Building a perfect ratings system

[Shrews]

I've been reading the ratings with interest and was wondering if there was a place for a more quantative approach based on statistics rather than opinion.  Rather than taking the place of opinion ratings, which are great and will always add value (however good or bad that opinion may be), it could run alongside opinion ratings and would add value through eliminating bias and any incorrect or limited knowledge.

The overall aim for a perfect ratings system would be to achieve a better score than the overall winner without the need for any trades.  Now that is almost certainly impossible, but is something to aspire to nevertheless.

If a perfect ratings system could be developed what factors would be included?  There are of course the obvious ones such as:

Last years score
Last years average
Previous years score
Age
Value for money

But then less obvious factors such as

% spend position (i.e 29% on backs, 29% on mids, 10% on rucks, 31% on forwards = 359k average to spend on backs (each), 440k on mids, 394k on rucks, 386k on forwards)
Draw (fixture) bias (how are we rating this at the moment Adelaide best followed by Sydney and St.Kilda ?)
Potential for injury

......and i'm sure you guys who have played the game for years could come up with all sorts of obscure factors that can affect the scores and therefore need considering before the first game begins.

Once the factors have all been highlighted, they would need weighting on importance and then it's just a formulaic process to arrive at a system of rating teams.

I'd give this a go, but i'm new to the game and don't really know how the game is played that well.  What I need is a load of help, so if you think it's an idea that would add value to the forum, then just let me know below.  If not, then i'll give it a go alone, let you know what team was created just after kick off and see how it does.  It will either go three ways:

1. Be exceptional and everyone will want their team to be rated using the method next season
2. Will show potential and will be worth fine tuning
3. Will be showere and have no worth at all.

If you think it's worth a go then give me your opinions below and i'll see if we can turn those opinions into a numerical process

[yorgis]

stats only go on past performances, and give no indication as to what type of role each player will have for the coming

year especially with a new coach. that being said a good player is a good player, no matter where he plays. stats are stats and

i think they just indicate how consistent a player can be with no variables

[Shrews]

True and maybe they can be called 'true stats', but opinions can also be turned into stats.  For example if I ask this question as a starting point, hopefully I will get a number of opinions that I can turn into a stat.

Who will be the top 10 point scorers of players valued 500k or more regardless of position (score 1 for who you think will score most and so on)

Here are the players (23, hopefully that's it):


S.Johnson
Beams
Swan
Pendlebury
Ablett
Boyd
J.Selwood
Rockliff
Mitchell
Murphy
Redden
Gibbs
Stanton
Priddis
Dal Santo
Thomas
Shiels
Judd
Beams
S.Thompson
Cox
Chapman
Franklin

If there's any other player over $400k who you believe will score as much as the one you rank 12 (above) then include his name and put an asterix next to his name.

[bomberboy0618]

I like this a lot. However as you will be aware there are too many unknown variables. Hidden injuries, decrease/increase in tog, positional changes that can't really be taken into account before the season starts(can't trust nab cup for positions etc.)

[Shrews]

I like this a lot. However as you will be aware there are too many unknown variables. Hidden injuries, decrease/increase in tog, positional changes that can't really be taken into account before the season starts(can't trust nab cup for positions etc.)

This is true, but even that can be put into a statistical format based on opinions.  We are all forming an opinion now to make a 'best guess' on who will play.  For example, you mention 'hidden injuries', now I would guess that not many of us will be aware of this, however there may be some of us who are much closer to a team than others.  For example the opinion of a 'Freemantle specialist' may hold statistical weight over the opinion of the masses.  He may be be privvy to information that the rest of us aren't

Formula - 70% of mass total affected by the specialists opinion.  If the specialist finds something out that we don't know then that has a massive sway on the rating of the player.  Even that can be broken down into further stats, for example if the opinion is 'fact' then it holds 70%, if it's a 'strong rumour' then only 50%, if it's just a 'rumour' then 40%.

Now I could do all this myself and create a formula for rating players but I don't think I know enough about the game to make that effective.  This is where I need help.

Yes great to have individual posters rating teams, but working alongside those how about a rating system based on stats, with specialist opinions and data created through as many people as possible.  Surely can only be a good thing

[TheHanger]

just a sidenote about the draw, in the age about a week ago they showed the hardest to easiest draws per team which was done by Champion Data. I cant remember it exactly but i know Collingwood had the Hardest and Melbourne the easiest but Melbourne also travelled the most of any Victorian team anyway thats only a small portion of your idea

[Shrews]

just a sidenote about the draw, in the age about a week ago they showed the hardest to easiest draws per team which was done by Champion Data. I cant remember it exactly but i know Collingwood had the Hardest and Melbourne the easiest but Melbourne also travelled the most of any Victorian team anyway thats only a small portion of your idea

Interesting.  Have you got a link to it ?  It may well be a small portion, (not quite sure what that means yet, 1%, 10% 25%?) but it's a factor nevertheless and is certainly a factor that i've spotted when i've seen others rating teams.

I'll see if I can pull together all the different factors over the next week and then ask for opinions on how much weighting should be placed on each factor. Then average those opinions out to arrive at a figure.

[TheHanger]

I can't find the exact link mate this is the closest i can find http://www.heraldsun.com.au/sport/afl/numbers-add-up-to-a-tough-draw-for-collingwood/story-e6frf9jf-1226262703781 i think if you want to do this seriously you will have to buy the AFL Prospectus

[Shrews]

I can't find the exact link mate this is the closest i can find http://www.heraldsun.com.au/sport/afl/numbers-add-up-to-a-tough-draw-for-collingwood/story-e6frf9jf-1226262703781 i think if you want to do this seriously you will have to buy the AFL Prospectus

If I knew what the AFL prospectus was I might consider it !  ;D

Thanks for the link, i'll take a good look at that.

I think the next stage is to provide a list of factors that might affect a players rating and then work out a weighting from there.  If anyone wants to do add their opinion then obviously 'more the better' really.  Hopefully, i'll have a list of factors by this time tomorrow and they can be weighted by the end of the week.  From there, each player can be 'rated'.  I doubt i'll get time to do them all but certainly the more popular ones