Topic: Statistics and Analytics - insight into our performance

[BEAST]

The most shocking stat I've seen this season is :

17 games against bottom 13 teams 26 goals conceded - that's 1.53 goals conceded per game!!

Extrapolated it means over the course of the season we'd concede almost 60 goals against bottom 13 at 1.76 points won per game. This is absolute and total lunacy.

Against Top 7 we win 2.20 ppg and concede 0.8 goals per game. Against top teams we're on higher level than against the dross, I had to double check the numbers as I couldn't believe it.

what's the xG against the bottom 13 teams?

I know we've played terrible for the most part vs the shit sides, but there also has to be some element of bad luck mixed in. 

Now before someone says I'm being overly optimistic, I wouldn't be surprised if the xG says we've been a little lucky against the top sides ..... don't get me wrong we're playing well but it's pretty rare to scoop so many points in this spot without a bit of luck

[JackWard33]

what's the xG against the bottom 13 teams?

I know we've played terrible for the most part vs the shit sides, but there also has to be some element of bad luck mixed in. 

Now before someone says I'm being overly optimistic, I wouldn't be surprised if the xG says we've been a little lucky against the top sides ..... don't get me wrong we're playing well but it's pretty rare to scoop so many points in this spot without a bit of luck

I haven't looked for the last month but I'd be shocked if you weren't right about this.
I know we've been unlucky with the lower teams chance conversion (most things they've hit has gone in!).

[BEAST]

I haven't looked for the last month but I'd be shocked if you weren't right about this.
I know we've been unlucky with the lower teams chance conversion (most things they've hit has gone in!).

I'm pretty sure we've lost at least one match this year where our Xg allowed was < 0.2 and our Xg for was > 1.0

I don't even know what to say about this season.  Of course people should be held accountable so to just say it's been super weird / unlucky is not right.  But at the same time, to demonize players or coaches isn't right either ..... this season has been fucked up! 

There's scenarios where we have the same number of points but the fanbase is much happier because the distribution of results is different (more wins vs bad teams and more draws vs good teams)

I'm pretty much at the point where I'm just ready to throw my hands up and say I have no idea about anything and will just let Klopp figure this out cause next year we might finish 1st or we might finish 6th who the hell knows.

[Chris~]

In some of those close matches it's been close in expected goals but it tends to be because we had a lot low quality shots vs fewer high quality chances, the Swansea and Hull ones are like this and I wouldn't be surprised if there's a couple others. In those games youd always want the high quality chances. Obviously there's some game state in this as well. Hull got a big chance to get it even when we were throwing everything for a goal so you could say we were unlucky not to be level. Sort of like how the City one is even but given we went in to full shutdown after an early goal it's fair to say they didn't create enough to say a draw would have been fair.

[McrRed]

I love Brundish's oft-repeated "if we cross the ball more than 25 times in a game, we lose".

[jepovic]

The most shocking stat I've seen this season is :

17 games against bottom 13 teams 26 goals conceded - that's 1.53 goals conceded per game!!

Extrapolated it means over the course of the season we'd concede almost 60 goals against bottom 13 at 1.76 points won per game. This is absolute and total lunacy.

Against Top 7 we win 2.20 ppg and concede 0.8 goals per game. Against top teams we're on higher level than against the dross, I had to double check the numbers as I couldn't believe it.

And it's been getting worse. We had a good start against the bottom teams, but lately it's been a nightmare

[SerbianScouser]

              via Ted Knutson

Phil is developing really nicely as a wide-forward and I think it's really important because otherwise we don't have an option where else to play him. Of course he can occasionally put a shift in as 8 but over the course of the season it's not a viable option imo. Klopp is doing a great job with him in a 'Reus' role and I'm sure he'll continue with a trend of getting better every year, he's only approaching his peak years.

[BabuYagu]

So I have a friend who (might) give me the opta player data he has for the 2016/17 season. This is the stuff sites like Squawka and whoscored use. It´s just raw data with 100 or more categories of data with things like goals, a break down of if they were left foot, right foot, headers, inside the box, outside the box etc or passes attempted, completed, forward passes, backward passes, through ball, etc.

I think the data even includes U23 matches, not sure on that one.

Would anybody be interested in having that? I warn you, there is no format to it. It´s a wall of text and in alphabetical order. You would need to import it into a spreadsheet manually I think. e.g.

Trent Alexander-Arnold, Defender
Aerial Duels lost: 1
Aerial Duels won: 0
Appearances: 7
Backward Passes: 20
Blocked Shots: 1
Blocks: 1
Catches: 0
Clean Sheets: 0
Clearances Off the Line: 0
Corners Taken (incl short corners): 0
Corners Won: 0
Crosses not Claimed: 0
Drops: 0
Duels lost: 10
Duels won: 7
Fifty Fifty: 0
Forward Passes: 28
Foul Won Penalty: 0
Games Played: 7
Goal Assists: 0
Goals: 0
Goals Conceded: 2
Goals Conceded Inside Box: 2
Goals Conceded Outside Box: 0
Goals from Inside Box: 0
Goals from Outside Box: 0
Headed Goals: 0
Interceptions: 3
Key Passes (Attempt Assists): 1
Left Foot Goals: 0
Leftside Passes: 20
Offsides: 0
Other Goals: 0
Passing Accuracy: 0
Penalties: 0
Penalties Conceded: 0
Penalties Saved: 0
Penalties Taken: 0
Penalty Goals: 0
Punches: 0
PutThrough/Blocked Distribution: 4
PutThrough/Blocked Distribution Won: 1
Recoveries: 9
Red Cards - 2nd Yellow: 0
Right Foot Goals: 0
Rightside Passes: 3
Saves from Penalty: 0
Saves Made: 0
Set Pieces Goals: 0
Shooting Accuracy: 0
Shots Off Target (inc woodwork): 0
Shots On Target ( inc goals ): 2
Starts: 2
Straight Red Cards: 0
Substitute Off: 1
Substitute On: 5
Successful Crosses & Corners: 1
Successful Crosses open play: 1
Successful Dribbles: 1
Successful Fifty Fifty: 0
Successful Long Passes: 3
Successful Passes Opposition Half: 29
Successful Passes Own Half: 18
Successful Short Passes: 43
Tackles Lost: 2
Tackles Won: 2
Through balls: 0
Time Played: 165
Total Clearances: 11
Total Fouls Conceded: 1
Total Fouls Won: 2
Total Passes: 71
Total Shots: 2
Total Successful Passes ( Excl Crosses & Corners ) : 46
Unsuccessful Crosses & Corners: 5
Unsuccessful Crosses open play: 5
Unsuccessful Long Passes: 8
Unsuccessful Passes Opposition Half: 21
Unsuccessful Passes Own Half: 9
Yellow Cards: 0
Aerial Duels: 1
Duels: 17
Total Tackles: 4
Total Red Cards: 0

[JackWard33]

And it's been getting worse. We had a good start against the bottom teams, but lately it's been a nightmare

you didn't see our run in at the end of last season I take it?

[Redcap]

That would be incredible.

[Kuks]

So I have a friend who (might) give me the opta player data he has for the 2016/17 season. This is the stuff sites like Squawka and whoscored use. It´s just raw data with 100 or more categories of data with things like goals, a break down of if they were left foot, right foot, headers, inside the box, outside the box etc or passes attempted, completed, forward passes, backward passes, through ball, etc.

I think the data even includes U23 matches, not sure on that one.

Would anybody be interested in having that? I warn you, there is no format to it. It´s a wall of text and in alphabetical order. You would need to import it into a spreadsheet manually I think. e.g.

Trent Alexander-Arnold, Defender
Aerial Duels lost: 1
Aerial Duels won: 0
Appearances: 7
Backward Passes: 20
Blocked Shots: 1
Blocks: 1
Catches: 0
Clean Sheets: 0
Clearances Off the Line: 0
Corners Taken (incl short corners): 0
Corners Won: 0
Crosses not Claimed: 0
Drops: 0
Duels lost: 10
Duels won: 7
Fifty Fifty: 0
Forward Passes: 28
Foul Won Penalty: 0
Games Played: 7
Goal Assists: 0
Goals: 0
Goals Conceded: 2
Goals Conceded Inside Box: 2
Goals Conceded Outside Box: 0
Goals from Inside Box: 0
Goals from Outside Box: 0
Headed Goals: 0
Interceptions: 3
Key Passes (Attempt Assists): 1
Left Foot Goals: 0
Leftside Passes: 20
Offsides: 0
Other Goals: 0
Passing Accuracy: 0
Penalties: 0
Penalties Conceded: 0
Penalties Saved: 0
Penalties Taken: 0
Penalty Goals: 0
Punches: 0
PutThrough/Blocked Distribution: 4
PutThrough/Blocked Distribution Won: 1
Recoveries: 9
Red Cards - 2nd Yellow: 0
Right Foot Goals: 0
Rightside Passes: 3
Saves from Penalty: 0
Saves Made: 0
Set Pieces Goals: 0
Shooting Accuracy: 0
Shots Off Target (inc woodwork): 0
Shots On Target ( inc goals ): 2
Starts: 2
Straight Red Cards: 0
Substitute Off: 1
Substitute On: 5
Successful Crosses & Corners: 1
Successful Crosses open play: 1
Successful Dribbles: 1
Successful Fifty Fifty: 0
Successful Long Passes: 3
Successful Passes Opposition Half: 29
Successful Passes Own Half: 18
Successful Short Passes: 43
Tackles Lost: 2
Tackles Won: 2
Through balls: 0
Time Played: 165
Total Clearances: 11
Total Fouls Conceded: 1
Total Fouls Won: 2
Total Passes: 71
Total Shots: 2
Total Successful Passes ( Excl Crosses & Corners ) : 46
Unsuccessful Crosses & Corners: 5
Unsuccessful Crosses open play: 5
Unsuccessful Long Passes: 8
Unsuccessful Passes Opposition Half: 21
Unsuccessful Passes Own Half: 9
Yellow Cards: 0
Aerial Duels: 1
Duels: 17
Total Tackles: 4
Total Red Cards: 0

If it's in that format, should be easy enough to manipulate into a format that can be analysed better and made a spreadsheet. I can have a crack at it if you don't mind sending it over?

[McrRed]

If it's in that format, should be easy enough to manipulate into a format that can be analysed better and made a spreadsheet. I can have a crack at it if you don't mind sending it over?

Excellent mate. That'd be cool.

As a minimum I'd ctrl-f it to find what I was after, so, yes Baba do it.
ps is it stuff available on those sites anyway? We wouldn't be handing a bibtheory type advantage to anyone?

[JCB]

So I have a friend who (might) give me the opta player data he has for the 2016/17 season.

Might be interested... is the data for every player in the EPL or just Liverpool? If it's the former then I'm in.
What I would really love is that kind of data on a granular level (per match) but damned if i know where you can find that information in bulk without having to pay for it. If anyone does please let me know.

Seeing as you're sharing, I've collected the data for every league game since the 2000-01 season (6,460 games) for some research/analysis I want to carry out on PPG.
The data is as follows:

Date = Match Date (dd/mm/yy)
HomeTeam = Home Team
AwayTeam = Away Team
FTHG = Full Time Home Team Goals
FTAG = Full Time Away Team Goals
FTR = Full Time Result (H=Home Win, D=Draw, A=Away Win)
HTHG = Half Time Home Team Goals
HTAG = Half Time Away Team Goals
HTR = Half Time Result (H=Home Win, D=Draw, A=Away Win)

Match Statistics (where available)
Attendance = Crowd Attendance      *
Referee = Match Referee
HS = Home Team Shots
AS = Away Team Shots
HST = Home Team Shots on Target
AST = Away Team Shots on Target
HHW = Home Team Hit Woodwork     *
AHW = Away Team Hit Woodwork      *      
HC = Home Team Corners
AC = Away Team Corners
HF = Home Team Fouls Committed
AF = Away Team Fouls Committed
HO = Home Team Offsides         *
AO = Away Team Offsides         *
HY = Home Team Yellow Cards
AY = Away Team Yellow Cards
HR = Home Team Red Cards
AR = Away Team Red Cards

* only first season or two
 
Happy to share it.

https://1drv.ms/f/s!AltsCVfWAaGYdDCsYewtTZYxZ18

[BabuYagu]

Excellent mate. That'd be cool.

As a minimum I'd ctrl-f it to find what I was after, so, yes Baba do it.
ps is it stuff available on those sites anyway? We wouldn't be handing a bibtheory type advantage to anyone?

Okay, once I get hold of it all I´ll pop you a PM, should be in the next couple of days

That would be incredible.

As above

If it's in that format, should be easy enough to manipulate into a format that can be analysed better and made a spreadsheet. I can have a crack at it if you don't mind sending it over?

As above, You could probably paste it into a spreadsheet and then transpose it into rows I think? I used to be good at all that shit but haven´t really fucked with Excel since 2007. Would you mind sharing the end product with me also?

Might be interested... is the data for every player in the EPL or just Liverpool? If it's the former then I'm in.
What I would really love is that kind of data on a granular level (per match) but damned if i know where you can find that information in bulk without having to pay for it. If anyone does please let me know.

Seeing as you're sharing, I've collected the data for every league game since the 2000-01 season (6,460 games) for some research/analysis I want to carry out on PPG.
/snip
https://1drv.ms/f/s!AltsCVfWAaGYdDCsYewtTZYxZ18

He offered to send me LFC´s as he knows it´s my team. He´s a work client so I don´t want to be cheeky. I might chance my arm later though and say "it would be good to have other teams data to compare it to" I´ll send you whatever he sends me though.

The "per match" stuff you can only get through opta (or similar) by paying. I might have something like that but it went totally over my head and I´ve no idea how to use it. Maybe you have more luck.

And thanks for sharing.

[Kuks]

As above, You could probably paste it into a spreadsheet and then transpose it into rows I think? I used to be good at all that shit but haven´t really fucked with Excel since 2007. Would you mind sharing the end product with me also?He offered to send me LFC´s as he knows it´s my team. He´s a work client so I don´t want to be cheeky. I might chance my arm later though and say "it would be good to have other teams data to compare it to" I´ll send you whatever he sends me though.

Haha of course would share it with you if I did do it! Send it over whenever and I'll have a crack.

If it's in the format you posted above where it's like:

Trent Alexander-Arnold, Defender
Aerial Duels lost: 1
Aerial Duels won: 0
Appearances: 7
Backward Passes: 20
Blocked Shots: 1
Blocks: 1

Transposing will probably not work but there's still a pretty quick way to do it. I think just create a new column pulling the player name and then create a new table and use If functions to populate player names down rows and stats across columns like:
             Aerial Duels Lost  Aerial Duels Won etc         etc         etc

Trent
Coutinho
Firmino
etc
etc
etc

[BabuYagu]

With the help of JCB (thanks again mate) we have a spreadsheet for this now. I´ll pm that to all those who expressed an interest above later today and anybody else that wants it.

In addition, we discovered some other tools too that we can use but would likely need a javascript coder to make the most of them. I´m reluctant to post more about it on a public forum but, if you are interested and know your way about javescript, give me a shout.

[Bob Sacamano]

538 projects us to finish 6th if anyone cares.

https://projects.fivethirtyeight.com/soccer-predictions/premier-league/

[BabuYagu]

538 projects us to finish 6th if anyone cares.

https://projects.fivethirtyeight.com/soccer-predictions/premier-league/

I still think that everything is too tight at the top to really predict anything. There will be 5 big transfers before the window shuts I think - any 1 of which will drastically change that table. The most important thing to take from it is this:-

Everton 5 times more likely to get relegated as win the league.

[Bob Sacamano]

I still think that everything is too tight at the top to really predict anything. There will be 5 big transfers before the window shuts I think - any 1 of which will drastically change that table. The most important thing to take from it is this:-

Everton 5 times more likely to get relegated as win the league.

Yeah I agree it's really hard to separate the top 6, though I think City will be a class above.

Also, I think the team ratings pick up where everything left off last year, and we limped to the finish line while Arsenal came on like a freight train. I also don't think it adjust for any transfers though I could be wrong.

[BabuYagu]

Yeah I agree it's really hard to separate the top 6, though I think City will be a class above.

Also, I think the team ratings pick up where everything left off last year, and we limped to the finish line while Arsenal came on like a freight train. I also don't think it adjust for any transfers though I could be wrong.

Well then it´s useless as a predictive model then really. If Chelsea sold their whole team and replaced with teh Sunderland team, their results would have them as a top 4 team. Ridiculous example of course - but we´ve all seen the affects losing key players has to sides.

[Bob Sacamano]

Well then it´s useless as a predictive model then really. If Chelsea sold their whole team and replaced with teh Sunderland team, their results would have them as a top 4 team. Ridiculous example of course - but we´ve all seen the affects losing key players has to sides.

I was wrong. The ratings are not a simple carryover from the end of last season. I have no idea what their formula is though for adjusting the ratings.

[joe ®]

538 projects us to finish 6th if anyone cares.

https://projects.fivethirtyeight.com/soccer-predictions/premier-league/

If Huddersfield finish bottom on 44 points I will eat my hat.

[BabuYagu]

I was wrong. The ratings are not a simple carryover from the end of last season. I have no idea what their formula is though for adjusting the ratings.

Then there is the fact we have the 6th best attack in the league despite adding Salah, one of the best attackers in europe. Burnley are 12th after selling Gray and Keane. It´s a weird table - would love to know how they did it

[BigJimFinn]

Then there is the fact we have the 6th best attack in the league despite adding Salah, one of the best attackers in europe. Burnley are 12th after selling Gray and Keane. It´s a weird table - would love to know how they did it

As a 538 fan, I tried to actually understand their methodology, and it seems like the system treats teams as singular entities and rates their strength solely on historical performance against all other teams, a kind of expanded ELO ranking system that looks separately at goal-related statistics. Unlike other sports rating and projection models used by them, mainly for basketball and baseball, this model does not look at individual players at all, and thus cannot project the impact of transfers. It is purely a backward-looking tool, and probably their new focus in trying to assess relative strength of different leagues is something where it can be useful, because this kind of approach should work best for the big picture rather than predicting individual team performance. It was seemingly created for international football and for that it also makes more sense, with no transfers and all teams playing each other in same competitions far more often than clubs split into different leagues.

Have to say to Joe that the table does not predict Huddersfield to get 44 points; their simulated points total at last place is 31.
Likewise City is projected to win the league with only 77 points. With this kind of approach, the presented average from a large number of simulations will always squeeze in the outliers, meaning that very likely the teams actually ending the season on top will have outperformed their projection and the bottom will have underperformed, and the points spread will be wider than in this predicted table.

All this with the usual caveats, based on a quick look and my flawed training in statistics. Again, I really love what 538 are doing in general, but "soccer" is not exactly within their core competence.

[Shaved Crossbar]

Good post ^

[BassTunedToRed]

If Huddersfield finish bottom on 44 points I will eat my hat.

Likewise. It says they'll finish bottom with 31 though!

[fredfrop]

This seems appropriate, loving the reasoned, well-argued narratives you guys can create.

[SerbianScouser]

https://understat.com/team/Liverpool/2017

Great site to track expected goals and assists.

[BassTunedToRed]

Based on the Euro Club Index (http://www.euroclubindex.com/) ranking, Liverpool have the easiest group in the CL.

[stejay007]

Based on the Euro Club Index (http://www.euroclubindex.com/) ranking, Liverpool have the easiest group in the CL.

And United statistically have the second hardest now do they?

[riismeister]

Based on the Euro Club Index (http://www.euroclubindex.com/) ranking, Liverpool have the easiest group in the CL.

Those numbers are ranking numbers, not a quantifiable measure of quality of the team.
Real Madrid aren't twice as good as Barcelona or 18 times better than us.

[BabuYagu]

Based on the Euro Club Index (http://www.euroclubindex.com/) ranking, Liverpool have the easiest group in the CL.

I don't disagree that maybe we have the easiest group. But you have to question the validity of that considering United's group comes out as the second hardest. For me that is by some distance the second easiest group IMO. It's a close call between us and theirs for easiest. I personally rate Seville above Benfica. Us above United (based on league last season) but rate the 3rd & 4th teams better in their group which makes it probably harder to escape from.

Low ranking teams seem to have way too much impact on the outcome. For example, Group C & H are an absolute nightmare which appear easy because of the huge score of the 4th team in the group. Then you have someone like Leipzig being rated as the 6th easiest team in the draw when I doubt anybody would believe that to be true.

A group of Real Madrid, Bayern Munich, Juventus & Maribor would come out as the easiest group simply because of the huge weighting effect of Maribor. Whereas Shakhtar, Porto, Beşiktaş & CSKA Moscow would come out as the hardest group.

[DanA]

In terms of making it to the next stage, the toughest group has to be the one with the highest ranking of the 3rd best team.
H = 13
C = 15
F = 19

I'd have group B as the easiest to qualify but ours the easiest to win.

[redmark]

If you take the ECIndex method BassTunedToRed posted, but then index within each pot (so every team is ranked 1-8, within their pots); or 1-32 across all teams, if you want to do it that way, you should get a more reasonable measure that isn't distorted by relative shitness of shit teams.

Both give D, H, C, F, B, A, G, E (compared to ECI - D, A, F, H, G, B, C, E). Which sounds about right, I think.

[BabuYagu]

If you take the ECIndex method BassTunedToRed posted, but then index within each pot (so every team is ranked 1-8, within their pots); or 1-32 across all teams, if you want to do it that way, you should get a more reasonable measure that isn't distorted by relative shitness of shit teams.

Both give D, H, C, F, B, A, G, E (compared to ECI - D, A, F, H, G, B, C, E). Which sounds about right, I think.

I believe Leipzig is still misleading though and would have G much higher. But yes, I like Dans suggestion as well as this one.

[redmark]

I believe Leipzig is still misleading though and would have G much higher. But yes, I like Dans suggestion as well as this one.

Yeah, 'most difficult to win' focusing on the top two teams, 'most difficult to qualify' on the top three, makes some sense.

[Hidegkuti]

With the help of JCB (thanks again mate) we have a spreadsheet for this now. I´ll pm that to all those who expressed an interest above later today and anybody else that wants it.

In addition, we discovered some other tools too that we can use but would likely need a javascript coder to make the most of them. I´m reluctant to post more about it on a public forum but, if you are interested and know your way about javescript, give me a shout.

Any chance I could get in on this action - would love to have access to that spreadsheet..

[BassTunedToRed]

Here are the ranking totals if you take out the weakest team. Still leaves ours as the weakest.

[Klippity Klopp]

[BabuYagu]

This is interesting. Just looking at this and ignoring fixtures, sample size, and the impact one big scoring win or loss has on things:-

I am guessing Huddersfield are looking similar to the Boro from last season? Very stingy, worries where goals come from. Swansea look like Hull. Could be in a lot of trouble if their keeper stops saving them and their attack doesn't learn how to score.

Newcastle, Burnley, Soton, West Brom look like the main contenders for top half / pushing the top 6. Arsenal and Liverpool have problems to solve to get back to top 4 form.

Everton, West Ham & Palace are in bad trouble. Bournemouth, Stoke & Leicester have problems to sort out to make sure they don't join them in trouble.