First thought, and I may have mentioned this once or twice through the season already - but how is Roko not getting picked?Somebody has developed an algorithim to measure player performance and put up a chart of the best 30 players in each position in the AP over the whole season based on it. Pretty interesting stuff, even without fully understanding how performance is being measured.
http://www.rugby.net/demo-responsive.html
One for the stat lovers.
Moderator: Puja
- Which Tyler
- Posts: 9038
- Joined: Tue Feb 09, 2016 8:43 pm
- Location: Tewkesbury
- Contact:
One for the stat lovers.
Stolen from peat, elsewhere (lazy bugger, not putting it here as well!)
-
- Posts: 20225
- Joined: Tue Feb 09, 2016 7:52 pm
Re: One for the stat lovers.
Dan Cole notable by absence! Be quite interesting to know how it's been developed tho.Which Tyler wrote:Stolen from peat, elsewhere (lazy bugger, not putting it here as well!)First thought, and I may have mentioned this once or twice through the season already - but how is Roko not getting picked?Somebody has developed an algorithim to measure player performance and put up a chart of the best 30 players in each position in the AP over the whole season based on it. Pretty interesting stuff, even without fully understanding how performance is being measured.
http://www.rugby.net/demo-responsive.html
-
- Posts: 1535
- Joined: Fri Feb 12, 2016 11:31 am
Re: One for the stat lovers.
Indeed. Charles Piutau 14th best FB? If he's 14th then there's some seriously odd parameters being used.
- Stom
- Posts: 5743
- Joined: Wed Feb 10, 2016 10:57 am
Re: One for the stat lovers.
Indeed. Only Roko of the top 3 wings has notable defence. Suggesting it hasn't been taken into consideration.Beasties wrote:Indeed. Charles Piutau 14th best FB? If he's 14th then there's some seriously odd parameters being used.
There are other iffy positions there. Botica as the 6th best FH...it suggests negative stats haven't been taken into consideration for me: missed tackles, knock ons, kicks on the full, etc.
-
- Posts: 6341
- Joined: Wed Feb 10, 2016 3:42 pm
Re: One for the stat lovers.
Quite interesting but some explanation on the methodology behind the scoring would be handy.
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
There are quite a few laughable results in there.
- Stom
- Posts: 5743
- Joined: Wed Feb 10, 2016 10:57 am
Re: One for the stat lovers.
Found some:
http://www.rugby.net/s/lmpAyv5MR4Q#Cont ... 8064758784
Individual Stats
To "roll up" the stats into a single number we first convert them to the same units. We use the tackle as the base unit and convert everything else into this unit. We chose the tackle because it is the most frequent and universal individual statistic. Here are the base conversion factors for the stats that are attributable to the players:
Tackles Made: 1
Tackles Missed: -2
Tries: 5
Try Assists: 5
Points: .5
Runs: .2
Passes: .2
Kicks: .2
Metres Run: .1
Clean Breaks: 3
Defenders Beaten: 1.5
Offloads: 3
Turnovers: -2
Lineouts won on Throw: .5
Lineouts stolen: 2
Penalties Conceded: -3
Yellow Cards: -5
Red Cards: -9
Minutes: .1
Example: A player who has 5 tackles (5), 3 runs (.6) for 7 metres (.7) and plays 80 minutes (8) will have a raw score of 14.3.
http://www.rugby.net/s/lmpAyv5MR4Q#Cont ... 8064758784
Individual Stats
To "roll up" the stats into a single number we first convert them to the same units. We use the tackle as the base unit and convert everything else into this unit. We chose the tackle because it is the most frequent and universal individual statistic. Here are the base conversion factors for the stats that are attributable to the players:
Tackles Made: 1
Tackles Missed: -2
Tries: 5
Try Assists: 5
Points: .5
Runs: .2
Passes: .2
Kicks: .2
Metres Run: .1
Clean Breaks: 3
Defenders Beaten: 1.5
Offloads: 3
Turnovers: -2
Lineouts won on Throw: .5
Lineouts stolen: 2
Penalties Conceded: -3
Yellow Cards: -5
Red Cards: -9
Minutes: .1
Example: A player who has 5 tackles (5), 3 runs (.6) for 7 metres (.7) and plays 80 minutes (8) will have a raw score of 14.3.
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
Not exactly an algorithm, then. More random and very basic allocated weighting and basic adding and subtraction.Stom wrote:Found some:
http://www.rugby.net/s/lmpAyv5MR4Q#Cont ... 8064758784
Individual Stats
To "roll up" the stats into a single number we first convert them to the same units. We use the tackle as the base unit and convert everything else into this unit. We chose the tackle because it is the most frequent and universal individual statistic. Here are the base conversion factors for the stats that are attributable to the players:
Tackles Made: 1
Tackles Missed: -2
Tries: 5
Try Assists: 5
Points: .5
Runs: .2
Passes: .2
Kicks: .2
Metres Run: .1
Clean Breaks: 3
Defenders Beaten: 1.5
Offloads: 3
Turnovers: -2
Lineouts won on Throw: .5
Lineouts stolen: 2
Penalties Conceded: -3
Yellow Cards: -5
Red Cards: -9
Minutes: .1
Example: A player who has 5 tackles (5), 3 runs (.6) for 7 metres (.7) and plays 80 minutes (8) will have a raw score of 14.3.
-
- Posts: 20225
- Joined: Tue Feb 09, 2016 7:52 pm
Re: One for the stat lovers.
did you follow the link?Mellsblue wrote:Not exactly an algorithm, then. More random and very basic allocated weighting and basic adding and subtraction.Stom wrote:Found some:
http://www.rugby.net/s/lmpAyv5MR4Q#Cont ... 8064758784
Individual Stats
To "roll up" the stats into a single number we first convert them to the same units. We use the tackle as the base unit and convert everything else into this unit. We chose the tackle because it is the most frequent and universal individual statistic. Here are the base conversion factors for the stats that are attributable to the players:
Tackles Made: 1
Tackles Missed: -2
Tries: 5
Try Assists: 5
Points: .5
Runs: .2
Passes: .2
Kicks: .2
Metres Run: .1
Clean Breaks: 3
Defenders Beaten: 1.5
Offloads: 3
Turnovers: -2
Lineouts won on Throw: .5
Lineouts stolen: 2
Penalties Conceded: -3
Yellow Cards: -5
Red Cards: -9
Minutes: .1
Example: A player who has 5 tackles (5), 3 runs (.6) for 7 metres (.7) and plays 80 minutes (8) will have a raw score of 14.3.
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
Nope, I'm at work and I didn't have time. Based on your post I now have followed the link and I can confirm that I don't have the time or the inclination to read it. Based solely on the length of their explanation of how the rankings work I withdraw my previous post.Banquo wrote:did you follow the link?Mellsblue wrote:Not exactly an algorithm, then. More random and very basic allocated weighting and basic adding and subtraction.Stom wrote:Found some:
http://www.rugby.net/s/lmpAyv5MR4Q#Cont ... 8064758784
Individual Stats
To "roll up" the stats into a single number we first convert them to the same units. We use the tackle as the base unit and convert everything else into this unit. We chose the tackle because it is the most frequent and universal individual statistic. Here are the base conversion factors for the stats that are attributable to the players:
Tackles Made: 1
Tackles Missed: -2
Tries: 5
Try Assists: 5
Points: .5
Runs: .2
Passes: .2
Kicks: .2
Metres Run: .1
Clean Breaks: 3
Defenders Beaten: 1.5
Offloads: 3
Turnovers: -2
Lineouts won on Throw: .5
Lineouts stolen: 2
Penalties Conceded: -3
Yellow Cards: -5
Red Cards: -9
Minutes: .1
Example: A player who has 5 tackles (5), 3 runs (.6) for 7 metres (.7) and plays 80 minutes (8) will have a raw score of 14.3.
-
- Posts: 2072
- Joined: Fri Feb 12, 2016 12:01 pm
Re: One for the stat lovers.
If it's over the whole season, I imagine Piutau is 14th because he only actually played half of it!
It's interesting if you wanted to get a "Moneyball" style squad what those stats show.
It's interesting if you wanted to get a "Moneyball" style squad what those stats show.
-
- Posts: 11963
- Joined: Sat Feb 13, 2016 5:10 pm
Re: One for the stat lovers.
I like the idea, but very odd weighting system as everyone else has said. You can contribute only a penalty and get yellow carded but if you play 80 minutes you've made up for it.
Wasn't there a similar, but better, thing done for the Lions last time. Where there was a weighting system for the effectiveness of each contribution. It's an interesting idea to see how much the stat-topping players actually add each time they get involved.
Wasn't there a similar, but better, thing done for the Lions last time. Where there was a weighting system for the effectiveness of each contribution. It's an interesting idea to see how much the stat-topping players actually add each time they get involved.
- Which Tyler
- Posts: 9038
- Joined: Tue Feb 09, 2016 8:43 pm
- Location: Tewkesbury
- Contact:
Re: One for the stat lovers.
Largely down to a player getting nill points if he's injured; playing for someone else (eg. England) or playing in a different position (eg. Piutau).Mellsblue wrote:There are quite a few laughable results in there.
Whilst the maths behind the whole thing could certainly do with improving - I welcome that someone has had the time, inclination and technical skills to put something together. It may not be how I'd do it; but I have none of those 3 things in sufficient quantity. Besides, any set of maths is going to come down to subjective opinion (how many points per meter run? how many against for giving away a penalty? was that yellow card for a team offense or a red mist?)
Just noticed that some players (well, I noticed Devoto) appear in 2 sections; 28th "best" FH in the league AND 22nd "best" centre their.
Add his FH and his centre scores together would make him 17th "best" centre. It also shows the benefit (for this purpose) of playing in a solitary position (like FH) rather than a doubled-up position (like centre); as he had 208 points for 28th at FH, and 935 points for 22nd CE.
Wonder how many other players appear in multiple sections?
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
Well, if the system tells us that Harry Sloan is a far better centre than Jonathan Joseph then it's not a system worth looking at. That's before you take in to account they have Mark Atkinson as a fullback when he's actually a centre (I must be missing something as this is a ridiculous mistake).Which Tyler wrote:Largely down to a player getting nill points if he's injured; playing for someone else (eg. England) or playing in a different position (eg. Piutau).Mellsblue wrote:There are quite a few laughable results in there.
Whilst the maths behind the whole thing could certainly do with improving - I welcome that someone has had the time, inclination and technical skills to put something together. It may not be how I'd do it; but I have none of those 3 things in sufficient quantity. Besides, any set of maths is going to come down to subjective opinion (how many points per meter run? how many against for giving away a penalty? was that yellow card for a team offense or a red mist?)
It's a system that makes a player like Tom Wood look great and Conrad Smith average. There's no nuance to it - no points for running the defence well, no points for a perfectly timed dummy run, no points for a perfectly placed tactical kick and no points lost for the opposite of all of those. It's pretty pointless beyond a rudimentary grasp of a player's influence on basic elements of the game, as the rankings of quite a large number of players testifies to.
As I say, I haven't really got the time to look deeply at it but the large amount of anomalies immediately make me think that when I do have the time I still won't be looking deeply in to it.
- Stom
- Posts: 5743
- Joined: Wed Feb 10, 2016 10:57 am
Re: One for the stat lovers.
It's like "missed tackles". It only counts attempted tackles that have been missed. But what if a player is woefully out of position? They're not punished in this system.Mellsblue wrote:Well, if the system tells us that Harry Sloan is a far better centre than Jonathan Joseph then it's not a system worth looking at. That's before you take in to account they have Mark Atkinson as a fullback when he's actually a centre (I must be missing something as this is a ridiculous mistake).Which Tyler wrote:Largely down to a player getting nill points if he's injured; playing for someone else (eg. England) or playing in a different position (eg. Piutau).Mellsblue wrote:There are quite a few laughable results in there.
Whilst the maths behind the whole thing could certainly do with improving - I welcome that someone has had the time, inclination and technical skills to put something together. It may not be how I'd do it; but I have none of those 3 things in sufficient quantity. Besides, any set of maths is going to come down to subjective opinion (how many points per meter run? how many against for giving away a penalty? was that yellow card for a team offense or a red mist?)
It's a system that makes a player like Tom Wood look great and Conrad Smith average. There's no nuance to it - no points for running the defence well, no points for a perfectly timed dummy run, no points for a perfectly placed tactical kick and no points lost for the opposite of all of those. It's pretty pointless beyond a rudimentary grasp of a player's influence on basic elements of the game, as the rankings of quite a large number of players testifies to.
As I say, I haven't really got the time to look deeply at it but the large amount of anomalies immediately make me think that when I do have the time I still won't be looking deeply in to it.
I think it's something that works a bit better on tight 5 forwards than on any other position...
-
- Posts: 1038
- Joined: Tue Feb 09, 2016 7:09 pm
Re: One for the stat lovers.
1) It's been a busy day, I was going to post it up here, honest... (and no, it's not mine).
2) I see a lot of people have spotted the surprise results, but no one's noticed the method by which they come about.
Each player's positional score is done by adding up all the scores for their games in that position together.
International duty? Tough. Injured? Tough. Playing in a different position? You guessed it...
So Charles Piutau is only the 14th best full-back because he missed the first three games of the season and is also the 21st best centre. Personally I think there should be a page where you can view players contributions regardless of position, but hey, I didn't do it.
Meanwhile, Mark Atkinson creeps in at full-back because he had one game there and not a lot of people play full-back regularly; he doesn't get in at centre because there's a lot of people playing centre. There's even more playing prop, so Cole and Marler both don't make it because of their international duties.
3) This doesn't tell you that Harry Sloan is a better centre than Jonathan Joseph. It tells you that he was a better centre to have in the AP this season, which is probably true.
4) There's some flaws to this to say the least (we've all been over passive tackles made vs dominant tackles missed and so on) but if you treat this for it is - a rating of how well players have done solely in the AP this season - I think it's interesting and useful. Do you compare Sloan with Joseph? Probably not, but it does tell you that Bath may as have not had Joseph on the books this season, and that Sloan lags a fair bit behind other young England hopefuls such as James, Slade, Hill and so on. Since you can track when they did play and didn't play, you can also see that Robshaw and Beaumont were having big domestic seasons until England and injury hit.
And so on.
Also, Which, you c&p the post I used and then call me lazy?
2) I see a lot of people have spotted the surprise results, but no one's noticed the method by which they come about.
Each player's positional score is done by adding up all the scores for their games in that position together.
International duty? Tough. Injured? Tough. Playing in a different position? You guessed it...
So Charles Piutau is only the 14th best full-back because he missed the first three games of the season and is also the 21st best centre. Personally I think there should be a page where you can view players contributions regardless of position, but hey, I didn't do it.
Meanwhile, Mark Atkinson creeps in at full-back because he had one game there and not a lot of people play full-back regularly; he doesn't get in at centre because there's a lot of people playing centre. There's even more playing prop, so Cole and Marler both don't make it because of their international duties.
3) This doesn't tell you that Harry Sloan is a better centre than Jonathan Joseph. It tells you that he was a better centre to have in the AP this season, which is probably true.
4) There's some flaws to this to say the least (we've all been over passive tackles made vs dominant tackles missed and so on) but if you treat this for it is - a rating of how well players have done solely in the AP this season - I think it's interesting and useful. Do you compare Sloan with Joseph? Probably not, but it does tell you that Bath may as have not had Joseph on the books this season, and that Sloan lags a fair bit behind other young England hopefuls such as James, Slade, Hill and so on. Since you can track when they did play and didn't play, you can also see that Robshaw and Beaumont were having big domestic seasons until England and injury hit.
And so on.
Also, Which, you c&p the post I used and then call me lazy?
- Stom
- Posts: 5743
- Joined: Wed Feb 10, 2016 10:57 am
Re: One for the stat lovers.
Oh, and it surely wouldn't be much work to add an "average score" function... That would surely iron out a few of the inconsistencies. The England players would surely figure higher for starters.
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
The most enlightening stat in this thread. I've no quibble over the algorithm for this onePeat wrote: Also, Which, you c&p the post I used and then call me lazy?
- Puja
- Posts: 17619
- Joined: Tue Feb 09, 2016 9:16 pm
Re: One for the stat lovers.
You'd've thought that'd be far more useful. At the moment, it's not so much "Who's the best in the AP" as it is "Who's played the most."Stom wrote:Oh, and it surely wouldn't be much work to add an "average score" function... That would surely iron out a few of the inconsistencies. The England players would surely figure higher for starters.
Puja
Backist Monk
- Stom
- Posts: 5743
- Joined: Wed Feb 10, 2016 10:57 am
Re: One for the stat lovers.
Except wing, where Wade just wins.Puja wrote:You'd've thought that'd be far more useful. At the moment, it's not so much "Who's the best in the AP" as it is "Who's played the most."Stom wrote:Oh, and it surely wouldn't be much work to add an "average score" function... That would surely iron out a few of the inconsistencies. The England players would surely figure higher for starters.
Puja
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
Who uses the 'I'm too busy' excuse and expects to keep their credibility (insert heavy sarcasm emoji).Peat wrote:1) It's been a busy day, I was going to post it up here, honest... (and no, it's not mine).
2) I see a lot of people have spotted the surprise results, but no one's noticed the method by which they come about.
Each player's positional score is done by adding up all the scores for their games in that position together.
International duty? Tough. Injured? Tough. Playing in a different position? You guessed it...
So Charles Piutau is only the 14th best full-back because he missed the first three games of the season and is also the 21st best centre. Personally I think there should be a page where you can view players contributions regardless of position, but hey, I didn't do it.
Meanwhile, Mark Atkinson creeps in at full-back because he had one game there and not a lot of people play full-back regularly; he doesn't get in at centre because there's a lot of people playing centre. There's even more playing prop, so Cole and Marler both don't make it because of their international duties.
3) This doesn't tell you that Harry Sloan is a better centre than Jonathan Joseph. It tells you that he was a better centre to have in the AP this season, which is probably true.
4) There's some flaws to this to say the least (we've all been over passive tackles made vs dominant tackles missed and so on) but if you treat this for it is - a rating of how well players have done solely in the AP this season - I think it's interesting and useful. Do you compare Sloan with Joseph? Probably not, but it does tell you that Bath may as have not had Joseph on the books this season, and that Sloan lags a fair bit behind other young England hopefuls such as James, Slade, Hill and so on. Since you can track when they did play and didn't play, you can also see that Robshaw and Beaumont were having big domestic seasons until England and injury hit.
And so on.
Also, Which, you c&p the post I used and then call me lazy?
Adding up scores leads to a result of who 'isn't good enough to be an international but stays fit' topping the charts. You could have a squad full of these and you wouldn't win the league. I'd certainly rather have 16 games from George Ford/Jonathon Joseph than 26 from Tom Heathcote/Harry Sloan.
As for Atkinson playing 1 game, out of position and still being a highly ranked fullback. That surely secures the award for whatever it was that I annointed him for at the start of the season.
- Mellsblue
- Posts: 15724
- Joined: Thu Feb 11, 2016 7:58 am
Re: One for the stat lovers.
They've gone with the basic idea that number of appearances is more important than quality during appearances. Something I'd strongly dispute. However, there is obviously a tipping point where it doesn't matter how good you are you just aren't playing enough. If the graph for this hasn't been invented yet, I'm naming it the Manu Tuilagi Curve.Stom wrote:Except wing, where Wade just wins.Puja wrote:You'd've thought that'd be far more useful. At the moment, it's not so much "Who's the best in the AP" as it is "Who's played the most."Stom wrote:Oh, and it surely wouldn't be much work to add an "average score" function... That would surely iron out a few of the inconsistencies. The England players would surely figure higher for starters.
Puja
- Which Tyler
- Posts: 9038
- Joined: Tue Feb 09, 2016 8:43 pm
- Location: Tewkesbury
- Contact:
Re: One for the stat lovers.
And it would be so easy to include - the data collected already includes minutes on the pitch; simply divide the final score by minutes played - easy. (X60 if you want to end up with larger numbers).Stom wrote:Oh, and it surely wouldn't be much work to add an "average score" function... That would surely iron out a few of the inconsistencies. The England players would surely figure higher for starters.
EG: Joseph ends up with 768 points; having played 716 minutes (total, statbunker, regardless of position); score of 1.07 per minute - or 64.4 per hour, or 85.8 per 80 minutes; or whatever.
Daly, on the other hand; has 1602 points from 1241 minutes (total... etc); score of 1.29 per minute - or 77.5 per hour, or 103.3 per 80 minutes; etc.
As for laziness - I make no attempt to defend against accusations of hypocrisy - I'm far to lazy for that.
-
- Posts: 15261
- Joined: Fri Feb 12, 2016 11:17 am
Re: One for the stat lovers.
Punning or laziness?Which Tyler wrote: I'm far to lazy for that.
-
- Posts: 1038
- Joined: Tue Feb 09, 2016 7:09 pm
Re: One for the stat lovers.
I have credibility here?Mellsblue wrote: Who uses the 'I'm too busy' excuse and expects to keep their credibility (insert heavy sarcasm emoji).
Adding up scores leads to a result of who 'isn't good enough to be an international but stays fit' topping the charts. You could have a squad full of these and you wouldn't win the league. I'd certainly rather have 16 games from George Ford/Jonathon Joseph than 26 from Tom Heathcote/Harry Sloan.
As for Atkinson playing 1 game, out of position and still being a highly ranked fullback. That surely secures the award for whatever it was that I annointed him for at the start of the season.
Being in the top 30 =/= being highly ranked. I'd like to imagine common sense alone tell us that and if it didn't, Atkinson's position in the list should.
And Joseph played 9 games, scoring in none of them. He's a class act, but he hasn't been this season. It's not like Harry Sloan is even all that high.
I'm not going to pretend this thing doesn't flaws, and fair enough if you don't like it, but you're misrepresenting this quite a lot. A squad full of players like Ayerza, Brits, Parling, Smith, Waldrom, Cipriani, Rokoduguni, Wade and Good wouldn't win the league?