Over the last several years, many AFL followers have argued that the Herald-Sun/Champion Data “supercoach” fantasy football game is far superior to the preceding “Dreamteam” game.
But only recently, I’ve discovered a fatal flaw in the “Supercoach” scoring that hopefully will have coaches screaming for answers.
We all know that with “Dreamteam”, we all know what we score. It’s a fixed scoring system for kicks, handballs, marks, etc. It’s the basics of football.
But, looking into the “Supercoach” scoring system, straight from the website, it says “points are awarded based on the comprehensive Champion Data scoring system, some of which are highlighted”. The operative word in that quote is “some”. We get points for effective & ineffective possessions, etc, which has always been a contentious issue & up for interpretation.
Now, it seems “interpretation” is the rule of thumb with Champion Data & they’ve been caught out. Let’s have a look at why they’ve been busted.
Let’s take a combined score of the 2009 round 1 games & see if you can spot the similarities in the totals.
- Richmond .vs. Carlton 3300
- Hawthorn .vs. Geelong 3302
- Collingwood .vs. Adelaide 3299
- Brisbane .vs. West Coast 3306
- St Kilda .vs. Sydney 3300
- Melbourne .vs. Nth Melbourne 3301
- Port Adelaide .vs. Essendon 3300
- Fremantle .vs. Western Bulldogs 3299
Average total 3301
Each teams combined total must have a standard distribution of 3300. Go ahead, check every game of every round & you’ll find the same result give or take. The question is how does this always end up being the result? It’s quite obvious some form of bell curve has to be used, that I do know. It raises another question that several of us have been asking for a few years & the answer now is, yes, the SuperCoach does also score on bias based on “superstar” status & where you play. How else does Chris Judd get so many high SuperCoach scores in games we knew he played poorly.
For those who are wondering, the Saints have 5 in the top 17, the Cats have the same number in the top 100, the same number as the Kangroos.
Collingwood have 8 & Carlton 7 in the top 100.
In other words, the high salaried, expected performances are not matching the actual performances of the 2009 premiers.
Take a look at the team totals for round 1 & the margin to see the examples
- Carlton by 83 pts (1880 to 1420)
- Western Bulldogs by 63 pts (1871-1428)
- Port Adelaide by 41 points (1756-1544)
- Nth Melbourne by 34 points (1721-1580)
- St Kilda by 15 points (1836-1464)
- Brisbane by 9 points (1920-1386)
- Geelong by 8 points (1842-1460)
- Adelaide by 4 points (1673-1626)
There is little doubt, as a result, the salaries don’t match the scores. So, how do we beat the system? Firstly, ignore the averages for they set the prices. Secondly, create your own bell curve based on the average score totals by this formula
- A = Players score
- B= average of league score
- C= number of players
- A – B
- (A – B) squared
- Average of ((A – B) squared) / C))
- (((A – B) – average of ((A – B) squared) / C)))
After completing this, we can get the form guide of each player, not counting the 2010 rookies, in a form of standard devitation.
- 9 players are the elite, or 3 times above the average
- 81 players are champions, or 2 times above the average
- 150 players are on the cusp of something big, or 1 times above the average
- 146 players are great contributors, or 1 times below the average
- 100 players are the list cloggers, or 2 times below the average
We want to pick from those 150 who are on the verge of something either big or returning to past glories, those who are above the average & yet have the breakout game like the elite without the price tag. Therefore, using their true score shows us some true bargains compared to the average score.
If you don’t think there’s a difference, here’s the sample of the salaries.
- 1 player is 4 times above the average
- 15 players are 3 times above the average
- 72 players are 2 times above the average
- 140 players are 1 times above the average
- 185 players are 1 times below the average
- 66 players are 2 times below the average
- 7 players are 3 times below the average
I’ve entered a team with $142,100 left remaining in the salary cap using 19 household names ($411,600 & over) & 11 cheap players.
Loading ...
March 5th, 2010 at 9:59 am
This argument has been made before and it is scientifically thorough. I quickly did a sum of two NAB cup games and they were 2902 and 2899 which is quite close but pretty far away from the 3301 mark. I think the only way to test that it’s a scam is to count all of the players stats yourself and mark them using their criteria (you can do this easily if you record the game). You will get different results because of interpretation but a real breakthrough is if they are significantly different. I think the reason that the superstars get more points is because the scorers are biased. It’s pretty hard to get around that.
Luke W´s last blog ..Western Bulldogs vs. Port Adelaide – NAB Cup Semi-Final Match Preview
[Reply to Comment]
March 5th, 2010 at 10:22 am
Hi, Luke. Thanks for reading.
It’s quite obvious that the SuperCoach uses a weighted system based on game importance for how else can you get 26 players a side equalling around 2900.
My bet is that it’s based on the FIFA soccer system (4 quarters going up by a higher % each quarter), but we just need to work out the weighting.
craig eyles´s last blog ..Week 18 NBA X-Win/Loss Totals
[Reply to Comment]
March 5th, 2010 at 12:39 pm
I got a comment from a reader on Twitter saying that the SuperCoach algorithm for 2010 is different to 2009
[Reply to Comment]
March 5th, 2010 at 1:39 pm
Jermayn
I’d like to see if the 2010 algorithm is different. Can you get the info??
craig eyles´s last blog ..Week 18 NBA X-Win/Loss Totals
[Reply to Comment]
March 12th, 2010 at 9:06 am
The supercoach site has the breakdown in stats and point scoring for all to see this year:
Stat Description Points : Awarded/Deducted
Effective kick : 4 Points
Ineffective kick : 0 Points
Clanger kick : -8 Points
Handball effective : 2 Points
Ineffective handball : 0 Points
Handball clanger : -6 Points
Handball receive : 1 Point
Hardball get : 4 Points
Loose-ball get : Points
Goal : 8 Points
Behind : 1 Point
Mark uncontested (maintaining possession) : 1 Point
Mark contested (maintaining possession) : 6 Points
Mark uncontested (from opposition) : 4 Points
Mark contested (from opposition) : 8 Points
Tackles : 4 Points
Free kick for : 4 Points
Free kick against : -6 Points
Hitout to Advantage : 3 Points
Danny´s last blog ..uncyherb: @fulltimecasual Congrats Uncle Pete.
[Reply to Comment]
March 12th, 2010 at 10:21 am
If that’s the case, Danny, then why isn’t the points breakdown set in concrete like Dreamteam?? It isn’t possible using the points system alone that each games points total is equal to 3300 give or take 1 or 2 points. That’s the main point of my post.
[Reply to Comment]
March 12th, 2010 at 10:58 am
@craig
Maybe it was different in 09 (I didn’t play supercoach until this year), but the point scoring “rules” suggest it IS set in concrete for 2010.
Essendon V WCE NAB Rd1 – 1396 to 1494 = 2890
Collingwood V StKilda NAB Rd1 – 1430 to 1472 – 2902
Sydney V Carlton NAB Rd1 – 1481 to 1408 = 2889
Bulldogs V Hawks NAB Rd2 – 1616 to 1273 = 2889
The question, with these being similar numbers, is “is this correct?”. Given that the game is fixed in time length, and that to change the overall score significantly would take many more effective stats it might be that what we’re seeing is a reflection of the fact that effective stats take a measurable amount of time, and that with time being the same between all game we’re seeing time as the limiting factor in scores.
[Reply to Comment]
March 12th, 2010 at 11:02 am
Danny
Would you agree that the 2889 points for the Syd .v. Carl & WB.v.Haw game show there is something wrong with the Swans/Blues game played at a terrible standard??
craig eyles´s last blog ..Week 4 Super 14s Picks
[Reply to Comment]
March 14th, 2010 at 11:06 am
It’s an interesting Q Craig. I didn’t watch the Syd/Car match, so I can’t really comment. I guess the question for those that did watch is “were Carlton (1408 pts) better than Hawthorn were (1273 pts)”?
As I said, I can’t comment because I didn’t see both games.
Danny´s last blog ..uncyherb: @Taezar I meant both, not just the SB. But the SB probably is better.
[Reply to Comment]
March 14th, 2010 at 1:13 pm
I only watched parts of the Hawthorn/ Dogs game but I would say they still looked and played better than Carlton. However from what I remember the Hawks where pressured a lot more than what Carlton where against Sydney. and in looking at the scores that Danny supplied you don’t score for ineffective handballs and kicks and from what I remember the Hawks where pressured a fair bit.
What gets me is the fact that all the score are around the same.
[Reply to Comment]
March 15th, 2010 at 11:08 am
Score may be lower for nab cup because time of game may be taken into account
[Reply to Comment]
March 27th, 2010 at 9:38 am
I have been reading from lots of sources as this year is my first year at Supercoach. From what I have been able to find.
The scores are indeed lower for NAB cup due to the amount of game time.
Ess vs Geel was 3327 if i’m correct.
Also I read that the weight doesn’t increase as the game gets longer more along the lines of if a team gets out to a 10 goal margin early and then cruises most of the points will be weighted to the start. And vice versa if a team wins in the last moments of the game massive points there!
So a juddy or in my case selwood last night that wasn’t dominating for the game, does some classy things in the tight sections of the game, an effective pass, goal etc.. it quickly adds up and he ends up with more points than ablett. Even though ablett had way more touches.
I think I actually like the idea behind supercoach even if the formula is a mystery. If a team is up by 40 points and plays a bit of pingpong in the back half, why should that be equal points to someone who gets the ball and hits their forward on the chest with 15 seconds to go when the scores are tied!
[Reply to Comment]
March 27th, 2010 at 9:42 am
The other bit to add was that the supercoach stats is to judge what effect a player had on the game. Thus why the points should match up between each game. One thing I did noticed was some games are a few points higher than others. Would be interesting to see if they had longer quarters or something…
[Reply to Comment]
March 29th, 2010 at 4:33 pm
The scores in Nab cup are 25% (or there abouts) lower than regular season because the games are 25% shorter. The process is that they score each player as per the points described above, give a little extra weighting to those that occur at important points in the game (i.e first quarter if winning team blows away losers or last quarter if it is a nail biter) and then they multiply the scores by 3300/total score. this brings the total of each teams score to 3300… hope that clears up a few things
[Reply to Comment]
March 29th, 2010 at 6:22 pm
@Damian The question remains how much weighting & at what point?? Some have pointed to me instances where scores are multiplied five fold to make up the 3300 required going on quarter by quarter.
craig eyles´s last blog ..Week 7 Super 14s Picks
[Reply to Comment]