Swans Academy and player watch

[liz]

We'll need to see how it pans out, but I interpret the very short description as using a statistical analysis to value picks, not draftees. I suspect what they are trying to do is come up with a formula for, say, pick 5=pick 18 + pick ?, not a formula to assess what pick a Heeney or Moore is worth. That assignment of draft value is either going to have to remain with a bidding system or be a matter of complete judgement by a panel of "experts".

There are still numerous difficulties involved in such a system. While it might be feasible to assign a value to a top 10 or even top 20 pick on the basis of using past picks to estimate the probability of a player taken in those ranges "making it", the problem will get much harder for later picks where there will be a high degree of dispersion. Also, how do they take into account that some draft pools are deeper than others, that some have a higher quality top tier, the impact of injuries on whether players "made it", the impact of development systems at different clubs, the impact of luck (ie which players never made it because an opportunity never arose when they were at their clubs), or even how you judge whether a player even made it. Will it just take into account number of games played, or superimpose some measure of the quality of those games?

[dimelb]

The formula is TAC Supercoach ratings divided by the number of Instagram followers times the number of selfies.

And all because Fat Bastard from Collingwood thinks the sky is falling. It would never have happened if Heeney wasn't in the mix. Kneejerk reaction to attempt a short-term "solution" which is not needed. These things tend to even out over time.

[The Big Cat]

And all because Fat Bastard from Collingwood thinks the sky is falling. It would never have happened if Heeney wasn't in the mix. Kneejerk reaction to attempt a short-term "solution" which is not needed. These things tend to even out over time.

Exactly. If there were no academy stand outs and we were still on the bottom of the ladder Eddie would be praising the academy concept. And Buddy and Tippett in Sydney would be for the good of the game. No one have whinged about the Pies swooping on Quentin Lynch!

[Ludwig]

We'll need to see how it pans out, but I interpret the very short description as using a statistical analysis to value picks, not draftees. I suspect what they are trying to do is come up with a formula for, say, pick 5=pick 18 + pick ?, not a formula to assess what pick a Heeney or Moore is worth. That assignment of draft value is either going to have to remain with a bidding system or be a matter of complete judgement by a panel of "experts".

There are still numerous difficulties involved in such a system. While it might be feasible to assign a value to a top 10 or even top 20 pick on the basis of using past picks to estimate the probability of a player taken in those ranges "making it", the problem will get much harder for later picks where there will be a high degree of dispersion. Also, how do they take into account that some draft pools are deeper than others, that some have a higher quality top tier, the impact of injuries on whether players "made it", the impact of development systems at different clubs, the impact of luck (ie which players never made it because an opportunity never arose when they were at their clubs), or even how you judge whether a player even made it. Will it just take into account number of games played, or superimpose some measure of the quality of those games?

I think you're right about this. It does seem that they will try to come up with a system to value each pick as opposed to the player and perhaps allow the bidding system to determine the value. I'm sure there will be statistical anomalies, like pick 9 may end up producing better players than pick 8, but that can be smoothed out. Some reasonable basis could be worked out to produce a draft pick value curve, which would be a good tool in a completely random and arbitrary system.

The more difficult part is valuing the investment part of the equation. For example, if a club spends $1 million a year on an academy program, how many draft value points does that buy you? And it would seem only fair that the investment value should start from the beginning of the academy program. So the Swans may have invested $4 million already before getting any player out of the system. This investment value should be bankable, so it can be 'spent' when high value draft picks arise in a given year. I don't think the system can work fairly by isolating individual years.

The equation would have to equate the value of an investment in a given player-generating program with the output. Some determination will have to be made as to how much value, represented by the number of places one moves up in the draft value, is bought by investing a particular amount of money in a program. So, does a $500,000 investment in an academy buy you a pick 5 player with your pick 15, for example.

Although I can see something emerging that could achieve fairness to some extent, I doubt if much or anything will be gained by replacing the current system with such a convoluted system which will surely be criticised from many quarters. In the end, if done fairly, will probably produce the same results as under the present bidding system.

All this because the Swans have been too clever and managed to squeeze some goodies out of the COLA and perhaps the academies and some other clubs don't like it. But these same clubs are happy to keep their cherished perks, like fixture preference, and jealously guard these advantages. Why don't we have a system where fixture value is put into a statistical framework and advantages are paid for in draft placement. I'm sure Collingwood would love that one.

[liz]

I can't see any significant discount being given to compensate for the degree of investment. The non-academy clubs would still argue that they don't have the opportunity to chose to make that investment to the extent we do, while it would be close to impossible to measure how much clubs invest in development of father son prospects because none will ever have a programme on the scale of the northern academies, even if they do set up some programme for their prospects.

Instead, other clubs will argue that, even without a discount, the northern clubs still have an advantage if they draft academy players because they have had a few years, pre-draft, to mould them into players to suit their needs, and that these players will hit the ground more quickly when drafted because they are familiar with the club and know something of its game style.

[dimelb]

I agree with Liz that the AFL is looking for some sort of mathematical formula that evens out the acquisition of talent available from the draft.

The insuperable problem is the variability of the draft from year to year, not to mention the frequent phenomenon of the first round pick that turns out to be a quite ordinary player. To try overcoming these difficulties by including some form of compensation at a later date is to amplify the complexity without any guarantee that it will work better than before.

The most foolish act of all would be to apply the new system to the next draft as a way of seeing how it works. Remember how long it took people to work out the best way to use the substitute? And that was a much less complex proposition. If they want to try a trial run they would do better to run a shadow draft behind the actual draft, in accordance with their new system and check the results down the track. Only in some such way can you have the benefit of trial without the most potentially damaging error.

[Ludwig]

I can't see any significant discount being given to compensate for the degree of investment.

But isn't this the crux of the issue? Clubs with academies have diverted funds from other programs to get some benefit out of this program, and they have been encouraged to do so because it also has a benefit to the AFL as a whole. If the AFL take all the value out of the program, then why continue?

Maybe we'd be better off sending a full time scout to Tonga.

[Steve]

I still can't see how they can accurately assign value to each draft pick which would stack up equally over time. Some drafts are considered shallow where picks 1-10 are great but anything 25+ is mostly rubbish, but in others little separates pick 10 and pick 40.

If they use trading history where picks are bundled like the Collingwood-West Coast example that was quoted, that doesn't take into account the players those clubs were looking at and thought they'd still get with their swapped picks, and the fact that those were the only picks those clubs had to swap.

Just thinking logically, if an Academy player is actually the standout best player and a club bids pick 1, and we win the flag and start with pick 18, how could any of our picks satisfy the 'fair value' system? In previous years you'd have traded picks 18, 36, 54, 72, 90 and however many more into the 100's to get pick 1 in a heartbeat, and have been getting a good deal.

The 'current proposal' also would seem to poorly cater for multiple academy and/or father-son picks in the same year. If there are 2 or 3 good players eligible, and we have late picks in each round, we'll need to use 20 picks to match the value they equate to 2 or 3 good players.

The credit they are suggesting also sounds OK in theory but no-one will be happy with it - either seen as too generous by VIC clubs, or negligible and therefore pointless by the northern clubs.

[Ludwig]

There's a headline on AFL website about Darcy Moore titled: Pies set for draft steal

[ShockOfHair]

I'm thinking the AFL will come up with some dazzlingly incomprehensible formula just to troll McGuire.

[Ludwig]

I still can't see how they can accurately assign value to each draft pick which would stack up equally over time. Some drafts are considered shallow where picks 1-10 are great but anything 25+ is mostly rubbish, but in others little separates pick 10 and pick 40.


The credit they are suggesting also sounds OK in theory but no-one will be happy with it - either seen as too generous by VIC clubs, or negligible and therefore pointless by the northern clubs.

They will concede that they can only achieve a statistical norm for the distribution of draft pick success over a period of time, and every year would produce variations from that norm. In a simple distribution analysis, they could plot draft pick number with games played over a long period of time, just an example of one of perhaps several criteria.

I agree with your point on the credit that it's unlikely to satisfy anyone unless it is blatantly unfair. In the end, if it's between making Eddie happy or the northern clubs happy, we know where the AFL will go.

[Conor_Dillon]

At the end of the day, if Sydney have to give up every single one of their draft picks in the next 2 drafts for Heeney, Davis, Mills and Dunkley then we are definitely coming out on top given that the best possible pick will be in the 16-18 range - and surely that is the worst case scenario?