Academy points system

[faible]

In an ideal system, should picks have points values that reflect their actual value?

The value also hinges on how many picks a club is planning on using; giving away more points via multiple later picks for few early ones, makes sense where some of those later picks were never intended to be used.

So if (say) Sydney traded pick 12 for later picks worth more points; it benefits Sydney through the acquisition of points allowing them to match a bid up to (say) Pick 5 instead of Pick 8. Whilst Adelaide goes in with a higher pick in exchange for a premium of points, made up of some picks they intended to use, and others they were never planning on taking.

Technically this does overvalue Pick 12, but part of the value is made up from picks where the club doesn't intend on taking them anyway. How do you then quantify the value of a Pick 40 that was never going to be used? Is it worth 429 points, or 0?

 

[Peel67skin]

The value also hinges on how many picks a club is planning on using; giving away more points via multiple later picks for few early ones, makes sense where some of those later picks were never intended to be used.

So if (say) Sydney traded pick 12 for later picks worth more points; it benefits Sydney through the acquisition of points allowing them to match a bid up to (say) Pick 5 instead of Pick 8. Whilst Adelaide goes in with a higher pick in exchange for a premium of points, made up of some picks they intended to use, and others they were never planning on taking.

Technically this does overvalue Pick 12, but part of the value is made up from picks where the club doesn't intend on taking them anyway. How do you then quantify the value of a Pick 40 that was never going to be used? Is it worth 429 points, or 0?

Isn't it you need to delist players on your current list for multiple later picks . If they consist of point .

 

[Suma Magic]

The value also hinges on how many picks a club is planning on using; giving away more points via multiple later picks for few early ones, makes sense where some of those later picks were never intended to be used.

So if (say) Sydney traded pick 12 for later picks worth more points; it benefits Sydney through the acquisition of points allowing them to match a bid up to (say) Pick 5 instead of Pick 8. Whilst Adelaide goes in with a higher pick in exchange for a premium of points, made up of some picks they intended to use, and others they were never planning on taking.

Technically this does overvalue Pick 12, but part of the value is made up from picks where the club doesn't intend on taking them anyway. How do you then quantify the value of a Pick 40 that was never going to be used? Is it worth 429 points, or 0?

Picks can always be traded for future picks, so they always have value.

 

[faible]

Isn't it you need to delist players on your current list for multiple later picks . If they consist of point .

They did bring in a rule about stock-piling more picks than available list spots, but between DFA / Rookie Upgrades / Flexible list sizes, you don't have to take every draft pick.

Picks can always be traded for future picks, so they always have value.

But is the value fixed? If Adelaide aren't planning on taking Pick 40 this year, that doesn't mean it holds the same value to them that it would hold to someone else.

 

[Suma Magic]

They did bring in a rule about stock-piling more picks than available list spots, but between DFA / Rookie Upgrades / Flexible list sizes, you don't have to take every draft pick.



But is the value fixed? If Adelaide aren't planning on taking Pick 40 this year, that doesn't mean it holds the same value to them that it would hold to someone else.

Value is always contested.

I'm just pointing out that if they weren't going to use it they could trade it for a future pick of similar value. So picks always have value, with higher picks having more value.

 

[faible]

Value is always contested.

I'm just pointing out that if they weren't going to use it they could trade it for a future pick of similar value. So picks always have value, with higher picks having more value.

Agreed; the issue being that value is very subjective. So whilst the AFL could re-value the points system, it would result in approximately the same outcome - clubs will trade up / down depending on needs and what's available.

We've already seen Port pay a premium to secure a pick that is likely (but not certain) to give them access to one of the three SA players, and that's completely outside of any bidding system.

Whereas in 2016 Port traded in extra lower picks with seemingly a goal in mind of securing more players, at the cost of their 2017 first round pick.

 

[Suma Magic]

Agreed; the issue being that value is very subjective. So whilst the AFL could re-value the points system, it would result in approximately the same outcome - clubs will trade up / down depending on needs and what's available.

We've already seen Port pay a premium to secure a pick that is likely (but not certain) to give them access to one of the three SA players, and that's completely outside of any bidding system.

Whereas in 2016 Port traded in extra lower picks with seemingly a goal in mind of securing more players, at the cost of their 2017 first round pick.

There will always be two sides to a trade.

But when academy f/s clubs can consistently trade high picks for lower picks of much greater value, it shows that there is a systematic over-valuation of lower picks from a points perspective.

In theory, higher picks should be able to be traded for lower picks of a similar value, on average over time.

 

[faible]

In theory, higher picks should be able to be traded for lower picks of a similar value, on average over time.

If the trade wasn't unequal it wouldn't occur; so for Sydney why would they trade 13 for 26, 28 and 40 if the points value was identical?

 

[Suma Magic]

If the trade wasn't unequal it wouldn't occur; so for Sydney why would they trade 13 for 26, 28 and 40 if the points value was identical?

They probably wouldn't.

But the point here is the magnitude of the points difference. They got over 600 extra points from the deal, as the points system over-values lower picks.

If the points system was different and lower picks were more accurately valued, and they had got say 200 extra points, they still would have done it. They just would have missed out on a huge windfall gain. And the blues and crows would have got exactly the same benefits. The only difference was that sydney would have less good live picks after matching blakely, which would indirectly benefit all other clubs.

 

[Taylor]

Do you think that picks should be assigned points that align with their market value?

How do you do that across years? At the top end of the draft the draft pick isn't a number, it's a player. That level changes between years as the quality of the top end changes.

It's hugely subjective. I think the picks at the top will always be worth more because the talent is better there but that level of talent fluctuates.

 

[Suma Magic]

How do you do that across years? At the top end of the draft the draft pick isn't a number, it's a player. That level changes between years as the quality of the top end changes.

It's hugely subjective. I think the picks at the top will always be worth more because the talent is better there but that level of talent fluctuates.

It's what the current system already attempts to do. It can never be 100% accurate, it is just the best attempt at a system where points reflect the value of picks over the long-term.

If trades are consistently leading to more points for the side giving up the highest pick, then the system is over-valuing lower picks. If the opposite is true, the system is over-valuing higher picks.

All data indicates that the current system is flawed as it over-values lower picks. An easy adjustment (reducing the value of lower picks) would better align the system to its original intent, namely to allocate points based on the actual value of picks.

 

[Taylor]

It's what the current system already attempts to do. It can never be 100% accurate, it is just the best attempt at a system where points reflect the value of picks over the long-term.

If trades are consistently leading to more points for the side giving up the highest pick, then the system is over-valuing lower picks. If the opposite is true, the system is over-valuing higher picks.

All data indicates that the current system is flawed as it over-values lower picks. An easy adjustment (reducing the value of lower picks) would better align the system to its original intent, namely to allocate points based on the actual value of picks.

I think the extra points are a market valuation of the difference between a top ten player and a top twenty player, not that the lower picks are too highly valued by points.

 

[Andre]

I've been clogging up other threads with this topic. Which can be annoying so I thought I'd start a thread.

The AFL aimed to allocate points to picks, that accurately reflect their actual value:

"We have found that the DVI is a good indicator of the general value of picks relative to one another and also agrees with trades actually executed by Clubs."
http://www.afl.com.au/staticfile/AFL Tenant/AFL/Files/Father-son-bidding-system.pdf

However, this hasn't proved to be the case in reality. In the vast majority of pick for pick trades, the team with the higher pick will get more points in return. This shows that the system systematically over-values lower picks, as compared to the real market.

Which brings me to the premise of this thread -

In an ideal system, should picks have points values that reflect their actual value?

For those who would answer no:

• how should their value differ from the market value?
• why should they differ?
• how much should they differ?
• should high or low picks be over-valued?

I get that this topic isn't everyone's cup of tea, but I was surprised by the number of posters who prefer a system that doesn't allocate points values in line with the actual market values of picks. So I was interested in finding out more about this way of thinking.

This is trying to band-aid over the half done implementation of points. The problem isn't that points have values, with trades not done to try and equalise them. The problem is the AFL should have ditched picks, as they currently exist, altogether.

How it should have been (and should still be) implemented.
1. Clubs get an allocation of draft points based on ladder position. To keep it simple, this would be the same as the points they'd get based on the picks they currently would get for their ladder position (assuming no F/S or Academy bids, picks traded in or out - ie. 'Clean').

2. Clubs instead of trading picks for players or future picks, would be able to trade players for points or future points. E.g. Shiel to Essendon for 2000 points from Essendon to GWS as a theoretical trade. Or Polec from Port to North for 500 points from 2018 and 250 from 2019. Or even if a club thinks the next years crop has more of the type of player they need next year, whilst another club prefers this years crop, then a 500 points from 2018 for 500 points from 2019 between two clubs.

3. Draft night changes from clubs getting a pick and no one can trump that to a sequence of player auctions. Ie. each club in reverse ladder order gets to nominate a player and an opening bid of points. So take 2018, Carlton nominates Walsh with a bid of 1000 points. Any other club can choose to bid higher (provided they have enough points left in their allocation). Bidding on a player continues (with a club able to bid multiple times, as long as each bid is higher than the last bid by another club) until there is only one club left. That club has their player and the point total they did in the final bid comes off their points total, for use on other players. The interesting one here is if someone is willing to pay more for Walsh than Carlton, then Carlton don't get to put forward another player until all the other clubs have, but of course can bid on any other clubs put forward.

4. Father / Son and Academy bids would change in that any club could put a player forward and have the chance to get them. What the F/S and Academy players would give for clubs is a 25% discount on what comes off their total. So if a club bids on a F/S with a winning bid of 1000 points, only 750 would come off their total. This still gives an advantage to the clubs, but not the current one of an absolute superstar F/S or Academy player picked up for a fair bit under.

5. The draft would continue with auctioning until all clubs have filled all the spots they want to (once a club is out of points they'd only be able to pick up players they put forward that no club with points still remaining don't want). The rookie draft would work in reverse draft order and F/S, Academy picks to those clubs as per now.

This addresses several issues.

1. The identified mismatch in pick values.
2. Clubs trading good picks for a bunch of lower picks to game the F/S and Academy bidding system as it currently is.
3. The impasse at trade week of a player worth pick 20 wanting to go to a club with picks 15 and 25, where their new club won't trade 15 and their existing won't take 25 and it takes future picks or trying to find 3rd or more parties to make it work. With this system it's just agree a points value and no need to unnecessarily involve other clubs.
4. Clubs trading future picks and then having injuries or other reasons for dropping unexpectedly down the ladder and trades costing more than predicted.

 

[Suma Magic]

I think the extra points are a market valuation of the difference between a top ten player and a top twenty player, not that the lower picks are too highly valued by points.

I don't quite understand what you're saying.

I'm saying that if a high pick is traded for a couple of lower picks, then at the moment the low picks have the higher points value. This suggests that teams don't actually value low picks as much as the points system suggests that they should. So lowering the value of lower picks will mean that trades will be more even from a points perspective.

 

[Taylor]

I don't quite understand what you're saying.

I'm saying that if a high pick is traded for a couple of lower picks, then at the moment the low picks have the higher points value. This suggests that teams don't actually value low picks as much as the points system suggests that they should. So lowering the value of lower picks will mean that trades will be more even from a points perspective.

I was saying that the trade of the better pick for two worse picks with a higher points value is the market adding value to better pick on account of the higher quality players there.

Pick #6 (1751) for Pick #18 (985) and #19 (948)
1751 for 1933

The market has decided that the better pick is worth more than the points because it will take a specific player, I don't think it shows the later picks have too high a value.

 

[Suma Magic]

I was saying that the trade of the better pick for two worse picks with a higher points value is the market adding value to better pick on account of the higher quality players there.

Pick #6 (1751) for Pick #18 (985) and #19 (948)
1751 for 1933

The market has decided that the better pick is worth more than the points because it will take a specific player, I don't think it shows the later picks have too high a value.

But if higher picks are systematically being traded for lower picks of greater points value, doesn't this say that the points system isn't achieving its intention of allocating points relative to the value of picks?

 

[Taylor]

But if higher picks are systematically being traded for lower picks of greater points value, doesn't this say that the points system isn't achieving its intention of allocating points relative to the value of picks?

I do see what you're saying, but those academy picks get a discount too..

 

[Suma Magic]

I do see what you're saying, but those academy picks get a discount too..

You can/should ignore the discount though, for the purposes of working out a points system/curve that best aligns to actual pick values. As Academy/f/s get the discount regardless of what trading they do.

 

[liz]

But if higher picks are systematically being traded for lower picks of greater points value, doesn't this say that the points system isn't achieving its intention of allocating points relative to the value of picks?

Only if you're fixated on market value, rather than intrinsic value.

Have you reads Footballistics? This was published earlier this year and contains around a dozen or so chapters mixing anecdotal commentary with quantitative analysis on various aspects of football. Its chapter on the draft concludes that the AFL's points index actually overvalues higher picks and undervalues later picks, based on their measure of the intrinsic value of picks (ie the actual output of players taken at various points in the draft).

You'd need to read the book yourself to make your own assessment of how credible you think their approach and conclusions are but, on face value, both the lead writer and the quantitative analysts who supported him appear decently credible based on their education, experience and the professional fields in which they work.

 

[Suma Magic]

Only if you're fixated on market value, rather than intrinsic value.

Have you reads Footballistics? This was published earlier this year and contains around a dozen or so chapters mixing anecdotal commentary with quantitative analysis on various aspects of football. Its chapter on the draft concludes that the AFL's points index actually overvalues higher picks and undervalues later picks, based on their measure of the intrinsic value of picks (ie the actual output of players taken at various points in the draft).

You'd need to read the book yourself to make your own assessment of how credible you think their approach and conclusions are but, on face value, both the lead writer and the quantitative analysts who supported him appear decently credible based on their education, experience and the professional fields in which they work.

What the system needs to achieve is the ability for teams with Academy or f/s picks to be able to trade their picks with other clubs. In doing deals that other clubs see as fair, the aim is for the points equation to be fairly similar.

An example of a failed system is where there is a consistent ability for clubs with academy and f/s picks to trade for large increases in points.

I haven't read the book, but I would guess they are right.

Lets say that the book is right, and that the points system over-rates high picks from an intrinsic value perspective. Given that all market value data suggests the opposite, this means that clubs consistently place a market value on high picks which is much higher than its intrinsic value. Based on this information:

• teams without academy or f/s players will be happy to trade up using lower picks, as teams generally are keen to do this if the price is right
• given the difference between market and the intrinsic values, teams with high picks that would be used to match academy or f/s bids can consistently trade down for lower picks of a greater points value. This is possible due to the gap between intrinsic and market value that exists, with the teams trading up willing to give up more for high picks than the intrinsic data tells us that they should.
• As a result, not only can teams with academy and f/s players get the 20% discount, they can consistently also get a further indirect discount by their ability to increase the points they have to match, sometimes significantly.

This indicates that market value, rather than intrinsic value, should be used to determine the points values of picks.

 

[liz]

This indicates that market value, rather than intrinsic value, should be used to determine the points values of picks.

Whether market or intrinsic value is a better goal is a matter of opinion, not fact. For reasons I have already given earlier in this thread, I think market value is impossible to determine because there are simply too few trades in a system with too many constraints. Intrinsic value brings its own issues - particularly that there is variation from year to year - but it is at least possible to come up with a systematic way to try and estimate it, at least on a long term, mean or median basis.

If some market transactions offer a wildly different assessment of value to measures of intrinsic value, this likely indicates a degree of irrationality in the market, as well as reflecting the lumpiness of this particular market.

And to an extent, the constraints operating around the draft and trade markets (particularly list sizes) also provide some explanation for the differences.

 

[Suma Magic]

Whether market or intrinsic value is a better goal is a matter of opinion, not fact. For reasons I have already given earlier in this thread, I think market value is impossible to determine because there are simply too few trades in a system with too many constraints. Intrinsic value brings its own issues - particularly that there is variation from year to year - but it is at least possible to come up with a systematic way to try and estimate it, at least on a long term, mean or median basis.

If some market transactions offer a wildly different assessment of value to measures of intrinsic value, this likely indicates a degree of irrationality in the market, as well as reflecting the lumpiness of this particular market.

And to an extent, the constraints operating around the draft and trade markets (particularly list sizes) also provide some explanation for the differences.

Market is better, assuming perfect information.

Market data has showed that the intrinsic approach is systematically flawed.

It doesn’t have to be one or the other. Market data can be used to alter intrinsic values. The AFL checked the intrinsic values with market data, and they thought they aligned.

But looking at points for points trades over 8 years they don't align - in fact the team with the higher pick got an average of 27% more points than they gave up. The team giving up the higher pick got points in 13 of 15 trades. It may be a small sample size but the trend is overwhelmingly clear. It's clear that the system isn't working as intended and there is clear scope to improve things via a tweak to the points system.

 

[Blood red]

How do you do that across years? At the top end of the draft the draft pick isn't a number, it's a player. That level changes between years as the quality of the top end changes.
It's hugely subjective. I think the picks at the top will always be worth more because the talent is better there but that level of talent fluctuates.

I would suggest that the academies and the f/s clubs know the quality (or lack of) of the following years future players.
They have been watching these kids since they were 11, with the ability to compare and predict the clubs positional needs.
Makes sense to build a points/picks war chest if the following year's perceived talent is greater than the current year.

 

[old55]

It's what the current system already attempts to do. It can never be 100% accurate, it is just the best attempt at a system where points reflect the value of picks over the long-term.

If trades are consistently leading to more points for the side giving up the highest pick, then the system is over-valuing lower picks. If the opposite is true, the system is over-valuing higher picks.

All data indicates that the current system is flawed as it over-values lower picks. An easy adjustment (reducing the value of lower picks) would better align the system to its original intent, namely to allocate points based on the actual value of picks.

It's possible that clubs will only trade down if they get a premium in points value. So if you reduce the value of lower picks, the clubs trading the earlier pick will demand even earlier late picks in the trade to retain this premium. You'll be chasing your tail. It's a free market now and clubs are willingly trading picks on the current points basis.

 

[Suma Magic]

It's possible that clubs will only trade down if they get a premium in points value. So if you reduce the value of lower picks, the clubs trading the earlier pick will demand even earlier late picks in the trade to retain this premium. You'll be chasing your tail. It's a free market now and clubs are willingly trading picks on the current points basis.

I would argue that most clubs are competing for the limited number of high picks held by clubs with highly rated academy or f/s picks each year (who currently seek to trade down to increase picks). So if the points value of later picks were made relatively smaller, then i'm not sure it would impact what clubs would be willing to give up to get higher picks. They are already competing with other clubs who have no interest in points or points values, they are only interested in how they value high picks compared to low picks.

Making later picks worth relatively less would just mean that the club with the high pick would be able to improve their points position by less. Or if they couldn't improve points at all then they would just keep the pick and use it to match a bid.