Opinion Questions For Rowsus

Rowsus · 21 Aug 2020

Heya Rumb,
just be thankful you don't have both Sicily AND Houston, like I do!

I think you either need to take Houston to the best you can (Haynes, Ryan), take a risk on Williams, or go value like Thurlow or Blakely. I believe Ryan is a bit of risk outside of WA, and I think Williams possibly misses a game somewhere, so I'd either go Haynes, or if like the possibility on taking a risk on Thurlow, you could use the leftover change to turn Rankine into Parker, or Zerrett.
I like a "different" road. I'd probably go Houston/Rankine to Thurlow/Parker.
Good luck, whichever way you jump.

Oh, and your % dilemma. You are safe.
You pretty much have to lose by 1,000, and the 9th team win by 1,000, for them to get ahead of you on %!

McLovin · 21 Aug 2020

Hi Rowsus,

I'm one of only 4 sides in the top 100 still to own Gawn and I'm considering holding him again this week to fix Sicily instead. Have I gone completely insane?

For reference I'm thinking of Sicily > Foot and Serong > last mid upgrade (yet to be decided on). That would leave me with Lukosius moving to D6 for now, Foot or Close covering F6 this week, Cripps M8 and Draper covering for Gawn again.

Gawn and Sicily to Lloyd (I still don't own him) and NicNat/whoever would be the obvious play if Gawn's out again, but going against the crowd is just so damn tempting.

nicohighscore · 21 Aug 2020

/!\ Sorry it is a long one... if it pollutes this topic we can move somewhere else ... and I might split the discussion in several messages.../!\

Hi Rowsus,
I don't know if this theme has been discussed in details in the past, but I can see that you "work" a lot with averages,
and I would like to have your view on the below...
(I had started to develop this in the other model, and I wonder if it is worth continuing and recreate it in the new one)

The idea is to a apply a analytical strategy, comparable to what you would do with stock prices,
and create dynamic indicators that allow you to anticipate variations and help with decisions.

What I have tried to do as a starting point to gather historical data (SC scores here) and defines 3 simples indicators :
- @22R average = rolling average of the last 22 rounds
- @11R average = rolling average of the last 11 rounds
- @3R average = rolling average of the last 3 rounds

Why these ?
@22R is representative of the long term performance - It is like dynamic version of the "last year average" which is often utilised. It should give a good idea of the next 22 rounds (in average)
@11R is more representative of the current form., it should be a good indicator. It should be a good indication of the next 11 rounds
@3R is the moment form - it is mostly an indicator of the volatility of player's performance, as well as his ceiling (and bottoms).

The 1M$ question is : how do we know in which direction these are going to go ?

The standard answer is generally : as long as the short average is above the long average, the long average should go up ( and vice versa),

which means we can define a set of values ( difference between the averages) and signals (change of signs of these values) to help identify the future movements for each players.
For example : If 3R>11R>22R : this is the best configuration, player performance is and should continue to improve compared to past performance.
when 3R crosses 11R - that's a first warning, if it also crosses 22R it is a second warning.
If eventually 11R crosses 22R, then it means players performance is expected to decrease...

or do you think we should consider the long average like "magnets" ? (if feels it more the sense of your answers in general)

It is probably a bit of both 🙂 - like for everything, the idea is to find the proper balance I guess....

so
do you think this approach can be relevant ( is it worth trying to salvage it from the previous model)
Do you think the parameters (22,11,3) are good ?
how about other advanced indicators (like bollinger band for ex) - more generally, can we apply the same logic ... ?

Below I am pasting a few examples ...
(It is only updated as of R8, and please ignore the few technical challenges I still need to solve - ( like, I have to use a standard scale, so the correspondence with year and round has to be done on the side table)

Petracca - I guess the "signal" to follow would have been around 225 ( end of 2019) when 11R crosses 22R up with a big angle
(however, so similar "fake signals" around 170 and 205 ... ) - maybe his premium potential was here already but didn't get to express...

Dusty - aside his volatility - we can see that the seeds of his great 2017 season were planted in the mid 2016 season ( 150) ...
then the plunge is impressive ... the last pike corresponds to the end of 2019 ..
If I had this at the start of the season, would I have picked him in my starting team ? probably not...

this is what made me buy Greenwood - well obviously it wasn't such a great move...

Rumbler · 21 Aug 2020

Rowsus said:

Thanks Rows

I'm worried about getting Thurlow as I think he may be over-performing looking at his stats from previous years and JS. Also with Rampe being out for the rest of the season, would it impact his scoring?
Considering getting Parker but I don't want to have too many Swans players! Already have Lloyd, Bell, Wicks & Gould (if he gets selected to play)

Rowsus · 22 Aug 2020

McLovin said:

Hi McLovin,
you've carried the Gawn can this far, I think you'd be crazy to trade him now, especially in a 1 of 4 in a 100 situation. While it will cost you points this week, it will cost those around you that want him a trade next week! That trade is probably worth more than what holding Gawn will cost you. Having him as a potential POD is a huge incentive, as is having a saved trade. Yes, you could hold him, and have him miss yet another week, or come back compromised and score an 80 or 90. That's the risk balanced against the reward. If there were 40 or 50 teams still with him, and you had say 11 or 12 trades left, I'd say dump him. I'm guessing you don't have those 12 trades, and only 4 have him. It's a risk, but I think it's worth hanging on now.
Well done, I hope you keep climbing the Rankings!!!

Rowsus · 22 Aug 2020

nicohighscore said:

/!\ Sorry it is a long one... if it pollutes this topic we can move somewhere else ... and I might split the discussion in several messages.../!\

Hi Rowsus,
I don't know if this theme has been discussed in details in the past, but I can see that you "work" a lot with averages,
and I would like to have your view on the below...
(I had started to develop this in the other model, and I wonder if it is worth continuing and recreate it in the new one)

The idea is to a apply a analytical strategy, comparable to what you would do with stock prices,
and create dynamic indicators that allow you to anticipate variations and help with decisions.

What I have tried to do as a starting point to gather historical data (SC scores here) and defines 3 simples indicators :
- @22R average = rolling average of the last 22 rounds
- @11R average = rolling average of the last 11 rounds
- @3R average = rolling average of the last 3 rounds

Why these ?
@22R is representative of the long term performance - It is like dynamic version of the "last year average" which is often utilised. It should give a good idea of the next 22 rounds (in average)
@11R is more representative of the current form., it should be a good indicator. It should be a good indication of the next 11 rounds
@3R is the moment form - it is mostly an indicator of the volatility of player's performance, as well as his ceiling (and bottoms).

The 1M$ question is : how do we know in which direction these are going to go ?

The standard answer is generally : as long as the short average is above the long average, the long average should go up ( and vice versa),

which means we can define a set of values ( difference between the averages) and signals (change of signs of these values) to help identify the future movements for each players.
For example : If 3R>11R>22R : this is the best configuration, player performance is and should continue to improve compared to past performance.
when 3R crosses 11R - that's a first warning, if it also crosses 22R it is a second warning.
If eventually 11R crosses 22R, then it means players performance is expected to decrease...

or do you think we should consider the long average like "magnets" ? (if feels it more the sense of your answers in general)

It is probably a bit of both 🙂 - like for everything, the idea is to find the proper balance I guess....

so
do you think this approach can be relevant ( is it worth trying to salvage it from the previous model)
Do you think the parameters (22,11,3) are good ?
how about other advanced indicators (like bollinger band for ex) - more generally, can we apply the same logic ... ?

Below I am pasting a few examples ...
(It is only updated as of R8, and please ignore the few technical challenges I still need to solve - ( like, I have to use a standard scale, so the correspondence with year and round has to be done on the side table)

Petracca - I guess the "signal" to follow would have been around 225 ( end of 2019) when 11R crosses 22R up with a big angle
(however, so similar "fake signals" around 170 and 205 ... ) - maybe his premium potential was here already but didn't get to express...

View attachment 20233

Dusty - aside his volatility - we can see that the seeds of his great 2017 season were planted in the mid 2016 season ( 150) ...
then the plunge is impressive ... the last pike corresponds to the end of 2019 ..
If I had this at the start of the season, would I have picked him in my starting team ? probably not...

View attachment 20235

this is what made me buy Greenwood - well obviously it wasn't such a great move...

View attachment 20234

Hi nhs,
for a member that's only made 84 posts, you've certainly made an impact!
I really enjoy posts like this, and I've read through it once, but want to re-read it, and digest it further before I respond.
I think the basic idea is useful, but doesn't account for a couple of factors, that are near as damn it impossible to accurately factor. I will go into this further in my response. I think the Bollinger band idea might have some merit. I'm a bit worried about a Taleb's "Black Swan" type scenario, that could drastically skew the data, though.
As I say, I need to digest it further. I will assume you are familiar with Taleb's Black Swan. If you aren't, I highly recommend the book to you. In the meantime, thanks again for your great efforts in improving the content on the site!

Rowsus · 22 Aug 2020

Rumbler said:

Happy to help, Rumb.
Keep in mind, Thurlow has only played 57 games (across 7 and a bit seasons!). He's playing a role, in probably the weakest team he's been involved with, that he hasn't played much before. While he's actually a smidge taller than Rampe, I don't see him playing Rampe's role, as they are different type players. As to having too many Swans, I'd only be worried about that if there were more Keepers involved. You've probably got one too many of their Rookies, but in the end, Rookies are temporary citizens in our team, and should be treated as such. Gould could just as easily been some other deadwood Rookie, from another team, so his presence shouldn't stop you getting another hopefully Keeper from the Swans.
All that aside, he's definitely a risky pick, and not for everyone's taste or purpose. Good luck, whichever way you jump.

McLovin · 22 Aug 2020

Rowsus said:

Thanks mate, I agree that it's worth the risk. Somehow I do have 11 trades left, but I'll hopefully find more creative ways to use them. If all goes well I'll have a finished side next week with 7 trades left for slingshotting and/or an extra premium to loop with my lower end players like Lukosius and the Big O.

Rumbler · 22 Aug 2020

Rowsus said:

Nicely explained! Just to let you know I made my trades last night before the Carlton vs. GC game and yes I did take the risk on Thurlow!! The extra 77k I got from trading Houston should hopefully help me slingshotting D.Smith to a Fyfe/Oliver next round. Hopefully it pays off!

However the other trade I made was Rankine to Gaff. Didn't go for Parker because I saw JPK return in the lineup. I think Gaff is good value at that price and I'm hoping he can have a strong finish to the season after a below par last 3-4 weeks.

Rowsus · 24 Aug 2020

@nicohighscore
I'm working my way through it, and I believe it has potential, but needs fine tuning, as the indicators are too "slow".
Let's look at Rozee as an example:

We can see his planets aligned in Rnd 1 this season, and he became a trade in option for Rnd 2. He scored well in Rnd 2, but then the wheels fell off, badly. The problem was, this wasn't indicated until Rnd 9's numbers were in the table, and that was too late!
Let me say, I traded him in for Rnd 2, and traded him out after Rnd 8, so I inadvertently followed the directions!

The problem is it doesn't and can't factor in some of the reasons for score swings. I don't think the idea is horrible, or without use, but maybe needs refining and re-directing. Things like role changes, which can be temporary, opposition, location, niggles are hard to factor in using a purely mechanical method.
I did some playing around with the Bollinger band idea, and tried limiting a players scores to one standard deviation, to try and smooth the curve a little. I even tried with 1.5 SD's up, and 1 down, as a players ceiling is quite often further from his average, than his basement is. It didn't quite have the affect I hoped for.
I will noodle with it further, and see if between us we can make it into some sort of usable model. I think one of the problems is a double edged sword, which Rozee perfectly depicts. The 3 Rnd average might be too small of a sample, as if you do what Rozee did, and hit 2 outlier scores within a 3 Round average, it will nearly always push you into a 3>11>22 situation. The other edge of the sword being, if you push the 3 Rnds out to 5 Rnds, it means you are reacting too slowly to the genuine discoveries the system found, and losing the points and dollars advantage you might have otherwise achieved.
In that regard, it might become more of a tool for indicating when a player might be dumped, rather than traded in. As always, the underlying thing is, is there a reason the players scoring has suddenly changed? Can we identify the reason? And, is that reason likely to persist in the longer term? They are the questions we should always be asking ourselves, when looking at a player, whose score has started to deviate in either direction, from where we might reasonably expect it to be.
I'll fiddle further, and let you know if I find anything.

nicohighscore · 25 Aug 2020

Rowsus said:

Hi @Rowsus,
Thank you for your answer.
I had to check the Taleb's Black Swan's wikipedia page, and I will probably need to check the book out - thank you for that.

Of course, this is very early stage and just on a technical point of view, there is plenty of possible enhancements.
I am already filtering out the injury impacted scores, but you could also re-align performance with a standardised opposition or location...
We could also easily switch to more "on field" figures which are "making" the scores - so a signal could be confirmed this way as well.

And I agree that overall, it can't be read in isolation : a signal (positive or negative) should always be checked with a "field" modification like role change. (and if the signal is related to a significant change, they the question is to know if the change will stay or not)

I am still doing baby steps with the tools, and I think I have reached the limits of Power BI and I need to get my hand on more robust platforms ( SQL & Python on the menu...)

Hopefully I will be able to have a stable and complete model soon ...
Very happy to entertain the discussion in the meantime.

Rumbler · 25 Aug 2020

Heya Rows! Hope you had a good weekend!

Performance of the two new recruits:
Thurlow

Gaff

As I'm facing two different elimination match-ups this week, I was hoping to get your advice (as mostly always!

) to be best prepared as possible!! Funnily enough, my upcoming opponent in my main league also has Thurlow so I'm not too worried about him this week.

As it stands I have 92k cash and 11 trades remaining.

I was thinking of trading Bell & D.Smith for a mid/def rookie and a mid worth up to 602k. Not sure who yet.
My opponent has Jelly, Cripps, Cogs, T.Greene, Menegola and your personal favourite A.Brayshaw

I'm worried a few of these players might go huge this round, especially Brayshaw if De Boer returns and tags Fyfe.
Is the strategy to pick up one of the players above to neutralise my opponent's scoring? If so, who?!

I could also trade out Close for another rookie to get more cash in order to get Oliver but I'm saving the third trade in case Gawn doesn't play again this week. Oliver may also get tagged by Ross/Steele.
Unfortunately Essendon are playing in the first game this round so I can't wait for any of the team announcements!!!

Cheers!

Rowsus · 27 Aug 2020

Rumbler said:

Hey Rumb,
11 trades puts you in an enviable position!!!
I have 2 bits of advice for you.
Do everything you can to win your most important League/s this week. Having 2 weeks without matches is worth gold this season.
Do not trade before Thursdays matches are played. I know you want to potentially use Smith, but tactically, you are better to wait. This gives you a chance to view any early trades they may have made, and also gives you a window as to how your match is going. Potentially you may need to trade Gawn AND Neale, just to try and secure that win. Hopefully you don't have to, but that's why you wait! Having 11 trades they are not dead to you, and both will be cheaper after next weeks non-game Round.
As to trading in one of his POD's against you to nullify it, I would only do that if you hold a lead, and nullifying that POD helps secure that lead.
Good luck, I hope you win both!

Rowsus · 27 Aug 2020

@Rumbler
hopefully you kept your finger off the trade button. Smith's 89 might possibly come in handy, and make it in your best 18.
His +$11,400 doesn't hurt, either.

Rumbler · 27 Aug 2020

Rowsus said:

Yep I did! Haven't made any trades yet! My opponent also had Smith on field and has already made his 3 trades. He got rid of W.Day, T.Papley and Neale for J.Lloyd, Riccardi & Macrae. That's some big ins!! Looks like it was veryyyyy wise to wait and hold off from making any trades.

He's currently projected to get 2075, while I'm projected at around 2070.
At this stage I think I'm gonna need to trade Neale. Just don't know to who though and I can't make my mind up. Bont? Jelly? Oliver? Steele? Ahhh!!

And if push comes to shove I may need to trade Laird as well...probably to Haynes but I wanna try and hold Laird if I can.
Will await and see if Gawn gets named too.

Rumbler · 5 Sep 2020

Hey Rows,
Firstly, can't believe this thread was all the way on the 2nd page! Especially at this stage of the season!

Two semi-final matches for me this round. I'm trying to complete my team but I'm struggling with the uncertainty of the Jelly situation and what trades to make. I think I'm being a bit too greedy so hoping you could assist in making my mind up on a few players!

7 trades and 380k left. Need two more upgrades (1 defender and 1 mid).
For this round I was trying to find a way to trade in both Steele and Neale but doing this would mean I have no cash to upgrade to a premo defender next round (ideally want to get Houli).
Looking to trade Wicks, Dusty & Rivers/Schoenberg at the moment but if Jelly misses then I will need to trade him. My opponent in my main league has both Neale & Oliver and I need to get one of them in this round to stand a chance!
Any trade suggestions you can give is greatly appreciated!

Thanks mate!

Imcjb · 5 Sep 2020

Rowsus,

I hope you are well. Thanks again for contributing to this thread for yet another year....we all appreciate it!

My question is mid related.

Have the following as my mids, need two more premiums, money not an issue. Who's the two best mids for the next 3 games.

Titch, Steele, Kelly, Danger, Duncan, Cripps, XXXX, XXXX

Oliver/Neale/Macrae? Or something left field? I'm going purely for leagues with one must win game this round.

Rowsus · 5 Sep 2020

Rumbler said:

Hey Rumbs,
7 trades AND $380,000 - WOW! What a position to be in!
Firstly, I think I'd turn Dusty into May/Ryan/Blicavs - depending on your desire for some POD action.
I think I'd also be tempted to trade Kelly, even if he plays. There is a fair chance he might still be affected, even if he plays, and you need those dollars working for you, to win these finals. I'd be looking at Oliver/Steele/Lyons/Neale.
I'd then hold that 3rd trade, and wait and see how the match ups are going. If it looks like you need a boost, then Grundy to ROB, to try and get you home/keep you ahead.
So, maybe Dusty/Kelly to May/Neale, or if you are feeling really frisky Blicavs/Lyons! Then as I say, hold, wait and assess, to see if you need ROB to get you home.
Good luck!

Rowsus · 5 Sep 2020

Imcjb said:

lmcjb,
thanks for the kind word, and I am well, thank you. I hope you are too.
You've got 2 or 3 Mids there that might be under a cloud, one way or another.
Certainly the 3 you have listed are strong picks, and hard to argue against. If you wanted to go something out there and POD'ish, then M Crouch has been in super form, and Ade have Carl and Rich, who can both give up a Mid score.
I'd probably pick 2 of the 3 you have listed, or Lyons, if you want to shy away from Neale.
Good luck!

Rumbler · 10 Sep 2020

Good evening Rows!

The preliminary finals are upon us except...
...this round has barely even started and it's already on the brink of being over for me unfortunately. I know that sounds pessimistic and although I'm really happy West Coast won, the fact that Tim Kelly almost got a double ton has screwed me over big time. Gaff's score, however, has slightly kept the flame still alive for me!
I'm playing against the guy who I beat in last year's final. He's out for revenge and right now it looks like he's gonna get it.

My team is on the left below. His team on the right. Assuming we both captain Gawn, I'm projected 2492 and he's projected 2559. Note that I can't see his current team because the website is not letting me, so gotta assume he still has two trades to make this round (likely has upgraded Riccardi to someone). The team on the right below is his one from round 16.

Trades are no issue as I have 4 left. 255k cash to spend. Injury concerns to Simpkin and Docherty only make things worse though.

What is the approach from here on how to go for the win? Make high risk/high reward trades? I will wait to see the line-ups for the Saturday games which are announced tomorrow. I need to trade out Wicks but don't know who to bring in and hoping you can suggest a genuine match winner who can go big these last two rounds.

Should I bring in a defender (move Rivers to mid) or a midfielder? VC/Captain choice to make? Currently thinking Gawn into Neale.

Also if Simpkin & Docherty both somehow do play, then trading Schoenberg from the bench to get extra cash may be an option.
Always grateful for any advice/suggestions you have! Don't think there's been a round this year where I need it as much as I do this round!
Many thanks!

p.s. Feeling a bit superstitious so gonna tag @Darkie for no apparent reason...well last year it worked and I won...

!!

Opinion Questions For Rowsus

Rowsus

Statistician

McLovin

nicohighscore

Rumbler

Rowsus

Statistician

Rowsus

Statistician

Rowsus

Statistician

McLovin

Rumbler

Rowsus

Statistician

nicohighscore

Rumbler

Rowsus

Statistician

Rowsus

Statistician

Rumbler

Rumbler

Imcjb

Rowsus

Statistician

Rowsus

Statistician

Rumbler