I have developed a formulae for basic differientiation between players/groups of players to determine which is a better option, titled by me as the "Jukes Coefficient". Below is a quote from it's first use, in the SwAblett thread from earlier today.
Quote from: Jukes on December 19, 2012, 10:14:20 PM
I reckon it's good to go with Beams and Cotchin over SwAblett. To find a quick way of valuation for players I like to add together three variables; previous season average (risk evaluation), 2013 predicted average (scoring evaluation) and the difference between these, improvement (cash value) then when comparing take into account the price difference by dividing the price difference by the magic number, 5150, and add that into the equation by adding it to the cheaper option.
I have predicted that Ablett will retain his average at approximately 125 PPG, while Swan will drop 5 points to 128, Cotchin gain 7 average to 118, and Beams 6 average to 122.
Ablett = 125 + 125 + 0 = 250
Swan = 133 + 128 + -5 = 258
Cotchin = 111 + 118 + 7 = 236
Beams = 116 + 122 + 6 = 244
That gives SwAblett a running total of 508, while CotchBeamsy a running total of 480.
Now you add price into the equation. SwAblett has a total of 1,330,600. CotchBeamsy has a total of 1,169,600. This means SwAblett costs $161,000 more than CotchBeamsy. 161000 / 5150 = 31.2621.
SwAblett = 508
CotchBeamsy = 480 + 31.2621
= 511.2621
Meaning CotchBeamsy shows a total of 3.2621 rating points over SwAblett. This may appear quite small seeing as their totals are both over 100, but seeing as each combination are scaled to be swung toward each other (through averages for SwAblett's benefit and pricing for CotchBeamsy's benefit) it is actually quite large.
The formula must be adjusted however for differentiating between mid-priced players.
Jukes Coefficient = Career best average + Predicted average for 2013 + xy
Then factor in price as done in the first formulae
'xy' is a relatively complex variable. To find a players xy, divide their price by the magic number in 5150. Then subtract this figure from their predicted average.
Below I have applied the Jukes Coefficient for Midpriced players for the battle between backline midpricers in Cameron Pedersen and Nathan Bock.
Pedersen: 71.3 + 75 + 17.5 = 163.8
Bock: 87.8 + 80 + 25.09 + 2.6 = 195.49
Bock wins comfortably by 32.41 points, meaning you should select Bock over Pedersen for your team.
Of course both of these formulae for premos and mid-pricers cannot be affected by other variables such as DPP and injury, although that could be factored in if you could be bothered finding an accurate proportion to the other variables.
Thoughts on these formula?
wow! good work jukesy, looks very thorough!
Quote from: Jukes on December 20, 2012, 12:49:27 AM
Pedersen: 71.3 + 75 + 17.5 = 163.8
Bock: 87.8 + 80 + 25.09 + 2.6 = 195.49
Why does Pedo have 3 numbers added up but Bock has 4 ?
Quote from: Scrads on December 20, 2012, 01:00:36 AM
Quote from: Jukes on December 20, 2012, 12:49:27 AM
Pedersen: 71.3 + 75 + 17.5 = 163.8
Bock: 87.8 + 80 + 25.09 + 2.6 = 195.49
Why does Pedo have 4 numbers added up but Bock has 4 ?
The 2.6 is Bock's price advantage over Pedersen's in terms of average. The price difference between them is 13,400. You divide that by the magic number then add the result, 2.6, to the coefficient of the cheaper player.
If anybody doesn't understand how to use the formula, you can request to me players/groups of players from whom you want me to calculate the coefficients for. Gonna be hard juggling rating other teams, doing formula and managing my own side but I'll have my best shot.
Quote from: Jukes on December 20, 2012, 01:22:00 AM
If anybody doesn't understand how to use the formula, you can request to me players/groups of players from whom you want me to calculate the coefficients for. Gonna be hard juggling rating other teams, doing formula and managing my own side but I'll have my best shot.
I will start you off with the change I made to my side.
Swan + Hrovat(or any other mid rookie)
or
Stants + Embley.
I haven't figured out a way to calculate how to do rookies :X
I don't get it..l lol. But I didn't read through it full :P
Quote from: Jukes on December 19, 2012, 10:14:20 PM
I reckon it's good to go with Beams and Cotchin over SwAblett. To find a quick way of valuation for players I like to add together three variables; previous season average (risk evaluation), 2013 predicted average (scoring evaluation) and the difference between these, improvement (cash value) then when comparing take into account the price difference by dividing the price difference by the magic number, 5150, and add that into the equation by adding it to the cheaper option.
I have predicted that Ablett will retain his average at approximately 125 PPG, while Swan will drop 5 points to 128, Cotchin gain 7 average to 118, and Beams 6 average to 122.
Ablett = 125 + 125 + 0 = 250
Swan = 133 + 128 + -5 = 258
Cotchin = 111 + 118 + 7 = 236
Beams = 116 + 122 + 6 = 244
That gives SwAblett a running total of 508, while CotchBeamsy a running total of 480.
Now you add price into the equation. SwAblett has a total of 1,330,600. CotchBeamsy has a total of 1,169,600. This means SwAblett costs $161,000 more than CotchBeamsy. 161000 / 5150 = 31.2621.
SwAblett = 508
CotchBeamsy = 480 + 31.2621
= 511.2621
Meaning CotchBeamsy shows a total of 3.2621 rating points over SwAblett. This may appear quite small seeing as their totals are both over 100, but seeing as each combination are scaled to be swung toward each other (through averages for SwAblett's benefit and pricing for CotchBeamsy's benefit) it is actually quite large.
Very nice formula, but the problem is in what one predicts. Keeping the same example I predict that Swan will keep his average of 133 (no reason for him not to have as good a year as the last).
Therefore
SwAblett = 518
CotchBeamsy = 480 + 31.2621
= 511.2621
and the CotchBeamsy advantage over SwAblett disappears.
*yawn* career best average take their TAC cup scores for for mature players, VFL/ SANFL/ NEAFL/ WAFL/ whatever tasmania is called then do everything else the same, but multiply the final result by about 0.75.
All about opinion there Presto buddy. I've been hearing bad things about Swannies personal life (potential drug habits, rifts with teammates, early retirement plans, etc) which have swung my average prediction downward.
Quote from: Jukes on December 20, 2012, 02:29:08 AM
All about opinion there Presto buddy. I've been hearing bad things about Swannies personal life (potential drug habits, rifts with teammates, early retirement plans, etc) which have swung my average prediction downward.
Well we've been hearing about this for a while (his weight was a major concern last year) but it hasn't affected him before. His average may drop a little though as a 133 average is so hard to maintain,
Quote from: henry on December 20, 2012, 11:07:48 AM
Quote from: Jukes on December 20, 2012, 02:29:08 AM
All about opinion there Presto buddy. I've been hearing bad things about Swannies personal life (potential drug habits, rifts with teammates, early retirement plans, etc) which have swung my average prediction downward.
Well we've been hearing about this for a while (his weight was a major concern last year) but it hasn't affected him before. His average may drop a little though as a 133 average is so hard to maintain,
my main worry is actually whether he will get pushed up to a forward pocket on a regular basis :-X
Now you just need to unlock the Lyon paradox.
Quote from: Mr.Craig on December 20, 2012, 11:57:04 AM
Now you just need to unlock the Lyon paradox.
hahaha quality Craigy ;D
Only problem with is Jukesy is that in the end the major factor of your work is the prediction of the players, which varies from person to person. :)
I like the mathematics you've put into this though, good effort.
Top work Jukes!
Quote from: Jukes on December 20, 2012, 12:49:27 AM
I have developed a formulae for basic differientiation between players/groups of players to determine which is a better option, titled by me as the "Jukes Coefficient". Below is a quote from it's first use, in the SwAblett thread from earlier today.
Quote from: Jukes on December 19, 2012, 10:14:20 PM
I reckon it's good to go with Beams and Cotchin over SwAblett. To find a quick way of valuation for players I like to add together three variables; previous season average (risk evaluation), 2013 predicted average (scoring evaluation) and the difference between these, improvement (cash value) then when comparing take into account the price difference by dividing the price difference by the magic number, 5150, and add that into the equation by adding it to the cheaper option.
I have predicted that Ablett will retain his average at approximately 125 PPG, while Swan will drop 5 points to 128, Cotchin gain 7 average to 118, and Beams 6 average to 122.
Ablett = 125 + 125 + 0 = 250
Swan = 133 + 128 + -5 = 258
Cotchin = 111 + 118 + 7 = 236
Beams = 116 + 122 + 6 = 244
That gives SwAblett a running total of 508, while CotchBeamsy a running total of 480.
Now you add price into the equation. SwAblett has a total of 1,330,600. CotchBeamsy has a total of 1,169,600. This means SwAblett costs $161,000 more than CotchBeamsy. 161000 / 5150 = 31.2621.
SwAblett = 508
CotchBeamsy = 480 + 31.2621
= 511.2621
Meaning CotchBeamsy shows a total of 3.2621 rating points over SwAblett. This may appear quite small seeing as their totals are both over 100, but seeing as each combination are scaled to be swung toward each other (through averages for SwAblett's benefit and pricing for CotchBeamsy's benefit) it is actually quite large.
The formula must be adjusted however for differentiating between mid-priced players.
Jukes Coefficient = Career best average + Predicted average for 2013 + xy
Then factor in price as done in the first formulae
'xy' is a relatively complex variable. To find a players xy, divide their price by the magic number in 5150. Then subtract this figure from their predicted average.
Below I have applied the Jukes Coefficient for Midpriced players for the battle between backline midpricers in Cameron Pedersen and Nathan Bock.
Pedersen: 71.3 + 75 + 17.5 = 163.8
Bock: 87.8 + 80 + 25.09 + 2.6 = 195.49
Bock wins comfortably by 32.41 points, meaning you should select Bock over Pedersen for your team.
Of course both of these formulae for premos and mid-pricers cannot be affected by other variables such as DPP and injury, although that could be factored in if you could be bothered finding an accurate proportion to the other variables.
Thoughts on these formula?
Only problem with the Pedersen vs Bock theory is Bock snapped his leg last year and Pedersen didnt, but i like the idea and effort put in Jukes
Yeah as I said in the OP there's obviously blind spots like that to be alert for, such as bye, early draw, etc.
Rookie and Rookie-Priced Formula; this will probably very unreliable but I'll give it my best shot.
JCE = (SA(M) - P - JS) x 2.5
= (Scoring Ability(Multiplier) - Price - Job Security) x 2.5
Scoring Ability: The average of a player in their state competition from the last season if a newly-drafted player OR the average of a player in their 2011 AFL season. You can find most of the stats from last season for newly-drafted players in Mr Craig's rookie thread.
Multiplier: 1 if player's 2012 season was in the seniors of their state competition. 1.2 if player's last non-injury affected season was in the AFL. 0.85 if player's 2012 season or the stats taken from scoring ability were in an U18 comp/carnival. 0.9 if stats taken from reserves competition.
Price: Simple. Divide the player's price by the magic number, 5150. Subtract this from the expression.
Job Security: This is where the opinion part of the formula takes place, like the predicted average component of the formula for premiums and mid-pricers. Determine how many players are ahead in the pecking order ahead of this player for their position at the club. Don't go into specifics eg tall, short, inside, outside etc, just look at the whole position. Multiply this by 2 and subtract from the expression.
Very confusing and probably doesn't work very well but w/e. This is all an experiment(al as anything).
All based on estimates... Nonetheless great work!
An example; Dean Terlich vs Sam Colquhoun
Terlich: 84.6(1) - 19.94 - 14
Colquhoun: 90.1(0.9) - 19.94 - 12
Terlich = 84.6 - 33.94
Colquhoun = 81.09 - 31.94
Terlich = 50.66 x 2.5
Colquhoun = 49.15 x 2.5
Players are very close, Terlich just winning out. IMO both offer great value and should both be highly considered as two of the best backline rookies for 2012.
Prof Juke can you give us your specs on Knights and varcoe.
Cheers
Quote from: greenmoon on December 21, 2012, 01:03:39 AM
Prof Juke can you give us your specs on Knights and varcoe.
Cheers
Knights: 87.7 + 80 + 30.6
Varcoe: 75.1 + 73 + 27.49 + 3.9
Knights = 198.3
Varcoe = 179.49
Knights wins comfortably, although this does not take into account the DPP of Varcoe, although I doubt that'd make up the 20 points needed.
Predicted averages taken from the following articles;
Chris Knights (Toga): http://www.fanfooty.com.au/forum/index.php/topic,67875.0.html
Travis Varcoe (ele): http://www.fanfooty.com.au/forum/index.php/topic,67883.0.html
Love your work Jukesy, could you do Knights vs Gray vs JJ Kennedy please? :)
Knights: 87.7 + 80 + 30.6 + 4.194
Gray: 81.8 + 80 + 26.389
JJK: 79.4 + 80 + 27.13 + 0.737
Knights = 202.49
Gray = 188.189
JJK = 187.267
Quote from: Jukes on December 21, 2012, 12:26:07 PM
Knights: 87.7 + 80 + 30.6 + 4.194
Gray: 81.8 + 80 + 26.389
JJK: 79.4 + 80 + 27.13 + 0.737
Knights = 202.49
Gray = 188.189
JJK = 187.267
Thanks mate :)
Awesome work Jukes!
Can't split Hanley and Broughton in my defence. Who wins using your formula? Cheers
Hanley: 82.5 + 87 + 4.5
Broughton: 79.2 + 88 + 8.8 + 3.36
Hanley = 174
Broughton = 179.36
Cotchin + Christenson v Martin + Mundy
Quote from: Holzman on December 21, 2012, 02:34:50 PM
Cotchin + Christenson v Martin + Mundy
Cotchin: 110.7 + 118 + 7.3
Christensen: 74 + 85 + 11
$951,400
Martin: 84.8 + 92 + 7.2
Mundy: 87.7 + 102 + 14.3
$888,900
CC = 406
MM = 388 + 12.136
Cotchin + Christensen = 406Martin + Mundy = 400.136
Quote from: whatlez on December 20, 2012, 01:37:50 AM
I don't get it..l lol. But I didn't read through it full :P
Are you going for an early nomination for an Elxam Award - Relton Roberts/Jimmy Bartel/Stephen Milne/Barry Hall combo award?
Hi Jukes your calculations please,thanks mate.
grimes/thomas/wines-----------broughton/varcoe/murphy
Cheers
Quote from: greenmoon on December 21, 2012, 03:31:17 PM
Hi Jukes your calculations please,thanks mate.
grimes/thomas/wines-----------broughton/varcoe/murphy
Cheers
Grimes: 92 + 96 + 4
Thomas: 93.2 + 103 + 9.8
Wines: (115.4(0.85) - 30.42 - 8) x 2.5.
$1,110,700
Broughton: 79.2 + 88 + 8.8
Varcoe: 75.1 + 73 + 27.49
Murphy: 101.3 + 110 + 8.7
$1,164,400
GTW: 192 + 196.2 + 149.175
BVM: 176 + 175.59 + 220
GTW: 537.375 + 10.427
BVM: 571.59
GTM = 547.802
BVM = 571.59This may be inaccurate as I had to factor in premiums, mid-pricers and rookies.
Cheers Jukes great info
Last one for me,could you rank these in order of priority thanks.
Robbo
thomas
grimes
goddard
broughton
Hey Jukes
Cox and Birchall or Ryder and Grimes?
Cheers :)
Quote from: greenmoon on December 21, 2012, 04:26:20 PM
Cheers Jukes great info
Last one for me,could you rank these in order of priority thanks.
Robbo
thomas
grimes
goddard
Robbo: 93 + 98 + 5 + 4.4
Thomas: 93.2 + 103 + 9.8 + 4.14
Grimes: 92 + 96 + 4 + 5.3
Goddard: 97.3 + 102 + 4.7
Robbo = 200.4
Thomas = 210.14
Grimes = 197.3
Goddard = 204
Meaning,
1st. Thomas
2nd. Goddard
3rd. Grimes
4th. Robbo
Quote from: KoopKicka on December 21, 2012, 04:33:01 PM
Hey Jukes
Cox and Birchall or Ryder and Grimes?
Cheers :)
Cox: 99.5 + 96 - 3.5
Birchall: 88.4 + 94 + 5.6
$967,800
Ryder: 90.3 + 92 + 1.7
Grimes: 92 + 96 + 4
$939,100
CB: 192 + 188
RG: 184 + 192 + 5.57
CB = 380
RG = 381.57
Quote from: SydneyRox on December 21, 2012, 03:10:32 PM
Quote from: whatlez on December 20, 2012, 01:37:50 AM
I don't get it..l lol. But I didn't read through it full :P
Are you going for an early nomination for an Elxam Award - Relton Roberts/Jimmy Bartel/Stephen Milne/Barry Hall combo award?
Already got the Relton, Jimmy and Stephin ones under my belt :P
Hey Jukes
Please don't take offence at the following, I appreciate that you are trying to provide some way to effectively players for selection.
Quote from: Jukes on December 20, 2012, 12:49:27 AM
I have developed a formulae for basic differientiation between players/groups of players to determine which is a better option, titled by me as the "Jukes Coefficient". Below is a quote from it's first use, in the SwAblett thread from earlier today.
Coefficient is a constant number, formulae is plural of formula.
Quote from: Jukes on December 20, 2012, 12:49:27 AM
Quote from: Jukes on December 19, 2012, 10:14:20 PM
I reckon it's good to go with Beams and Cotchin over SwAblett. To find a quick way of valuation for players I like to add together three variables; previous season average (risk evaluation), 2013 predicted average (scoring evaluation) and the difference between these, improvement (cash value) then when comparing take into account the price difference by dividing the price difference by the magic number, 5150, and add that into the equation by adding it to the cheaper option.
I have predicted that Ablett will retain his average at approximately 125 PPG, while Swan will drop 5 points to 128, Cotchin gain 7 average to 118, and Beams 6 average to 122.
Ablett = 125 + 125 + 0 = 250
Swan = 133 + 128 + -5 = 258
Cotchin = 111 + 118 + 7 = 236
Beams = 116 + 122 + 6 = 244
That gives SwAblett a running total of 508, while CotchBeamsy a running total of 480.
Now you add price into the equation. SwAblett has a total of 1,330,600. CotchBeamsy has a total of 1,169,600. This means SwAblett costs $161,000 more than CotchBeamsy. 161000 / 5150 = 31.2621.
SwAblett = 508
CotchBeamsy = 480 + 31.2621
= 511.2621
Meaning CotchBeamsy shows a total of 3.2621 rating points over SwAblett. This may appear quite small seeing as their totals are both over 100, but seeing as each combination are scaled to be swung toward each other (through averages for SwAblett's benefit and pricing for CotchBeamsy's benefit) it is actually quite large.
I can understand what you are trying to do in including the average of the previous year. However, if you are going to add the difference between the previous years average and 2013 predicted average, you are essentially doubliing your 2013 predicted average.
so if you are assessing them with the formula JC = 2012 average + 2013 predicted average + difference
now consider: difference = 2013 predicted average - 2012 average
so, now JC = 2012 average + 2013 predicted average + (2013 predicted average - 2012 average)
= 2012 average + 2013 predicted average + 2013 predicted average - 2012 average
= 2013 predicted average + 2013 predicted average + 2012 average - 2012 average
= 2*2013 predicted average
e.g. for Swan,
JC = 133 (2012 average) + 128 (2013 predicted average) - 5 (difference)
= (133-5) + 128
= 128 + 128 (2*128)
Buggered if I know how to actually analyse them though.
Hahaha, nice work Jukesy.
Wouldn't fully listen to it but an interesting point of view.
@JackBeQuick
Pretty sure that first "formulae" was just a typo :P
By including predicted average, 2012 average and difference instead of just predicted x2 it allows you to greater compare players on multiple fronts by just reading off the formula.
2012 average lets you compare players on their history, ie scoring security. You're more likely to pick a player with a history of good scoring than a player with no past good scoring, right.
Predicted average allows comparation between players on their scoring potential for 2012.
Difference lets you see their room for improvement, ie money to gain.
Bringing price into it gives you value options.
Each of these are now together in one place, in addition to the JC result.
Quote from: Jukes on December 22, 2012, 03:24:30 PM
By including predicted average, 2012 average and difference instead of just predicted x2 it allows you to greater compare players on multiple fronts by just reading off the formula.
2012 average lets you compare players on their history, ie scoring security. You're more likely to pick a player with a history of good scoring than a player with no past good scoring, right.
Predicted average allows comparation between players on their scoring potential for 2012.
Difference lets you see their room for improvement, ie money to gain.
Bringing price into it gives you value options.
Each of these are now together in one place, in addition to the JC result.
I just think it is a little misleading.
1. This formula is for premiums, by definition they are not a player with "no past good scoring".
2. There are 2 values dependent on subjective assessment.
3. Players will need to improve their average to maintain their price as season progresses. (i.e. difficult to truly assess money to gain for predicted improvement. Additionally, as a premium you are not picking them to make money)
I'm not quite sure what you mean by:
Quote from: Jukes on December 22, 2012, 03:24:30 PM
Bringing price into it gives you value options.
Do you mean that in determining the difference in price and dividing by the magic number, you can determine the scoring differential based on initial pricing.
i.e. if assessed by prices, differential between Swan/Ablett and Cotchin/Beams is 31.26
but assessing by predicted averages of Swan/Ablett 253 (128/125) and Cotchin/Beams 240 (118/122), difference is 13
That is, though you are paying for 31.26 more points in picking Swan/Ablett, but effectively (assuming predicted averages are correct) you will only score 13 more points.
*17
captain bonus ;)
Happy new year Jukesy,
Your Analysis thanks.
boyd,knights,varcoe
or
l.mitchell,cox,thomas
Cheers
I have one please Jukes.
J.Grimes + B.Moloney
vs
J.Redden + C.Pederson
Cheers!