(Written & Created By The Salamander)
In last season’s AFLW Fantasy competition, I introduced a new scoring system (called SalamanderCoach) that was was designed to roughly approximate the SuperCoach scoring system, by rewarding the kinds of actions that contribute most to winning (or losing) a game of football: winning the ball on the inside, ball movement, pressure/defence, and goalkicking.
In all, it served its purpose quite well. However, in my preparations for this year’s competition, I have stumbled across a new, much easier to use source of data for AFLW games. Although it doesn’t have everything I want, this new data source has some more exotic stats to which I did not have access last season. These include metres gained, tackles inside 50, intercepts, and other things. Naturally, I decided to work some of these new stats into this year’s scoring system.
Old scoring system (SalamanderCoach 1.0):
● Ineffective Disposal: 0 points
● Effective Disposal: 3 points
● Contested Possession: 3
● Clanger: -3 points
● Clearance: 5 points
● Mark: 3 points
○ additional: 5 points if contested
○ additional: 5 points if inside 50
● Tackle: 4 points
● One-percenter: 4 points
● Inside-50: 3 points
● Hitout: 2 points
● Free-for: 1 points
● Goal: 12 points
● Behind: 1 point
New scoring system (SalamanderCoach 2.0):
● Effective metres gained† : 1 point per 5 metres (rounded to nearest point)
● Contested possession : 3 points
● Clearance : 5 points
● Inside-50: 3 points
● Hitout: 2 points
● Mark : 2 points
○ Additional 6 points if contested
○ Additional 6 points if inside-50
● Tackle : 4 points
○ Additional 4 points if inside 50
● One-percenter : 4 points
● Intercept : 4 points
● Goal: 12 points
● Behind: 1 point
● Non-scoring shot: -1 point
● Goal assist: 6 points
● Score involvement: 3 points
● Turnover: -1 point
● Clanger: -3 points
† Effective metres gained is one of the best predictors of the outcome of an AFL game. Unfortunately, Champion Data doesn’t give us access to this particular stat, but we can get a usable approximation by multiplying metres gained by disposal efficiency. This formula doesn’t behave the way we want when metres gained is negative, so I have chosen to give this stat a minimum value of 0. It’s not perfect, but I think it is preferable to the alternatives – rewarding gains in metreage that go straight the the opposition (and come straight back), or rewarding useless but nominally ‘effective’ disposals. N.B. because of the way I am calculating this stat, if you wish to verify my work and add up the scores yourself, you will have to do so on a game-by-game basis – multiplying the season averages by their respective values won’t work.
Check out the upcoming stats guide to see how these changes affect the scoring of each player.
Questions and Answers:
What happened to frees-for?
The short answer is that I simply forgot about it until I was writing this article (at which point I remembered that it featured in the old system), by which time I didn’t want to go back and change all the numbers. However, although the immediate cause might be one of mere forgetfulness, I actually think its removal is an improvement. Outside of a holding-the-ball decision (in which case the player is already rewarded with 4 points for a tackle), a free kick is not something that is earned – it either comes about by an opposition player breaking the rules, which has nothing to do with the player receiving the free kick, or by the receiving player doing something (such as ducking) to draw a free kick, which is not something that I personally think ought to be rewarded. So a free kick, in and of itself, is not worth anything; it’s what the player does with it that counts.
Aren’t a turnover and a clanger the same thing?
No. There is some overlap between the two, but not all turnovers are clangers, and not all clangers are turnovers. A turnover is simply when a player gives the ball back to the opposition. A clanger can be a particularly bad turnover, a free-kick/50-metre-penalty against, or a general act of incompetence or stupidity, such as fumbling what should be an easy mark.
Kicking the ball to the opposition is undesirable, but it happens. Even the best players do it sometimes. It’s generally a nuisance, but not the end of the world. Hence the -1. A clanger, on the other hand, be it in the form a bad turnover or otherwise, is very bad for the team, hence the much harsher -3 points.
Why are marks worth less this year?
Taking a mark – catching the ball – is a basic skill of the game. Doing so from a teammate’s kick, under no physical pressure, and not within scoring range is not much of an achievement in my book. Ideally, I would make a mark worth nothing in and and of itself, with points only handed out for specific subcategories. However, I am reluctant to do so in the absence of available stats for intercept marks and marks taken on a lead. Thus, the present compromise.
And what’s so special about a mark inside 50, if it’s taken uncontested? A basic skill is a basic skill, regardless of which part of the ground it’s in, right?
If a player can get themselves into a position where they are able to take an uncontested mark that close to goal, they deserve to be rewarded (or their direct opponent should be docked a dozen or so points. As it stands, we don’t have the data do be able to make that distinction).
There’s a stat for non-scoring shots? Also, why are you docking points for it?
It turns out that yes, Champion Data does in fact record all shots at goal – even those that don’t score anything. I’m docking a point for it because although a behind adds to the team’s overall score, a completely missed shot is just a wasted opportunity. It’s quite possible for the player to more than recover that point, however – they’ll get points for an inside-50 if the shot is taken from outside 50 (or if they run the ball inside 50 then have a shot), and can potentially be rewarded with a score involvement and/or goal-assist if it ends up in the hands of a teammate.
Let me know below if you have any questions or comments.
Leave a comment / Scroll to bottom
I reckon that’s excellent. We should run this scoring through the 2019 AFL season too and see what it reveals. Bound to reward players by degree of influence on the game, and that for me is what SC distinctly lacks, to the point of frustration.
Great effort Jack! And not just for the scoring system, but the whole comp. Must have taken AGES!!
I’d rather not think about how many hours have gone into all of this… 😉
The scoring is fully automated, so I’m quite happy to keep tabs on how the men go in this system. How many people on here would care is another matter.
We have better stats for the men than we do for the women (including the Intercept Mark and Mark on Lead stats I so desperately wanted, plus things like Hitouts-to-Advantage), so it would be tempting to do a slightly different version for the men, but then it would be harder to compare their scores with the women. If comparability is desirable, then it might also be worth applying some (uniform) scaling so that we end up dealing with similar sized numbers. I had been toying with the idea of doing this for this year’s AFLW season anyway, but decided it would just be creating unnecessary extra work at a time when I already had a lot on my plate. Scaling the scores would mean that it wouldn’t matter if it was a high/low-scoring game, of if was wet/dry weather, etc., and unlike Champion Data, mine would be uniform and 100 % transparent. There’s a good chance I’ll implement it next year.
This might just be vanity on my part, but at the back of my mind, I do like the idea that this could one day become a viable third scoring system. It certainly combines my favourite features of both Dreamteam and SuperCoach – like DT, it’s fully transparent, so anyone can calculate the scores, but, like SC, it focuses on much more pertinent stats. Then again, I may be a tad biased. 😉
That’s brilliant mate. I don’t even wanna think how much time went into it 😉
May I ask what your source is for all these stats?!
The only thing I reckon is missing from all the stats we can get is that perceived pressure that often leads to a turnover, without necessarily impacting play directly (am thinking the Kane Lambert types) that go largely unrewarded. Still, they end up winning B&Fs and I guess that’s way more important to the players!
Once again, fantastic effort. Congratulations!
Hey, if there was a stat for that kind of thing, it would definitely be included! I’m pretty sure Champion Data counts shepherding as a one-percenter, but last time I checked, they didn’t include corralling. You’ll have to ask them why.
As I said in the comments section of the stats guide, I am more than willing to send you the raw data if you’re interested.
As to where I got it, the short answer is Champion Data. The long answer is by capturing the data Champion Data sends to the AFL website’s stats section. When you use the AFL’s stats site, it basically works as follows:
#1. You go to the stats section of the AFL website, and ask it to load some particular data.
#2. The website then uses Champion Data’s API to ask for the data.
#3. Champion Data’s servers look at the request, and see that it was sent by the AFL website (unfortunately it doesn’t work if you just send a request directly – I’ve tried).
#4 Champion Data’s servers send the relevant data in JSON format to your browser.
#5. Your browser uses that data to populate the AFL website’s tables.
So far, nothing unusual there. That’s how any AJAX site works. The trick comes with #4: before you’ve actually made the request, open up your browser’s developer tools, and capture then network traffic. You’ll be able to see Champion Data’s JSON package directly, and copy/paste its contents to a file on your computer. Congratulations, you now have raw AFL stats, directly from Champion Data! If you look inside, you’ll notice that there’s a lot of data that the website’s table doesn’t even give you the option of loading.
Before you ask, yes, this is 100 % legal. Champion Data was always sending this data to the browser anyway. This way, we’re just making better use of it. It may not be the way they intended us to use it, but that’s not our problem. 😉