Jump to content
  • Sign up for free and receive a month's subscription

    You are viewing this page as a guest. That means you are either a member who has not logged in, or you have not yet registered with us. Signing up for an account only takes a minute and it means you will no longer see this annoying box! It will also allow you to get involved with our friendly(ish!) community and take part in the discussions on our forums. And because we're feeling generous, if you sign up for a free account we will give you a month's free trial access to our subscriber only content with no obligation to commit. Register an account and then send a private message to @dave u and he'll hook you up with a subscription.

Goalimpact - LFC Scouting


Cij
 Share

Recommended Posts

‘Moneyball’ is just a media buzzword and what Liverpool actually appear to be using to evaluate potential transfers is an algorithm known as ‘Goalimpact’, which was created around 2004 by a German physicist. It statistically measures the extent to which a player has contributed to the goal difference per minute of his team.  

 

The basic stats that Match of the Day, Sky and WhoScored focus upon -  such as tackles, completed passes, goals and assists – play no part in the formula. They have little value as they are so subjective; why should a player be penalised if a teammate fails to control an excellent pass or misses a sitter laid on a plate? Instead, Goalimpact (“GI”) employs a mathematical equation to evaluate how much more likely a team is to win if an individual is selected in the team (and the impact that player will have upon the club’s points total at the end of the season if he plays every game). Players with a current GI rating of around 110 are Premier League quality, 130 is Champions League level and 170+ is elite.

 

Basically, Goalimpact ascribes a numeric value to reflect the ‘influence’ that a specific player has on the goals scored and conceded by his team while he is on the pitch. It also predicts what his peak GI Rating will be at age 26. The more games played by a player during their career, the more accurate the GI rating.

 

Wijnaldum is a perfect illustration of the model in action. I could not fathom why the club pursued him out of all the options available until I learned he was ranked 17th in the world on the GI index in April 2015 (182). Liverpool fans endlessly debate what exactly he brings to the table but there is pretty much universal agreement he offers more than Emre Can and the team performs significantly better with him in it.

 

GI does not explain precisely why a team is more or less successful whenever a particular player is included but does objectively determine the extent to which the team’s performance is affected. It therefore provides a much better insight into actions which are difficult to quantify like positioning, marking, tracking, and creation of space than traditional ‘on the ball’ statistics. Our best performers this season on a GI basis have been Mane, Henderson, Wijnaldum and Lallana. The fact that two are new signings is probably no coincidence.

 

FSG infamously favour a statistical approach to player trading and it is little wonder they like GI as a metric. The creator, Jorg Seidel, recently selected groups of young players using his algorithm and benchmarked their increases in value against the Transfermarkt website to show its predictive capabilities. The returns on investment ranged from 67% to 145%; this is the prism through which investors like FSG determine value. It allows them to clinically evaluate whether a player has conformed to expectations and added to the performance of the team.

 

Until I stumbled across GI I could not work out the common thread between our recent transfer targets; Konoplyanka (c. 138 when we were linked), Gotze (200), Draxler (), Dembele, Teixeira, Brandt. Their backgrounds and career achievements are so disparate that it seemed impossible to draw any comparison. However, they all have excellent GI rankings. In fact, Brandt and Gotze’s ratings are staggeringly high and it is therefore little wonder they attracted our attention. Sadio Mane’s GI rank is 177 and that is the territory Brandt is currently projected to reach.

 

Goalimpact is by no means a secret in the football world and goes a long way towards explaining why our targets so often mirror Dortmund and Spurs; similar tools are being used to inform decisions on transfers at various clubs around Europe.  Wolfsburg, Southampton and the Red Bull clubs have used this approach to excellent effect.

 

Don’t be surprised to see us linked to Martin Hinteregger (Augsburg @ 190), Ilsanker (178), Werner, Keita, Sabitzer (Red Bull Leipzig), Kolasinac (Schalke) or Brandt (Leverkusen @ 142 presently) this Summer. The full GI Index is a closely guarded secret but the Hamburg-based company that produces it periodically releases limited information through Twitter and its website. I have no doubt that the GI Index rankings from 20 – 300 would provide a very clear picture of our targets over coming transfer windows if you focused on players under 26.  

 

TL;DR Goalimpact is how we scout players. If a player linked to us does not have a GI rating of 120+ we probably are not buying him – certainly not for any serious money. We need an average GI in excess of 135 to have a good shout of Champions’ League qualification.  

  • Upvote 8
Link to comment
Share on other sites

Did you write this Cij, or is it from a blog/article?

 

I like Goal Impact, but I'm not sure if many clubs are using something like it, at least in house anyway, as most clubs tend not to hire people with the capabilities to do it. I know Swansea have hired someone who has a similar model, so will be interested to see what signings they make.

Link to comment
Share on other sites

Goalimpact is how we scout players.

It isn't.

 

(If we use Goal Impact either on it's own or part of a wider package of analytic tools) It feeds into the data gathering and is a part of how we scout players.

 

It's an important distinction given the shit storm that is about to take place.

Link to comment
Share on other sites

‘Moneyball’ is just a media buzzword and what Liverpool actually appear to be using to evaluate potential transfers is an algorithm known as ‘Goalimpact’, which was created around 2004 by a German physicist. It statistically measures the extent to which a player has contributed to the goal difference per minute of his team.

 

The basic stats that Match of the Day, Sky and WhoScored focus upon - such as tackles, completed passes, goals and assists – play no part in the formula. They have little value as they are so subjective; why should a player be penalised if a teammate fails to control an excellent pass or misses a sitter laid on a plate? Instead, Goalimpact (“GI”) employs a mathematical equation to evaluate how much more likely a team is to win if an individual is selected in the team (and the impact that player will have upon the club’s points total at the end of the season if he plays every game). Players with a current GI rating of around 110 are Premier League quality, 130 is Champions League level and 170+ is elite.

 

Basically, Goalimpact ascribes a numeric value to reflect the ‘influence’ that a specific player has on the goals scored and conceded by his team while he is on the pitch. It also predicts what his peak GI Rating will be at age 26. The more games played by a player during their career, the more accurate the GI rating.

 

Wijnaldum is a perfect illustration of the model in action. I could not fathom why the club pursued him out of all the options available until I learned he was ranked 17th in the world on the GI index in April 2015 (182). Liverpool fans endlessly debate what exactly he brings to the table but there is pretty much universal agreement he offers more than Emre Can and the team performs significantly better with him in it.

 

GI does not explain precisely why a team is more or less successful whenever a particular player is included but does objectively determine the extent to which the team’s performance is affected. It therefore provides a much better insight into actions which are difficult to quantify like positioning, marking, tracking, and creation of space than traditional ‘on the ball’ statistics. Our best performers this season on a GI basis have been Mane, Henderson, Wijnaldum and Lallana. The fact that two are new signings is probably no coincidence.

 

FSG infamously favour a statistical approach to player trading and it is little wonder they like GI as a metric. The creator, Jorg Seidel, recently selected groups of young players using his algorithm and benchmarked their increases in value against the Transfermarkt website to show its predictive capabilities. The returns on investment ranged from 67% to 145%; this is the prism through which investors like FSG determine value. It allows them to clinically evaluate whether a player has conformed to expectations and added to the performance of the team.

 

Until I stumbled across GI I could not work out the common thread between our recent transfer targets; Konoplyanka (c. 138 when we were linked), Gotze (200), Draxler (), Dembele, Teixeira, Brandt. Their backgrounds and career achievements are so disparate that it seemed impossible to draw any comparison. However, they all have excellent GI rankings. In fact, Brandt and Gotze’s ratings are staggeringly high and it is therefore little wonder they attracted our attention. Sadio Mane’s GI rank is 177 and that is the territory Brandt is currently projected to reach.

 

Goalimpact is by no means a secret in the football world and goes a long way towards explaining why our targets so often mirror Dortmund and Spurs; similar tools are being used to inform decisions on transfers at various clubs around Europe. Wolfsburg, Southampton and the Red Bull clubs have used this approach to excellent effect.

 

Don’t be surprised to see us linked to Martin Hinteregger (Augsburg @ 190), Ilsanker (178), Werner, Keita, Sabitzer (Red Bull Leipzig), Kolasinac (Schalke) or Brandt (Leverkusen @ 142 presently) this Summer. The full GI Index is a closely guarded secret but the Hamburg-based company that produces it periodically releases limited information through Twitter and its website. I have no doubt that the GI Index rankings from 20 – 300 would provide a very clear picture of our targets over coming transfer windows if you focused on players under 26.

 

TL;DR Goalimpact is how we scout players. If a player linked to us does not have a GI rating of 120+ we probably are not buying him – certainly not for any serious money. We need an average GI in excess of 135 to have a good shout of Champions’ League qualification.

Fuck me, this shit is boring.

  • Upvote 2
Link to comment
Share on other sites

Did you write this Cij, or is it from a blog/article?

 

I like Goal Impact, but I'm not sure if many clubs are using something like it, at least in house anyway, as most clubs tend not to hire people with the capabilities to do it. I know Swansea have hired someone who has a similar model, so will be interested to see what signings they make.

 

Just my own views Woolster. When you check the players in which the usual reliable sources have confirmed we hold an interest, they all have a high GI rating. I can't state as a fact LFC use Goalimpact specifically but there is too much of a correlation for the club not to be employing a very similar algorithm. It is definitely much closer to the truth than the simplistic term 'Moneyball'. 

 

Your point on Swansea makes sense as Llorente and Narsingh are both signings who chime with this sort of approach.

  • Upvote 1
Link to comment
Share on other sites

It isn't.

 

(If we use Goal Impact either on it's own or part of a wider package of analytic tools) It feeds into the data gathering and is a part of how we scout players.

 

It's an important distinction given the shit storm that is about to take place.

 

I agree and hope an algorithm like this is simply used to identify players worth a further look by scouts or to rule out targets which are 'overvalued' by the management team / scouts (it could have eliminated Benteke and Carroll as potential signings for example) as part of a multi-stage process. 

  • Upvote 1
Link to comment
Share on other sites

liv.png

 

This was the GI Rankings of our squad in January 2016.

 

Despite being based on an equation, it strikes me as a pretty accurate assessment. We only had 8 players with a GI Rank above 130, which definitely wasn't sufficient for CL qualification.

 

Benteke is a slightly above average Premier League footballer at best and was never worth £32.5m.

 

Caulker was not of Premier League standard, making him a pointless signing.

 

Joe Allen was Champions League level, as was illustrated by his performance for Wales in the Summer. However, his injury record made him a bit of liability and not worth persevering with. 

 

Ibe was just about Premier League level and there was virtually no prospect of him mirroring Sterling's development.

 

Firmino has actually bucked the trend this season and performed above expectations so his GI has been adjusted accordingly this season - he is now of CL standard. 

 

The surprising one for me was Clyne. I always rated him as a more than decent PL full back but perhaps I overvalue his contribution based on some of the shit we have had to endure over the years!  

  • Upvote 1
Link to comment
Share on other sites

That isn't quite how to read it Strontium. GI shows the relative benefit over the alternative; so the numbers show that Mignolet was significantly better than the alternatives (Brad Jones / Danny Ward) but his development has under-performed what was expected of him. He should have reached 166 but only managed 144 at his peak. He is basically a CL quality keeper but not an elite one.

 

Sakho and Henderson were two of our best performers and have reached their potential. They are both CL level footballers which strikes me as correct. Pity about the drugs and messing from Mamadou but there you go.

 

There is a reason why Klopp has kept Lucas and this is perhaps explained by these figures. He is not our best player but has a positive impact when he plays and is therefore reliable.

 

That particular chart omits Lovren - who is rated slightly worse than Sakho - and Coutinho.  

  • Upvote 1
Link to comment
Share on other sites

The following table lists the Top 50 most influential players in the world in December 2016. It shouldn't be read literally; Gotze is obviously not better than Messi. The point though is that Dortmund's level drops off more without Gotze than Barca's without Messi. The GI equation also happens to be a pretty good barometer of the best players and when supplemented by scouting should help to sort the wheat from the chaff. All I am trying to illustrate is how 'Moneyball' actually works in practice. This type of information is what is informing Michael Edward's decisions on our transfer targets.

 

Rank Player Team Goalimpact Player Age PeakGI Nationality Previous
Rank

1 Mario Götze Borussia Dortmund 216,51 24,50 221,74 Deutschland 3

2 Thomas Müller Bayern München 212,59 27,25 215,05 Deutschland 1

3 Busquets FC Barcelona 207,43 28,42 212,11 Spanien 10

4 Lionel Messi FC Barcelona 203,50 29,42 209,97 Argentinien, Spanien 8

5 Piqué FC Barcelona 199,92 29,83 207,05 Spanien 15

6 Pedro Chelsea FC 199,34 29,33 205,66 Spanien 11

7 Manuel Neuer Bayern München 198,81 30,67 210,94 Deutschland 9

8 Toni Kroos Real Madrid 197,03 26,92 198,87 Deutschland 20

9 Karim Benzema Real Madrid 195,17 28,92 200,82 Frankreich 5

10 Mesut Özil Arsenal FC 193,66 28,17 197,91 Deutschland 2

11 Cesc Fàbregas Chelsea FC 193,65 29,58 200,36 Spanien 6

12 Robert Lewandowski Bayern München 193,29 28,25 197,79 Polen 4

13 David Alaba Bayern München 192,70 24,42 198,13 Österreich 42

14 Marcelo Real Madrid 192,14 28,58 197,12 Brasilien 14

15 Theo Walcott Arsenal FC 192,10 27,75 195,56 England, Barbados 7

16 James Rodríguez Real Madrid 192,00 25,42 194,12 Kolumbien 46

17 Thiago Bayern München 191,19 25,67 192,43 Spanien, Brasilien 172

18 Martin Hinteregger FC Augsburg 190,77 24,25 196,92 Österreich 24

19 Neymar FC Barcelona 190,51 24,83 194,61 Brasilien 47

20 Víctor Valdés Middlesbrough FC 190,20 34,92 199,67 Spanien 31

21 Toby Alderweireld Tottenham Hotspur 189,54 27,75 193,07 Belgien 13

22 Jérôme Boateng Bayern München 189,53 28,25 193,97 Deutschland 52

23 Mats Hummels Bayern München 188,09 28,00 192,04 Deutschland 18

24 Holger Badstuber Bayern München 186,01 27,75 189,48 Deutschland 175

25 Kevin Kampl Bayer Leverkusen 184,73 26,17 185,02 Slowenien,
Deutschland 29

26 Gianluigi Buffon Juventus 184,24 38,83 184,61 Italien 102

27 Bartra Borussia Dortmund 184,10 25,92 184,51 Spanien 246

28 Dani Carvajal Real Madrid 183,88 24,92 187,74 Spanien 221

29 Koke Atletico Madrid 183,69 24,92 187,52 Spanien 198

30 Ángel Di María Paris Saint-Germain 183,49 28,83 188,87 Argentinien 40

31 Eden Hazard Chelsea FC 183,46 25,92 183,80 Belgien 41

32 Iker Casillas FC Porto 183,16 35,50 191,28 Spanien 55

33 Yaroslav Rakitskiy Shakhtar Donetsk 182,95 27,33 185,63 Ukraine 54

34 Gregory van der Wiel Fenerbahce 182,88 28,83 188,32 Niederlande, Curaçao 44

35 Henrikh Mkhitaryan Manchester United 182,71 27,83 186,47 Armenien 35

36 Alex Teixeira Jiangsu Sainty 182,54 26,92 184,36 Brasilien 264

37 Luis Suárez FC Barcelona 182,02 29,83 189,19 Uruguay 59

38 Cristiano Ronaldo Real Madrid 181,92 31,83 200,52 Portugal 17

39 Siem de Jong PSV Eindhoven 181,08 27,83 184,80 Niederlande 75

40 İlkay Gündoğan Manchester City 180,64 26,08 180,86 Deutschland 150

41 Xherdan Shaqiri Stoke City 180,58 25,17 183,56 Schweiz 284

42 Arturo Vidal Bayern München 180,57 29,50 187,19 Chile 23

43 Petr Čech Arsenal FC 180,11 34,50 189,85 Tschechien 32

44 Luuk de Jong PSV Eindhoven 179,64 26,25 180,18 Niederlande 130

45 Jeremain Lens Fenerbahce 179,12 29,00 184,88 Niederlande, Suriname 43

46 André Ramalho 1. FSV Mainz 05 178,82 24,83 183,03 Brasilien 259

47 Bastian Schweinsteiger Manchester United 178,74 32,33 199,90 Deutschland 25

48 Isco Real Madrid 178,35 24,58 183,17 Spanien 302

49 Marcel Schmelzer Borussia Dortmund 178,22 28,83 183,71 Deutschland 100

50 Stefan Ilsanker RB Leipzig 178,13 27,58 181,24 Österreich 12

  • Upvote 1
Link to comment
Share on other sites

I think GoalImpact is interesting, and I have little doubt that a similar algorithm is being employed by Edwards and the analytics team at LFC as a part of their recruitment strategy.

 

I personally think there are better predictors than GI when it comes to looking at a young player and his future development, but I'm sure that it's used to help weed out the list some when it comes to which players we should scout in more detail.  I think attitude is pretty big, and for attacking players I think history also shows we really, really look hard at things like xG and xA, as well as possession-chain stats that show which players contribute to possessions that end in goals (Wijnaldum does really well in this area).

 

One thing I don't like about GI as much as other statistical measures is that it has a very hard time accounting for team effects.  In other words, put the same player in a better team and their GI always goes up, and while the algorithm tries to account for this the effect is still clearly seen.  

 

I do think that if you look through the top GI lists and look for younger players who aren't at big clubs you can get some really interesting names to look at.  It really likes Naby Keita who I am a huge fan of, as well as Kevin Kampl who is massively underrated in the Bundesliga.

  • Upvote 4
Link to comment
Share on other sites

The following table lists the Top 50 most influential players in the world in December 2016. It shouldn't be read literally; Gotze is obviously not better than Messi. The point though is that Dortmund's level drops off more without Gotze than Barca's without Messi. The GI equation also happens to be a pretty good barometer of the best players and when supplemented by scouting should help to sort the wheat from the chaff. All I am trying to illustrate is how 'Moneyball' actually works in practice. This type of information is what is informing Michael Edward's decisions on our transfer targets.

 

Rank Player Team Goalimpact Player Age PeakGI Nationality Previous

Rank

1 Mario Götze Borussia Dortmund 216,51 24,50 221,74 Deutschland 3

2 Thomas Müller Bayern München 212,59 27,25 215,05 Deutschland 1

3 Busquets FC Barcelona 207,43 28,42 212,11 Spanien 10

4 Lionel Messi FC Barcelona 203,50 29,42 209,97 Argentinien, Spanien 8

5 Piqué FC Barcelona 199,92 29,83 207,05 Spanien 15

6 Pedro Chelsea FC 199,34 29,33 205,66 Spanien 11

7 Manuel Neuer Bayern München 198,81 30,67 210,94 Deutschland 9

8 Toni Kroos Real Madrid 197,03 26,92 198,87 Deutschland 20

9 Karim Benzema Real Madrid 195,17 28,92 200,82 Frankreich 5

10 Mesut Özil Arsenal FC 193,66 28,17 197,91 Deutschland 2

11 Cesc Fàbregas Chelsea FC 193,65 29,58 200,36 Spanien 6

12 Robert Lewandowski Bayern München 193,29 28,25 197,79 Polen 4

13 David Alaba Bayern München 192,70 24,42 198,13 Österreich 42

14 Marcelo Real Madrid 192,14 28,58 197,12 Brasilien 14

15 Theo Walcott Arsenal FC 192,10 27,75 195,56 England, Barbados 7

16 James Rodríguez Real Madrid 192,00 25,42 194,12 Kolumbien 46

17 Thiago Bayern München 191,19 25,67 192,43 Spanien, Brasilien 172

18 Martin Hinteregger FC Augsburg 190,77 24,25 196,92 Österreich 24

19 Neymar FC Barcelona 190,51 24,83 194,61 Brasilien 47

20 Víctor Valdés Middlesbrough FC 190,20 34,92 199,67 Spanien 31

21 Toby Alderweireld Tottenham Hotspur 189,54 27,75 193,07 Belgien 13

22 Jérôme Boateng Bayern München 189,53 28,25 193,97 Deutschland 52

23 Mats Hummels Bayern München 188,09 28,00 192,04 Deutschland 18

24 Holger Badstuber Bayern München 186,01 27,75 189,48 Deutschland 175

25 Kevin Kampl Bayer Leverkusen 184,73 26,17 185,02 Slowenien,

Deutschland 29

26 Gianluigi Buffon Juventus 184,24 38,83 184,61 Italien 102

27 Bartra Borussia Dortmund 184,10 25,92 184,51 Spanien 246

28 Dani Carvajal Real Madrid 183,88 24,92 187,74 Spanien 221

29 Koke Atletico Madrid 183,69 24,92 187,52 Spanien 198

30 Ángel Di María Paris Saint-Germain 183,49 28,83 188,87 Argentinien 40

31 Eden Hazard Chelsea FC 183,46 25,92 183,80 Belgien 41

32 Iker Casillas FC Porto 183,16 35,50 191,28 Spanien 55

33 Yaroslav Rakitskiy Shakhtar Donetsk 182,95 27,33 185,63 Ukraine 54

34 Gregory van der Wiel Fenerbahce 182,88 28,83 188,32 Niederlande, Curaçao 44

35 Henrikh Mkhitaryan Manchester United 182,71 27,83 186,47 Armenien 35

36 Alex Teixeira Jiangsu Sainty 182,54 26,92 184,36 Brasilien 264

37 Luis Suárez FC Barcelona 182,02 29,83 189,19 Uruguay 59

38 Cristiano Ronaldo Real Madrid 181,92 31,83 200,52 Portugal 17

39 Siem de Jong PSV Eindhoven 181,08 27,83 184,80 Niederlande 75

40 İlkay Gündoğan Manchester City 180,64 26,08 180,86 Deutschland 150

41 Xherdan Shaqiri Stoke City 180,58 25,17 183,56 Schweiz 284

42 Arturo Vidal Bayern München 180,57 29,50 187,19 Chile 23

43 Petr Čech Arsenal FC 180,11 34,50 189,85 Tschechien 32

44 Luuk de Jong PSV Eindhoven 179,64 26,25 180,18 Niederlande 130

45 Jeremain Lens Fenerbahce 179,12 29,00 184,88 Niederlande, Suriname 43

46 André Ramalho 1. FSV Mainz 05 178,82 24,83 183,03 Brasilien 259

47 Bastian Schweinsteiger Manchester United 178,74 32,33 199,90 Deutschland 25

48 Isco Real Madrid 178,35 24,58 183,17 Spanien 302

49 Marcel Schmelzer Borussia Dortmund 178,22 28,83 183,71 Deutschland 100

50 Stefan Ilsanker RB Leipzig 178,13 27,58 181,24 Österreich 12

I knew Gotze would've been a brilliant signing for us. Fucking Ayre.

Link to comment
Share on other sites

That isn't quite how to read it Strontium. GI shows the relative benefit over the alternative; so the numbers show that Mignolet was significantly better than the alternatives (Brad Jones / Danny Ward) but his development has under-performed what was expected of him. He should have reached 166 but only managed 144 at his peak. He is basically a CL quality keeper but not an elite one.

 

Sakho and Henderson were two of our best performers and have reached their potential. They are both CL level footballers which strikes me as correct. Pity about the drugs and messing from Mamadou but there you go.

 

There is a reason why Klopp has kept Lucas and this is perhaps explained by these figures. He is not our best player but has a positive impact when he plays and is therefore reliable.

 

That particular chart omits Lovren - who is rated slightly worse than Sakho - and Coutinho.

 

rowan.gif

  • Upvote 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share


×
×
  • Create New...