Discussion Perfect SuperCoach

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Actually the more I stare at it the more I like it...

View attachment 7348
I'm not sure I like constraining it to pick so many DPPs. Specifically I don't think Westhoff is a good pick and at 12.9% ownership I don't the crowd agrees. The R/F FD is fine.

Also if you are wishing to make it like the crowd, you might want to scan as many RMT threads to pick up how many premos, mid pricers, and rookies people are running. I'm not sure where it's falling out this year... this might give some more quantitative data behind the 16 rookie threshold....

I do like it shoves the PTA DPPs in the mids though, that is cool and against what I'd do, but that the beauty of it.

I do like that I can compare my team and pick my bets against the crowd more accurately.

Any news on what the optimal structure would be based on generic premos, mid pricers and rookies?

Cheers
 
Joined
21 Apr 2018
Messages
25
Likes
106
I'm not sure I like constraining it to pick so many DPPs. Specifically I don't think Westhoff is a good pick and at 12.9% ownership I don't the crowd agrees. The R/F FD is fine. Also if you are wishing to make it like the crowd, you might want to scan as many RMT threads to pick up how many premos, mid pricers, and rookies people are running. I'm not sure where it's falling out this year... this might give some more quantitative data behind the 16 rookie threshold.... I do like it shoves the PTA DPPs in the mids though, that is cool and against what I'd do, but that the beauty of it. I do like that I can compare my team and pick my bets against the crowd more accurately. Any news on what the optimal structure would be based on generic premos, mid pricers and rookies?
Cheers
Yeah when I originally pitched the idea to the HS it was to help people compare their team against the crowd more accurately so I am glad you find it helps.

That's a good point about Westhoffs low percentage. I just don't believe that Fort will get a game and dislike having cover in the rucks. But I guess that it me trying to interfere with the crowd.

The most recent team I have has a 11-4-15 structure which seems fine.

Regarding the optimal structure I feel that we'd have to make too many assumptions to make it meaningful. Basically we would be tweaking the numbers until we got the answers we wanted to see. Not really my jam. For example, what are you trying to maximise in the structure? The expected score in round 1? The expected score across the season? If so would you take into account likelihood that the players would play each week? You mentioned something about having 20 trades to save some for injuries? But if you were only considering generic players you wouldn't ever trade because they're all the same? Unless some players only start at different points across the season? You could add complexity until the cows come home but every time you make assumptions the model becomes less reliable so it feels a bit like a time sink to me. Sorry probably overcomplicating what you meant.
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Yeah when I originally pitched the idea to the HS it was to help people compare their team against the crowd more accurately so I am glad you find it helps.

That's a good point about Westhoffs low percentage. I just don't believe that Fort will get a game and dislike having cover in the rucks. But I guess that it me trying to interfere with the crowd.

The most recent team I have has a 11-4-15 structure which seems fine.

Regarding the optimal structure I feel that we'd have to make too many assumptions to make it meaningful. Basically we would be tweaking the numbers until we got the answers we wanted to see. Not really my jam. For example, what are you trying to maximise in the structure? The expected score in round 1? The expected score across the season? If so would you take into account likelihood that the players would play each week? You mentioned something about having 20 trades to save some for injuries? But if you were only considering generic players you wouldn't ever trade because they're all the same? Unless some players only start at different points across the season? You could add complexity until the cows come home but every time you make assumptions the model becomes less reliable so it feels a bit like a time sink to me. Sorry probably overcomplicating what you meant.
What i was trying to optimise is max season score by varying only the number of generic rookie mid pricer and premos. Assuming an infinite supply of each as they are generic :).
20 trades because this model wouldn't have rookie corrections, bye round strategies and injuries to deal with because generic player don't get injured and play every game.

The only variable I see left is the scoring output of the rookie/midpricer/premo. You could vary that to get an outcome you wanted hence starting with some ex ante assumptions about what they score cuts that out.

I don't want it to be a time sink...

If I was pitching to the hsun it would be 'how many mid pricers is too many?'

But cool if you don't want to do, no probs. Cheers Andrew
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
What i was trying to optimise is max season score by varying only the number of generic rookie mid pricer and premos. Assuming an infinite supply of each as they are generic :).
20 trades because this model wouldn't have rookie corrections, bye round strategies and injuries to deal with because generic player don't get injured and play every game.

The only variable I see left is the scoring output of the rookie/midpricer/premo. You could vary that to get an outcome you wanted hence starting with some ex ante assumptions about what they score cuts that out.

I don't want it to be a time sink...

If I was pitching to the hsun it would be 'how many mid pricers is too many?'

But cool if you don't want to do, no probs. Cheers Andrew
Other variable is when the trades occur...hope the model can optimise that.
 
Joined
21 Apr 2018
Messages
25
Likes
106
Other variable is when the trades occur...hope the model can optimise that.
Yeah we could totally model trades my main concern is how many assumptions you would need to make the solutions useful. Your suggestions definitely help though. It could also be used to help determine when to cash in rookies as well, yeah?

I think the neatest way to do this would be to create something interactive so people could play with the numbers themselves as opposed to my normal trick of saying "these are the answers to a question you didn't ask". Will put it on the list of stuff to play with when I have time.

Thanks for the suggestion :)
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Yeah we could totally model trades my main concern is how many assumptions you would need to make the solutions useful. Your suggestions definitely help though. It could also be used to help determine when to cash in rookies as well, yeah?

I think the neatest way to do this would be to create something interactive so people could play with the numbers themselves as opposed to my normal trick of saying "these are the answers to a question you didn't ask". Will put it on the list of stuff to play with when I have time.

Thanks for the suggestion :)
Good man, I very much look forward to it.
 
Joined
21 Apr 2018
Messages
25
Likes
106
Don't want to talk about it hahah

I ended up removing the constraint that forced DPP players and just went with ultimate cookie cutter. Coming 100k-th! Disaster!

Didn't help that I didn't take Cripps' VC and stayed with Grundy. I blame @randomcliche ;-).
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Don't want to talk about it hahah

I ended up removing the constraint that forced DPP players and just went with ultimate cookie cutter. Coming 100k-th! Disaster!

Didn't help that I didn't take Cripps' VC and stayed with Grundy. I blame @randomcliche ;-).
It was correct in the theory of the cookie cutter! Sorry!

For what it's worth, I'm surprised that given the information signalling that comes from HSun, and many dedicated websites going with a cookie cutter team would end up with anything other than about the the mean. Without the information sharing, I think you would be on winner.

My team only has 3 onfield players with sub 25% ownership (Rocky, Goldy, Dunkley 13-15%) (1 bench @ 3%). For me it's the communal sharing of informed participants on SCS (as well as about 50 'rules' I have, that I invariably break) that has the potential to make improvements on the cookie cutter, so thanks to @Adrenaline and others who suggested not taking Collins and others over the long pre-season that began back at xmas!

PS I hope for your sake he stinks it up next week and you forget I ever made the suggestion :)
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Don't want to talk about it hahah

I ended up removing the constraint that forced DPP players and just went with ultimate cookie cutter. Coming 100k-th! Disaster!

Didn't help that I didn't take Cripps' VC and stayed with Grundy. I blame @randomcliche ;-).

Stayed with Grundy and don't have Westhoff..... you can thank me now. Round 3 second most purchased ruck, Grundy, most sold this round Westhoff. Hope your overall rank is soaring, while your trades left is 30!
 
Joined
21 Apr 2018
Messages
25
Likes
106
Stayed with Grundy and don't have Westhoff..... you can thank me now. Round 3 second most purchased ruck, Grundy, most sold this round Westhoff. Hope your overall rank is soaring, while your trades left is 30!
Yep, definitely! You were right and I was wrong. In fact BOTempelli got in the top 1% in round 2, which was pretty motivating! Still sitting at 40kth overall though (much better than 100kth). My main problem at the moment is that all 30 in the squad are playing so not being able to use the captain loophole.

Thanks again
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Yep, definitely! You were right and I was wrong. In fact BOTempelli got in the top 1% in round 2, which was pretty motivating! Still sitting at 40kth overall though (much better than 100kth). My main problem at the moment is that all 30 in the squad are playing so not being able to use the captain loophole.

Thanks again
Good man! And now my problem is not having Grundy, and two KOs last round...

Well played BOTtompellie :p Plenty of scope for rank improvement!

Anyways love to hear what your trades plans are. Cheers I'm two down with dusty and Balta out and t.kelly and LDU in. No bank. TRYING. FOR. NO. TRADES. THIS WEEK.
 
Joined
21 Apr 2018
Messages
25
Likes
106
here comes BOTempelli!

Why no trades this week @randomcliche? I’m normally pretty loose with trades before the byes and then coast in the last few rounds. BOTempelli always wants to make the maximum amount of trades and I haven’t really got a rule to decide when to or not so any advice would be appreciated.
 

Attachments

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
here comes BOTempelli!

Why no trades this week @randomcliche? I’m normally pretty loose with trades before the byes and then coast in the last few rounds. BOTempelli always wants to make the maximum amount of trades and I haven’t really got a rule to decide when to or not so any advice would be appreciated.
Boom! after that start you're now back on the train! Another 120 points and you'll be passed me!

1. I was wondering if you were going to have some crowd bot rule, e.g. 1) look at top 2 purchases for the week and top 2 sales from your team; or b) look at top ownership of premiums and look to buy those... Or whether you had some other data points you were focusing on.

2. Why no trades? No trades was really about making sure I've got trades for proper upgrades, bye round swaps and injuries, and not using them on questionable sideways trades now. Dunkley to Marshall I probably should have done...but that fitted perfectly in the category of sideways trades that may/may not pay off. But now after 4 weeks of not playing primarily mid i really just need to pray for few injuries to other dogs mids...which as an other of libba and macrae I'd rather not have :p Grateful for your thoughts.

3. My rules for rookies is I try to buy everyone that that has BE of -80 or better and most with BE of -60 or better, provided they are selected rnd 3 and there is not some easy to spot rationale as to why they'll be out soon. My general principle is good rookies keep their spot.

4. My big connumdrum this week is Rowan Marshall. At $406k has tonned up in every game, seems a no brainer... but does he keep Billy Longer out of the team for the rest of the season??? If
 
Joined
21 Apr 2018
Messages
25
Likes
106
1. I was wondering if you were going to have some crowd bot rule, e.g. 1) look at top 2 purchases for the week and top 2 sales from your team; or b) look at top ownership of premiums and look to buy those... Or whether you had some other data points you were focusing on
Yeah you have the wisdom of crowds stuff pretty right. If >10% of people haved traded someone out or in I definitely do that. If its around 8% then I allow myself some discretion (for now).

The other half of the equation is using the pretty crappy SC gold projections. I input the forecast data into my optimiser and that spits out all the decisions I should do to get the overall optimal score based on my current team / remaining budget, etc. If I like any of the trades then I do it if I don't then I wait. So far they have all been pretty terrible suggestions but this generally gets better throughout the season. For now that's all I'm doing.

3. My rules for rookies is I try to buy everyone that that has BE of -80 or better and most with BE of -60 or better, provided they are selected rnd 3 and there is not some easy to spot rationale as to why they'll be out soon. My general principle is good rookies keep their spot.
This is absolutely fantastic info. Will build this into BOTempelli. "not some easy to spot rationale as to why they'll be out soon." is that for example some of the richmond players in the absense?

Yeah not sure about Marshall I'll see what the optimiser tell me. Doesn't look like he'll get over 8% trade in.
 

randomcliche

Rising Star Nominee
Joined
3 May 2017
Messages
320
Likes
359
AFL Club
Hawthorn
Yeah you have the wisdom of crowds stuff pretty right. If >10% of people haved traded someone out or in I definitely do that. If its around 8% then I allow myself some discretion (for now).

The other half of the equation is using the pretty crappy SC gold projections. I input the forecast data into my optimiser and that spits out all the decisions I should do to get the overall optimal score based on my current team / remaining budget, etc. If I like any of the trades then I do it if I don't then I wait. So far they have all been pretty terrible suggestions but this generally gets better throughout the season. For now that's all I'm doing.



This is absolutely fantastic info. Will build this into BOTempelli. "not some easy to spot rationale as to why they'll be out soon." is that for example some of the richmond players in the absense?

Yeah not sure about Marshall I'll see what the optimiser tell me. Doesn't look like he'll get over 8% trade in.
I think the easiest one to spot this week is R. O'brien (Adl) this week. Sam Jacobs comes back in, there is no spot for him. Stack this week has better LT security due to Short's elbow. But yeah, I wouldn't load up on 3 richmond rookies.
 
Top