Topic: Statistics and Analytics - insight into our performance

[ScouserAtHeart]

Mods, please merge this with the Klopp Template thread if you feel this isn't new thread worthy, but I found this to be the oddest stat ever:

The KU Leuven team’s sophisticated xT-like metric, called VAEP, credited the player on the ball for the full value of each action, including the result. The 0.1 xG value of that through-ball would be assigned to the midfielder who played the pass, and the winger who made a well-timed run and controlled a tricky ball while a defender nipped at his heels would get no credit for helping his team until he tried the next pass or dribble.

They created an alternative “atomic” version of VAEP that divides a pass’s value between the passer and the receiver. The passer gets credit for however much moving the ball from Point A to Point B changes his team’s probability of scoring or conceding. Crucially, though, this number doesn’t include the result of the pass. Whether a defender intercepts the ball or a team-mate receives it, the pass value stays the same.

If the pass is complete, the receiver gets the rest of the value for beating the defence to Point B and successfully collecting the ball. Depending on the kind of pass and how likely it was to be completed, that receiving value may be worth next to nothing or it may be bigger than the passer’s share.

Adding receiving value to the equation changed how VAEP rates players. “You often see two different types of strikers: strikers that are good at scoring goals and strikers that are good at receiving the ball, keeping possession of it, and then playing to another player who scores the goal,” explained Pieter Robberechts, a PhD student who works on VAEP. The atomic version rewards that second type of striker that adds value by receiving dangerous passes and linking play.





https://theathletic.com/3098569/2022/01/29/wont-somebody-please-credit-the-pass-receiver/

It's a pretty fascinating article but it's behind a paywall so can't paste the whole thing. This blog has more details for anyone interested-- https://dtai.cs.kuleuven.be/sports/blog/valuing-on-the-ball-actions-in-soccer-a-critical-comparison-of-xt-and-vaep/

[newterp]

where is jota in the second illustration?

[JackWard33]

Given that our front 3 are the top 3 in this there must be huge team/system effects in whatever ‘receiving effects’ is

[PoetryInMotion]

Isn't this similar to Packing stats by Impect?

[Dave McCoy]

https://theathletic.com/3098569/2022/01/29/wont-somebody-please-credit-the-pass-receiver/

It's a pretty fascinating article but it's behind a paywall so can't paste the whole thing. This blog has more details for anyone interested-- https://dtai.cs.kuleuven.be/sports/blog/valuing-on-the-ball-actions-in-soccer-a-critical-comparison-of-xt-and-vaep/

In conjunction with this article you should read this:

https://absoluteunit.substack.com/p/theory-of-soccer-pt-3-balls-bouncing

It's honestly some of the best writing about the thinking behind analytics and how it matters or doesn't matter.

Isn't this similar to Packing stats by Impect?

No, they are trying to give a value to off-ball actions to the receiver of the pass. Packing was only for the person doing the passing or dribbling. The problem here is the value is only given if they receive a pass. The player can make the same run multiple times but if they don't receive a pass then there is no value in it. People have to keep in mind that almost all stats are "on-ball" due to the way data tracking works. This gets at what the article I linked is saying in that xG is great but only for it's specific thing. There's a whole world of "off-ball" actions that go into making that xG happen that can mean a lot or a little but currently there is no way to know. Or if there is then clubs like LFC, who have tracking installed at Anfield, aren't disclosing it.

Edit: And funnily enough LFC's website has a feature on Spearman today who said the following:

Tracking data sees the positioning of all 22 players and the ball tracked by cameras placed around the stadium at a rate of 25 frames per second. Along with ‘event data’ such as ball touches, this is used in Liverpool’s match analysis and recruitment work.

“With the tracking data I’m able to answer a lot of the same questions as you can with event data, but usually with a lot more context,” Spearman details.

https://www.liverpoolfc.com/news/cern-lfc-weird-journey-william-spearman-liverpools-lead-data-scientist

So LFC does try to answer those questions but whatever they've come up with they're not sharing.

[royhendo]

Worth a look/listen: https://mobile.twitter.com/AIUnderPressure/status/1516136774023913472

This squad is primed to do special things you might think.

[royhendo]

A lot of good stuff on the high line and the concession of through balls.

First Carragher on MNF and this: https://twitter.com/Carra23/status/1518674825941819395

Then this: https://twitter.com/AIUnderPressure/status/1518855151041355776

[royhendo]

There's an echo in here.



Statsbomb commentary: https://twitter.com/StatsBomb/status/1519712066411679745

Liverpool 15 Game Rolling xG Trendlines, Premier League 2015 - 2022

Seven seasons of Jürgen Klopp. Another four on the way.

-

Two things:

1) You can see the movement from Good Team to Elite Team in the second half of 2017/18 (before they hit 97 then 99 points in the following two seasons)

2) That attack has gone SUPERNOVA this season

[newterp]

There's an echo in here.



Statsbomb commentary: https://twitter.com/StatsBomb/status/1519712066411679745
 

Thank goodness it's not the Liverpool Echo - the ads would be unbearable.

[royhendo]

Just on Sky Sports News - our 139th goal for the season in all comps - highest ever total we've had in a single season... with 6 games left to go.

[grenny158]

Just on Sky Sports News - our 139th goal for the season in all comps - highest ever total we've had in a single season... with 6 games left to go.

Incredible .. and, allied to that stat is the following one:

We currently have 4 players who have scored 10 or more goals for us this season (all competitions), and 3 who have already topped 20:

Salah 30 goals / 15 assists
Jota 21 / 5
Mane 21 / 2
Firmino 11 / 4

That is very good in itself but, bubbling just under 10 goals are:

Minamino 9 / 1 (from only 23 games)
Fabinho    8 / 1

.. and then let's not forget Luis Diaz who has scored 5 and assisted 3 in 21 games - no question he would have scored over 10 had he played a full season.

Would love to see Minamino get 1 more, and Fab get 2 more, and we end with 6 players scoring 10 or more goals for us in a single season .. also not out of the realms of possibility that Luis Diaz could score another 5 goals from the remaining 6 games either.

Talk about placing your GOALden eggs in 6 or 7 baskets 

(PS. Oh, and not forgetting Divock and his 6 goals and 3 assists from only 17 games, most of which were from the bench)

[Dave McCoy]

Dream job for someone:

William Spearman
@the_spearman
·
2h
We're hiring an Academy Data Analyst:

https://recruitment.liverpoolfc.com/vacancies/jobs/view/869

[No666]

Not sure if this should go in the injury thread but would probably cause a few palpitations if that were bumped:

“I always say we are a little bit like criminal detectives.” So says Dr Andreas Schlumberger, Liverpool’s head of recovery and performance, but it is hard to imagine even the most enlightened FBI agent employing anything more sophisticated than the extraordinary technology that is helping Liverpool in their quest for an unprecedented quadruple.

Invented by the artificial intelligence company Zone7, who are based in California’s Silicon Valley, Liverpool have been using cutting-edge computer algorithms that both detect injury risk and recommend preemptive action.

It all helps to explain why manager Jurgen Klopp chose to rotate nine key players in beating Southampton on Tuesday night, barely three days after needing 120 minutes to win the FA Cup.

It perhaps also helps to explain how Liverpool have slashed their number of lost days this season to injury by more than a third and retained such remarkable performance levels across 61 games.

News of the partnership can be disclosed today for the first time, as well as Liverpool’s decision to extend their use of the artificial intelligence platform by a further two seasons, as well as with both the women’s and Under-23 teams.

The above is the key bit from:
https://www.telegraph.co.uk/football/2022/05/18/revealed-silicon-valley-algorithm-helping-liverpool-cope-history/

[royhendo]

The Anfield Index stuff tracking the minutes and workload has been really interesting and it sounds like their insight that we're managing that side perfectly is compounded by this - an expert systems approach effectively. It's bloody fascinating to think the data science approach is extending into the day to day management of other areas than recruitment and on-field analysis.

It's an incredible time to support this football club - everything we used to talk about in theoretical terms and more is being explored and actually applied eh mate?

[No666]

Indeedy. And made possible by a manager who is prepared to commit to it in the search for small margins.

[Dave McCoy]

From John Muller today:



Here are the definitions of each metric:

Possess

Build-up: likelihood that an open-play possession starting in team’s own third will reach the final third
Field tilt: team’s percentage share of both teams’ attacking third passes
Safety: likelihood that team will retain possession or move the ball out of the first 60 per cent of the pitch within eight seconds of winning it there (see counter-press in the Press section)

Disrupt

Progressive passes: average number of progressive passes (successful, open play passes that gain at least 25 per cent of the remaining distance to goal) per possession
Switches: average number of switches (successful passes that cross at least half the width of the pitch) per possession
Dribbling: average combined number of progressive carries (carries that move the ball at least 15 per cent of that remaining distance to goal) or successful take-ons per possession

Finish

Transition: likelihood that a possession starting in final 60 per cent of pitch will end in a shot within 12 seconds
Chance creation: average non-penalty expected goals per each possession that reaches final third
Set pieces: average non-penalty expected goals within eight seconds of a corner kick or free kick taken fewer than 35 metres from goal

Press

Counter-press: likelihood that team will recover a ball in the highest 60 per cent of the pitch within eight seconds of losing it there (this is the inverse of Safety in the Possess section at the top)
High press: passes allowed per defensive action in highest 40 per cent of pitch
Start distance: average starting distance from opponent’s goal of team’s open play possessions

Defend

High defence: likelihood that team will prevent opponents’ open play possessions starting in opponent’s third from reaching team’s third (see build-up in the Possess section)
Low defence: average non-penalty expected goals conceded per opponent possession that reaches team’s third (see chance creation in the Finish section)
Set-piece defence: average non-penalty expected goals conceded within eight seconds of opponent corner kick or free kick taken fewer than 35 metres from goal (see set pieces in the Finish section)

[Nobby Reserve]

Apologies if someone has already done the stats on this, but I've had the impression for a while this season that we almost always conceding the first goal so thought I'd do some mild statting and see if it's true.

Using PL games only, the overall split is:

Concede first - 11 games
Score first - 6 games
0-0 draws - 2 games


Where we've conceded first breaks down home/away as:

Home - 5 games   (out of 10)
Away - 6 games    (out of 9)


Checking how we fared after conceding first:

Home
W - 2
D - 2
L - 1

Away
W - 0
D - 1
L - 5


Overall
W - 2
D - 3
L - 6


It's a pretty damning set of statistics.

Incidentally, we've won every PL game when we have scored first (4 home / 2 away)

[A-Bomb]

Some interesting numbers in the below article - including defensive ones which could help begin to explain some of our woes.

https://www.skysports.com/football/news/11669/12778997/liverpool-form-darwin-nunez-mohamed-salah-trent-alexander-arnold-and-more-analysed

[jillcwhomever]

Some interesting numbers in the below article - including defensive ones which could help begin to explain some of our woes.

https://www.skysports.com/football/news/11669/12778997/liverpool-form-darwin-nunez-mohamed-salah-trent-alexander-arnold-and-more-analysed

That article to my mind shows how much we have missed Jota, he just gives us something extra and his tally of goals has surely been missed.

[Fromola]

The slow starts are a big issue. In both halves we're often completely switched off for the first 5 minutes and usually concede a big chance and if it goes in then our heads go. Unlike last season there's not the confidence we'll score 2 or 3 goals to come back and we know the opponent will have more big chances.

Also because we're too open. We may as well just have two banks of four, give nothing away and then pick our moments to break with whatever pace we've got, than the way we are playing that's not working. Our underlying numbers are just abysmal this season.

The full backs bombing on just catches us out now as well.

[El Lobo]

They're a shocking set of statistics and point to a fundamental flaw in how we're starting games (and it'd be nice to keep this away from 'oh but a midfielder, FSG out' stuff). And it does go back to last season where we were quite often having to come from behind too.

[A-Bomb]

That article to my mind shows how much we have missed Jota, he just gives us something extra and his tally of goals has surely been missed.

He sure has been an important member of the squad since he has joined, and we've certainly missed him this season.

Our defensive numbers are clearly lower than in previous seasons, and is partly why we're not winning games as we're conceding too many. In fact we're conceding at a rate closer to when Klopp first joined!

We can all analyse why that might be, but we are without doubt conceding more chances than in the past 4 or 5 seasons. Our pressing from the front is less effective, our midfield are not winning enough balls high enough up the pitch - our defenders are not winning their duals as much, our defensive line has dropped by several metres to compensate so our 'squeeze' between the lines is wider. In addition we've a lot of injuries and are lower in confidence.

It's a personnel and a system issue. We've got new players upfront who are not the finely tuned pressing machine and we're yet to find the right blend of those forwards for our starting x11 (for example Darwin playing centrally alters the pressing points that Mo was so familiar with when Bobby plays). Our midfield is clearly lacking in energy / sprints and grass covered....meaning we win less balls higher up the pitch. We've dropped deeper creating more space to play around us.

We clearly need new midfielders, we need to start finding combinations up front to establish the right approach using the players we have (easier said than done with all the bloody injuries)

Konate looks the natural successor in defence, however his injury record is somewhat concerning.....VVD levels have dropped and Matip is approaching the end zone of his career at the very highest level I suspect. Fabinho who has been such an important component to our success has been involved in some Alien space jam ability sucking situation.....

[jillcwhomever]

He sure has been an important member of the squad since he has joined, and we've certainly missed him this season.

Our defensive numbers are clearly lower than in previous seasons, and is partly why we're not winning games as we're conceding too many. In fact we're conceding at a rate closer to when Klopp first joined!

We can all analyse why that might be, but we are without doubt conceding more chances than in the past 4 or 5 seasons. Our pressing from the front is less effective, our midfield are not winning enough balls high enough up the pitch - our defenders are not winning their duals as much, our defensive line has dropped by several metres to compensate so our 'squeeze' between the lines is wider. In addition we've a lot of injuries and are lower in confidence.

It's a personnel and a system issue. We've got new players upfront who are not the finely tuned pressing machine and we're yet to find the right blend of those forwards for our starting x11 (for example Darwin playing centrally alters the pressing points that Mo was so familiar with when Bobby plays). Our midfield is clearly lacking in energy / sprints and grass covered....meaning we win less balls higher up the pitch. We've dropped deeper creating more space to play around us.

We clearly need new midfielders, we need to start finding combinations up front to establish the right approach using the players we have (easier said than done with all the bloody injuries)

Konate looks the natural successor in defence, however his injury record is somewhat concerning.....VVD levels have dropped and Matip is approaching the end zone of his career at the very highest level I suspect. Fabinho who has been such an important component to our success has been involved in some Alien space jam ability sucking situation.....

What won't be helping our defence is the lack of a consistent line up, we've had to use all of them as the injuries have struck which doesn't really give any of them a chance to find any consistency in performance. The pressing has obviously been a huge miss in midfield in particular which has really opened us up and with at times a makeshift defence has meant more mistakes have been made by us, especially with the chances missed down the other end. Having Jota, even Bobby back will be a help in so much as we can mix it up a bit more in attack and maybe try and find the right players to go together to bring us a less predictable attacking shape. The team needs to find some breathing space which is why more players was a must this transfer season.

[oldman]

It would be interesting to see how many big chances we missed while games were 0 - 0

if our forwards didnt miss so many then perhaps our midfield and defence wouldn't look so bad as our confidence would be up
and the opposition would be down

just a thought

[liverbloke]

treacle

no, it's not my pet name for you nobby but how i would describe our game

too slow in the build up and when we lose/give away possession then the defence doesn't look too sharp either - i'll give ali, konate and robbo (coz i like him) a pass there

so we're bound to concede first because we're like treacle whereas when we were good - so now we're bad? - we would look like a hot knife cutting through butter

but now we look languid and laboured

yeh the odd brilliant defence splitting pass is still shown at times - as is the rare solo slalom bit of skill - but overall: we look slow

[mikey_LFC]

When trying to forecast how the rest of the season my go, it is important to remember that extreme performances will more than likely regress to the mean. That means, the chances are that we will see much improved performance in the coming months without any other explanation required as to why it happened.

[killer-heels]

When trying to forecast how the rest of the season my go, it is important to remember that extreme performances will more than likely regress to the mean. That means, the chances are that we will see much improved performance in the coming months without any other explanation required as to why it happened.

We are one of the worst performing sides in terms of giving up big chances. We are not unlucky here, our underlying numbers are as bad as we are performing. This is not an extreme this is our level now.

The only thing keeping us in the top 10 was the performance of Alisson and the attack.

Where there is scope for better performance is in our finishing as Nunez and Salah have under performed their xG. But even then, thats not enough to get us 4th.

[Haggis36]

treacle

no, it's not my pet name for you nobby but how i would describe our game

too slow in the build up and when we lose/give away possession then the defence doesn't look too sharp either - i'll give ali, konate and robbo (coz i like him) a pass there

so we're bound to concede first because we're like treacle whereas when we were good - so now we're bad? - we would look like a hot knife cutting through butter

but now we look languid and laboured

yeh the odd brilliant defence splitting pass is still shown at times - as is the rare solo slalom bit of skill - but overall: we look slow

Basically. In every single part of the team, too, more or less. Just too slow to the ball all over the pitch. Slow on it, slow off it, slow in our movement, slow in transition, even slow on the break at times. Which is why I'm not sure buying midfielders completely fixes it... It helps, but I think we need to be extending the scrutiny to a few other positions in the team/squad too.

[StL-Dono]

What strikes me about those stats is how much more accurate opponents' shots are this year than before.  They're not from obviously "better" positions but can it be accounted for is that opponents have just slightly more time and are facing slightly less pressure and are having to shoot through slightly less pairs of legs than before.  I wouldn't think it would indicate that Alisson is at fault, but I suppose you could read that into it as well.  Just the eye-test wouldn't suggest that to me though....

[Fromola]

It would be interesting to see how many big chances we missed while games were 0 - 0

if our forwards didnt miss so many then perhaps our midfield and defence wouldn't look so bad as our confidence would be up
and the opposition would be down

just a thought

We're often 1-0 down before we have a chance.

That was true at the back end of last season but we just don't have the goals in us now to come back from that (unless a team like Leicester score them for us).

Ultimately we've got 3 or 4 of our main forwards out injured at the moment though and our midfield has never had many/any goals in it. Corners and set pieces were always a good way to break the deadlock at 0-0 for us but we're conceding these rather than scoring them. I don't think we've had a penalty in the league this season either.

I think much of our attacking issues have been down to injuries. If we had Diaz, Jota, Firmino, Nunez, Salah and Gakpo all fully fit right now we wouldn't be having scoreless games, or maybe one goal a game.

[BassTunedToRed]

It would be interesting to see how many big chances we missed while games were 0 - 0

if our forwards didnt miss so many then perhaps our midfield and defence wouldn't look so bad as our confidence would be up
and the opposition would be down

just a thought

I can't comment on at 0-0 specifically, but:
Liverpool have had the first big chance in 14/31 matches this season: scored 3, missed 11
They've obviously conceded the first 17 times: opponents scored 6, missed 11.

Even a switch of those numbers might've made a difference.

[harleydanger]

What’s the stats on ball recoveries this season vs last two?

[groove]

22/23: 59.8 recoveries per 90
21/22: 59.6
20/21: 57.9

[killer-heels]

22/23: 59.8 recoveries per 90
21/22: 59.6
20/21: 57.9

Probably because we are having less possession?

[Dave McCoy]

I think it's hard to parse as the tactics we've used this season have swung so drastically from game to game at times. But what has been constant is we are allowing the highest quality shots in the league and more of them then ever before. We may still get the same amount of regains or whatever but where in the past the times that we don't would lead to nothing or very rarely a shot whereas now when we don't regain the ball it's going right down our throats with a best shot opportunity in the league on top of it.

[JackWard33]

I don’t know anything about the recoveries stat in terms of how valid it is but I’d expect that where the recoveries are happening on the pitch is far more important than how many

[paddysour]

Simon Brundish posted some interesting stats regarding switches of play a couple of weeks ago

"Games last season Liverpool attempted >30 switches of play

Since the World Cup, they average 2.7 per game"

This was basically our biggest weapon with Trent, Robertson, and VVD all heavily involved. We still have these players so why aren't we attempting them anymore?

[naYoRHa2b]

Simon Brundish posted some interesting stats regarding switches of play a couple of weeks ago

"Games last season Liverpool attempted >30 switches of play

Since the World Cup, they average 2.7 per game"

This was basically our biggest weapon with Trent, Robertson, and VVD all heavily involved. We still have these players so why aren't we attempting them anymore?

Just guessing here but earlier in the season Trent was almost playing as an 8 in attack, he wasnt exactly wide. Since we've gone to a more pragmatic approach he's not really advanced enough for the switch of play. Salah is stuck on the wing, so he isn't dragging fullbacks inside which means the space isn't there for a switch.

Robbo has suffered from Mane leaving and Diaz being unavailable, there's no movement that side so the switch of play isn't on to Robbo. He's usually just marked by the winger. We're so static and predictable those patterns of play just aren't on.

[groove]

Probably because we are having less possession?

My guess would be yes, it would pretty much be the inverse of possession. I don't think it's that interesting. Ditto who scores first. The sample size for that is so small, I don't think it really tells us anything.

The switches of play stats are interesting. It seems such a weapon for us because it allows the fullbacks to make forward progress. However, neither Trent nor Robbo's number of crosses or xA are down.

[groove]

I don’t know anything about the recoveries stat in terms of how valid it is but I’d expect that where the recoveries are happening on the pitch is far more important than how many

As in we're seeing a drop in pressing (and therefore recoveries) high up the pitch?