Strategy List Management 101

[robbieando]

No. Everyone knows their list is good, this is a club I've seen criticised a lot but I think people have overlooked a key factor.

If its Norf I'm carding you

 

[ScrewInStuds]

No. Everyone knows their list is good, this is a club I've seen criticised a lot but I think people have overlooked a key factor.

My guess is North or Tiges

 

[Leg End]

He was taken at #29, that's about average for a key defender. Certainly fits the bracket.

Due to positive feedback I have started writing Episode 2 - List Construction and Depth (AKA why a certain club's list is better than people are giving them credit for)

Love your owrk - most informative.

If you need some further information, i could contribute "How List and Performance are Compromised by Cheating Corrupt Victorian Umpires who Obviously Hate Sydney". It's possibly more qualitative than quantitative, but still has merit.

 

[swansfan51]

If its Norf I'm carding you

It is not norf, please let me survive a bit longer.

Episode 2: 28! Its a magic number (How I Learned to Stop worrying about Morons on the D+T Board and Love Stephen Wells)

This post refers heavily to my previous thread on depth. To avoid repeating it all again, here is the summary:

1) Contenders need 30-35 players to play AFL football during a season (good teams sometimes use more than this, see Fremantle this year, however this is not a major issue and is less important than the following points)

2) On average, your team each week will include 3-4 players ranked between 23-28 on your list (this top 28 is crucial)

3) The 29th-35th players will play <5 games each

4) The 36th-44th players are almost completely irrelevant

The concept of a Best 22 is nominal at best, and serves only to allow posters to debate each other on the board as to whether 3rd rebounding halfback X deserves the final bench spot over small forward Y. What is actually important is your Best 28. In practice, almost every team has a couple of its nominal Best 22 missing each week, meaning that halfback X and small forward Y play 20+ games each, and the debate didn't serve any purpose.

The top 4 this year was Hawthorn, West Coast, Fremantle and Norf. At the start of 2015 each club had the following number of players with 50+ games experience:*

Fremantle - 23
West Coast - 23
Fremantle - 24
Norf - 23

Sydney - 17
Adelaide - 17
Richmond - 21
Bulldogs - 13

See that gap between the top 4 and the next 4? Not a coincidence. The top 4 clubs having more 50+ game players than any other club in the league? Not a coincidence. Sydney being less able to cover injuries? Not a coincidence.

Winning a premiership inevitably requires a degree of luck. No team would win the premiership with as many injuries as Brisbane or the Gold Coast suffered in 2015 (sidenote: look out for GWS in 2016, they are better placed than most sides). You might avoid suspensions, you might get an easier finals draw, you might get an extra home game or 2 along the way. If you're Brad Scott, you may even avoid playing at a stadium with the roof open.

However you cannot simply rely on luck, and complain when something doesn't go your way. You must engineer your list to play the percentages, and the best 4 teams had a full 6 players stronger depth than the Swans did. They could each suffer significantly more injuries before having to call on untested players.

*50 games is not a magical milestone, it was just a very handy sorting point. Some clubs (the Bulldogs being the most obvious recent example) have a significant number of very talented players in years 1-3, who are under this level, but certainly are good enough to play when called upon. The 50+ stat is only a sidenote to the more important number (28).

So what does all this have to do with Geelong anyway? Get to the ****ing point!

Geelong will not win the premiership in 2016. Sorry to disappoint. However the criticism of Geelong is that they are taking a high risk approach to list management, trading away their future, and they will have no depth because they have lots of empty spaces on their list and will need to take a bunch of gutter trash draft picks.

Are they really trading away their future? I may cover this in a future post later, but for now a quick summary will do. Geelong traded away 2 first round picks, one was #9 and lets assume the one next year is #10. They also traded away #28. The historical failure rate for picks 9+10 (1997-2005, allowing for players to have developed a full career) is that 20% of them will fail to play 25 games. On average a pick 9-10 plays 85 games.

Therefore your average pick 9-10 is not as good as Dangerfield (thank you captain obvious) and about level with Henderson. #28 is a handy steak knife however it has a historical failure rate of ~45% and the average pick plays ~50 games. If you're playing the percentages, which you should always do, study the historical success of draft picks and you will see they are often overvalued. I see no fault in those trades of 3 picks for 2 high quality players. If pick 9 or 10 becomes a star it doesn't necessarily invalidate the deal either. This point usually confuses people and/or leads to disagreement, so a quick example:

I have a magic coin. When flipped, it lands on heads 55% of the times and tails 45% of the times. I offer you even money (a $1 stake gives a $1 profit), and I win if the coin lands on heads. If the coin lands on tails, you have won but you still made a poor decision in electing to bet. If you trade Nathan Fyfe for pick 76 and 77, and are able to select 2 repeated All-Australian club champions at those selections, it does not make your trade any better.

The argument that Geelong have traded away a large amount of players/depth and left list spots open is 100% correct. At the moment I believe they have only 36 players on their list for 2016, so they would need to use 8 selections at shitty picks of 60+. However we've already established that the bottom 8 spots on your list are not very relevant - if you are using them and relying on those players to play regular games and get you wins, you are already ****ed.

More importanly, what is Geelong's Best 28? Here's the importance of the magic number. Geelong have 19 players with 50+ games experience, a solid enough starting point. They then have Horlin-Smith, Smedts, Thurlow, Menzel, Vardy, Bews, Lang, Gregson, Cockatoo and Kolodjashnij who have shown the ability to be at least partially competent at AFL level. Account for the fact that Menzel and Vardy combine to equal 1 sort-of-healthy player, and that's 28 players who Geelong could feel confident in selecting to make a contribution.

And what will happen with those gutter trash draft picks? History says they have a success rate of ~20% and will play on average 30 games. With no consideration of either the reported weakness of this draft, or Wells' history as an above average recruiter, on percentages they get 1-2 competent players even from these shitty selections. Next year they walk into the draft with another batch of picks and a chance to add another 2-3 competent players. The bottom of the list starts to build again.

Summary: Geelong have enough players to get through a season with average injuries, without any trouble. Depth doesn't need to extend to the bottom 8 spots on your list. The Best 28 is what matters.
Disclaimer: This is a quantitative analysis only. If those 28 players are shit, no dice. Sorry Stephen.

 

[The King!]

No. Everyone knows their list is good, this is a club I've seen criticised a lot but I think people have overlooked a key factor.

Mind blown if its Sydney

 

[Tuco]

This top 28 formula really highlights how our trade restrictions have crimped our style.

Interestingly, it looks like Geelong are moving in a direction not dissimilar from Sydney (sans trade restrictions).

Like us, they're shortening the 'expected' list of players getting a game (meaning the 30+ players are becoming ballast) offsettingh this by topping up at the top end (Dangerfield).

Like us, they are moving on from a hardened crew who have led the club to Premiership success for a prolonged period.

Our list of players under 50 games likely to play a role in 2016 are:

Mitchell 39 (Regular 22)
Heeney 14 (Reg 22, 50 by 2017)
Jones 11 (should be a regular 22 in 2016)
Sinclair 29 (will be a mainstay. 50 in 2016)
Drex 15 (depth/never to be seen again)
Towers 17 (looks like will end up with a reg spot)
BJ 27 (depth, pushing for regular spot)
Lloyd 43 (could be 1st 22)
Bird 137 (depth unless traded)
Robinson 4
Rose 2
(Hewett, Hiscox, Naismith, Nanka, Davis, Mills)

We also have Johnson under 50

The difference between us and Geelong is they were able to bring in players to help (Henderson, Smith, Selwood) to add to that top 22/top 28.

We haven't been able to do that for two years. We got Laidler and Drex in 2012.

So it looks like we'll be running with Heeney, Lloyd, Jones and Towers in our starting 22. When the preference would have been to only run with maybe two of them.

That means our top 28 (pending trade period ending) becomes:

BJ, Bird, Rose, Robinson, plus 2 assorted rookies/listed players (Mills? Hewett, Hiscox, Davis, Naismith, Nanka, Aliir?)

Basically we're out of whack until we get games into those players, or until we trade in 3 or 4 guys to help in 2016.

 

[mtooler]

This is really interesting stuff, good job buddy!

 

[The King!]

#swansfan51tothewarroomplease

 

[Wayne Swan]

Fabulous stuff, really well researched and documented.

Puts the numbers and analysis to back up what a number of us have been saying the last couple of years, we lack depth and that's what killed our 2014. It's why Hawthorn are such an awesome side, quality depth.

It's why we really need to pick up a DFA or two that can play a role whilst we see if our next batch of kids are good enough. The 2016 best 22 threads show that we barely have enough experienced/quality players due in part to the cap clearing for Buddy and the illegal trade ban. When guys like Towers and Mills appear in your best 22, you know you won't cope with even a small injury list.

Thanks heaps for the article.

 

[swansfan51]

I'm working on a draft edition ahead of next week's National Draft. Focusing on the value of draft picks, drafting needs vs talent, and hopefully swansrule100's question about mature age draftees if I can. Any other topic requests?

 

[RW]

Not a request, but I'm curious to hear your thoughts on something Kinnear Beatson said in his draft preview on Swans TV the other day. He said something to the effect that you can always find picks in the 30-40 range (and later) who end up playing more games than those picked in the first round, because we're dealing with 18 year olds who have a long way to go in their development. Intuitively this does make some sense, but I know you're a cold hard evidence kind of guy, so what say you in response to his claim?

 

[RUNVS]

Not a request, but I'm curious to hear your thoughts on something Kinnear Beatson said in his draft preview on Swans TV the other day. He said something to the effect that you can always find picks in the 30-40 range (and later) who end up playing more games than those picked in the first round, because we're dealing with 18 year olds who have a long way to go in their development. Intuitively this does make some sense, but I know you're a cold hard evidence kind of guy, so what say you in response to his claim?

Well given the likes of Parker and Hannebery I am going to have to agree with Beatson.

 

[swansfan51]

His statement is correct (not an opinion). And I'll be posting stats reflecting exactly that (games played by players at various levels). But note this in conjunction with the consideration of positions I posted earlier. You can always find back pockets and ok inside midfielders late, a fair bit harder to find key forwards etc.

 

[GotTheGoodes]

I'd love to see some analysis on ruckmen.

Correlation between draft spot, age until they started performing, likelihood of drafting a gun ruckman late in the draft etc etc. Just think it's often overlooked how much additional development the ruck crew need in comparison to other spots. Linking this with natural talent, age, time on one list etc would also be interesting..... Anyway, more mermering to myself than asking the question really...... Carry on!

 

[swansfan51]

Episode 3A - Value of draft picks

Are the AFL values on picks accurate?

TLDR: Yes, but

Longer answer:

This year for the first time the AFL have attributed a points value to picks in the draft. These start at 3000 points for pick 1 and tail off into the 70s. Pick 72 is worth 19 points, 3 more than Craig Bird.

The nature of the points system has lead to controversy as academy clubs traded away their top selections for lower picks that collectively had a greater points worth. This came with a great deal of criticism, and the popular view was that the later draft picks had been given too high a value. The argument was that no club would normally agree to a trade such as those that the academy clubs accepted - Sydney's exchange of pick 14 for picks 23 and 44 being a prime example.

Is pick 14 for 23 and 44 really a bad trade?

A pick in the banding 13-16 (better data set than just using pick 14) plays on average 80+ games. There's a bust rate of close to 25% (a scrub that plays <25 career games).

A pick in the banding 21-24 plays on average ~60 games. Bust rate of ~40%.
A pick in the banding 41-48 plays on average 50+ games. Bust rate of ~45%.

Therefore it looks like picks 23 and 44 are indeed worth more than pick 14, just like the AFL bidding system says there are. Not only do they go on to play more AFL matches combined, but you get 2 chances of a decent player and there is that 1/4 chance that your pick 14 is a waste of space.

BUT

The AFL system does a good job of attributing a points value to selections based on their actual performance. However there are 2 caveats

1) The 'tail-off' isn't right

I feel the AFL system pretty much nails the early picks. In reality there is a gap between picks 1-4 and 5-8, and a gap between 5-8 and 9-20.

However the gap between picks 20 and 50 has historically been far lower than the AFL's points system suggests. Below in blue is the AFL's line, and in red is what I feel is closer to historical accuracy.

Picks 17-20 average 72.25 AFL games
Picks 21-24 average 53 AFL games
Picks 25-28 average 53.75 AFL games
Picks 29-32 average 62 AFL games
Picks 33-36 average 47.25 AFL games
Picks 37-40 average 64.25 AFL games
Picks 41-44 average 49 AFL games
Picks 45-48 average 50 AFL games
Picks 49-52 average 38 AFL games
Picks 53-56 average 52 AFL games
Picks 57-60 average 30.5 AFL games
Picks 61-64 average 32.25 AFL games
Picks 65-68 average 28.5 AFL games
Picks 69-72 average 30.25 AFL games

Note the very significant gap between pick 20 and the rest, and then how little difference there is in picks 21-56.

2) There is not an unlimited number of list spots

The above statistics clearly demonstrate that if there was no maximum list size, and you could fit them in the salary cap, it is far better to have pick 51-70 (totalling 2872 points) than to have pick 1 (3000 points).

BUT (that word again) we don't have unlimited list space, and there is a salary cap. This is where the AFL points system breaks down. A top 10 pick is more valuable than the collection of lower draft picks because it is statistically proven to have a higher success rate than any 1 lower pick, which has to occupy one of the limited spots on the club's list.

For all clubs the potential to get 1 star player (of which the chances are significantly higher with a top 10 pick than elsewhere) outweighs the opportunity to pad their list with 3-4 lower percentage prospects. That is exactly what we are about to do with Mills, and every club would do the same if they were given the opportunity. This list management reality is what the AFL have not taken into account in their points allocations.

Picks 33, 36, 37 and 44 are worth pick 3 mathematically. They are not worth pick 3 in reality. While the above graph demonstrates the statistical performance of players at those draft selections, in practicality the drop in value should be steeper than it currently is.

What does this mean for list management?

Even though these worked examples were addressing a hypothetical question about the new bidding system (particularly topical for us this year), they also give you a lot of information about the success rates of various draft picks and what a realistic outcome is.

In a perfect world, with a competent recruiter, this is how an AFL club should build their list:

1) Go deep into the draft
2) Rapidly identify the players that cannot become above average quality at AFL level
3) Aggressively and quickly cut those players from your list
4) Go deep into the draft

In a perfect world a club should be taking at least 6 players each year in the draft. They have minimum 2 year contracts, meaning that over a rolling 2 year period you would have a minimum commitment to 12 players in your squad of 44. This allows for an established core of 32 players, sufficient to meet the magic number of 28 (see last episode) competent players with a few spares. Statistically 1-2 of your 6 picks will be strong (who will migrate into that senior core), and the rest should be immediately cut at the end of their 2 year contracts.

When every draft pick you take in the top 60 has a 10-20% chance of becoming a strong AFL player, you want to maximise your turnover and the number of rolls of the dice you get. Holding onto players outside your best 28 who are nothing more than mediocre or list cloggers provides very limited practical value (remembering that if the players 28+ are playing regularly, you are already stuffed) and means that you are sacrificing chances to have another dip at the above odds.

 

[MarkP]

What do you mean by go deep into the draft - just want to make sure I understand the point it precisely.

 

[Millky95]

What do you mean by go deep into the draft - just want to make sure I understand the point it precisely.

I think he means, take a pick in each round say 1-6 instead of only using your 1st 3 or trading up so you have only 3 picks to use

 

[RW]

Good to see the Craig Bird measure lives on

Overall, nice analysis. I suspect the reason why the AFL has assigned the values that it has to later draft picks is because of the higher bust rate. Yes, they are probably undervaluing those late picks, but perhaps part of the reason is there's a bigger chance that you'll end up with nothing? That is, the later you go in the draft, the greater the chance that you'll end up with a player who doesn't play a single AFL game.

 

[RW]

Picks 33, 36, 37 and 44 are worth pick 3 mathematically. They are not worth pick 3 in reality. While the above graph demonstrates the statistical performance of players at those draft selections, in practicality the drop in value should be steeper than it currently is.

I feel this is one aspect of the system that actually helps us, since we are one of the teams that almost always finishes in the top half of the ladder and under normal circumstances wouldn't be able to get pick 3 without trading away a quality player. This gives us a mechanism to trade our way up in the draft order, as it were.

You are right about the valuations, but realistically it is the only way to ensure the Academy clubs can still get priority access to their players in a year where there is a highly rated prospect such as Mills or Hopper, otherwise we could have been placed in the ridiculously unfair position of having to give up more value just to be able to draft one of our own Academy prospects. I suspect we would not have agreed to a system that would have put us in that position (what club in their right mind would!), as it would have taken away the incentive for us to develop Academy players at all.

 

[swansfan51]

I feel this is one aspect of the system that actually helps us, since we are one of the teams that almost always finishes in the top half of the ladder and under normal circumstances wouldn't be able to get pick 3 without trading away a quality player. This gives us a mechanism to trade our way up in the draft order, as it were.

You are right about the valuations, but realistically it is the only way to ensure the Academy clubs can still get priority access to their players in a year where there is a highly rated prospect such as Mills or Hopper, otherwise we could have been placed in the ridiculously unfair position of having to give up more value just to be able to draft one of our own Academy prospects. I suspect we would not have agreed to a system that would have put us in that position (what club in their right mind would!), as it would have taken away the incentive for us to develop Academy players at all.

I agree completely. Every club in the league would give up 3-4 lower selections for a chance at the top 3, it is a fantastic advantage and a great ability for us to use.

In case I don't get time to post Episode 3B before the draft, here is the TLDR:

1) Should you ever draft for positional need over talent? No.
2) (Swansrule100 question) What about first round picks on mature age players? Always worth a look if they are good enough and not a reach. Lower bust rate.
3) (Swansrule100 question) The other thing id be interested in is the trading of first round picks, has it been a win in the past? Our history - absolutely yes (but this is a comparison with the poor performance when we use the picks). For other clubs - mixed bag. One TLDR part of the answer is be wary tradeing for other clubs' 1st former round picks unless they have a big discount, success rate is low.
4) (RW question) Can you can always find picks in the 30-40 range (and later) who end up playing more games than those picked in the first round? Yes.

 

[Wayne Swan]

Fabulous analysis yet again, a very big well done.

The idea of turning over players after two years is interesting. To add to that you can turn over rookies every year as they are on 1 year contracts.

The aim of this fast and significant turnover is to build the latter part of your list with gems from this process rather than have list cloggers in those mid 20s out spots. A team that already has their mid 20s and out stocked with gems/potential gems is GWS. That juggernaught is about the steamroll the comp and watch Eddie and Newbold complain about the concessions they got/are getting once they start dominating. GWS is coming in 2016.

 

[melbournehammer]

one question I am always interested in is whether it is the size of the fight in the dog or the size of the dog that matters.

ie as recruiting has become more scientific is a good player with extreme levels of competititveness better than a naturally talented but slightly more reclusive player - ie do you get more from a bryce gibbs than a luke parker (i know what we will all say about that but we need a better example)

 

[Tedeski]

This is nothing new.
Paul Roos identified when he took over the reigns, that in the Sydney market, top 10 picks in the draft were worth far less to us than they would to teams in pro AFL states due to the high risk we faced of losing such a player back to his home state, on the back of a high probability of homesickness.

We became the masters at picking mature age players with our higher picks that filled a required role for our club. Richards, Shaw, Jolly, Mumford etc.

Roos put greater value on using pick 15 or 16 on a mature player with known quantities than he did on pick 8 or 10 for a young player with unknown quantities.
Longmire has continued the same pathway.

Picking up Heeney with pick 18 last year was a steal!
Trading pick 14 (& Bird) in order to gain Mills was just pure brilliance by our recruiters. Mills is the prize we want. We don't value pick 3 like a struggling Vic club values pick 3, but we have somehow managed to pull off, for the second year running, the best trade of the year in my opinion. Where as in the past we have recruited a Josh Kennedy as a 22 year old from another state & with the potential to lose him on the back end of his career, we have now picked perhaps the best talent NSW has ever seen two years running. Still early days I know but of any of the top 5 players to be drafted in the last two years, Heeney & Mills are the two most prepared to begin their senior careers in their first year in the AFL.

Thanks to the academy, they have learned our game style early, thus allowing us to get the most out of these boys for much longer than many of the other top 5 picks.

So even if pick one is bid on Mills, we are a mile ahead in that we have a known quantity on a young gun for a change, something that we never knew when Roos took over as coach. Not only that, we will never, ever lose these boys to homesickness. Count our blessings & pray that we can develop a few more before Eddie & Newbitch shut down our academies after having to watch our academy stars tear their bottom 4 teams new a..eholes for years to come.

Sorry for being so positive but I ran out of lemons!

 

[swansfan51]

Picks 17-20 average 72.25 AFL games
Picks 21-24 average 53 AFL games
Picks 25-28 average 53.75 AFL games
Picks 29-32 average 62 AFL games
Picks 33-36 average 47.25 AFL games
Picks 37-40 average 64.25 AFL games
Picks 41-44 average 49 AFL games
Picks 45-48 average 50 AFL games
Picks 49-52 average 38 AFL games
Picks 53-56 average 52 AFL games
Picks 57-60 average 30.5 AFL games
Picks 61-64 average 32.25 AFL games
Picks 65-68 average 28.5 AFL games
Picks 69-72 average 30.25 AFL games

Note the very significant gap between pick 20 and the rest, and then how little difference there is in picks 21-56.

Further evidence in support: the AFL's salary data. Look at the data point for pick 21 and the data points in the early 50s, virtually identical. Other than that spike in the 36-37 range, no significant difference.

 

[Lakeside Swan]

Trading pick 14 (& Bird) in order to gain Mills was just pure brilliance by our recruiters.

I agreed with everything you said, except for this. I think you're missing the point in what I and others have been complaining about. No one is complaining about us trading Bird and pick 14. We're complaining about trading them for nothing.

If we wanted to offload Bird we should have done it for a fifth round pick (i.e. nothing) and then traded pick 14 in another trade. We were in a position of strength when we possessed pick 14. Yet, what did we do? We swapped it for two picks (23 & 44) which gave us no increase in points. It was a complete waste. Plenty of clubs would have been happy to trade two second round picks for pick 14, especially in a shallow draft.

Look at the trade that GWS did with Melbourne. That's what we should have done.