Agreed. But to then sign Tonali, Bruno G, Wharton Baleba or Elliot Anderson is not going to change that, we need a regista, a player who can dictate the tempo of the game, controll the ball and pass, then we have to look elsewhere.
Centre Midfielder Transfer Links
- Thread starter luke511
- Start date
-
In the latest edition of the United Hour Podcast, we interview Duncan Drasdo of MUST - covering ticket prices, new stadium, INEOS and what fans should be demanding from our club.
Also, due to technical issues we couldn't put a notice up for the previous episode which was about the game against Bournemouth and other stuff. If you'd like to listen to it, click here.
I agree with this, I just don't think that we would find the fee Everton would demand palatable considering our history with the player. If I was the one signing the cheque I'd go with him, Anderson and Gomes which, like I said in the previous post, would catapult us to the best midfield in the league. He is as sure a bet as you can get, knows the league and has a point to prove at a club that didn't give him a chance. Dealing with Everton, though....
I'm fairly happy with this as an overarching summary of midfielders' Offensive vs Defensive output. It seems a fairly decent amalgamation of a number of different key metrics, and shows a fairly comprehensive view of players. Top 5 leagues for now.
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented (Padj)' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points for transition value. I like this measure overall, as unlike say just PAdj tackles+interceptions, this seems to give a good weight to defending the penalty box; this seems valuable type of information for United given the loss of Casemiro who is excellent in these areas.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented (Padj)' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points for transition value. I like this measure overall, as unlike say just PAdj tackles+interceptions, this seems to give a good weight to defending the penalty box; this seems valuable type of information for United given the loss of Casemiro who is excellent in these areas.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
Last edited:
I'm fairly happy with this as an overarching summary of midfielders' Offensive vs Defensive output. It seems a fairly decent amalgamation of a number of different key metrics, and shows a fairly comprehensive view of players. Top 5 leagues for now.
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points from the transition value.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
Tchouameni would be the perfect Case replacement.
He would indeed be the single ideal possible replacement, yes.
Seems very unlikely, mind - though if Real really are desperate for someone like Enzo, and are happy with keeping Chema Andres instead, then who knows?!
Thanks! It's tough to make out who should be the clear targets based on this. Most of the guys we've been linked to don't grade out well defensively (super rough numbers for Fernandes and Wharton).I'm fairly happy with this as an overarching summary of midfielders' Offensive vs Defensive output. It seems a fairly decent amalgamation of a number of different key metrics, and shows a fairly comprehensive view of players. Top 5 leagues for now.
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented (Padj)' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points for transition value. I like this measure overall, as unlike say just PAdj tackles+interceptions, this seems to give a good weight to defending the penalty box; this seems valuable type of information for United given the loss of Casemiro who is excellent in these areas.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
Taking these numbers at face value, Alex Scott seems like good value for the money based on how well his OOP work grades (or Tchouameni if available). He'd be a good defensive anchor, paired with a more expansive passer like Anderson, Stiller, or Wharton. We could target someone like a Camara/Kone/Sangare, but given they're coming from lower-level leagues I wouldn't expect them to be instant starter-level players next season.
Yeah, knowing how different players may adapt is of course tricky.
Even allowing for issues of moving, considering that the PL is typically more difficult right now than other leagues, when looking at similar players in the chart and things like Uefa coefficient I wonder if an estimate average c.10-15% drop in output of key metrics could be expected on a move from the lesser leagues to the PL? Though at least with younger signings they would still have more time to adapt and further improve after the transfer.
Uniteds scandinavian Supporters Club asked 2 of Norways biggest football experts to rank 6 United-linked midfielders as to who would fit United best - and included style of play and age
Both ranked Bruno G as number 1, Both had Elliot Anderson as number 3, they ranked Tonali as 2nd and 4th respectively and Wharton as 5th and 2nd respectively. Baleba and Joao Gomes were the 2 lowest ranked combined. But both agree that all players would in some way strengthen United.
Both ranked Bruno G as number 1, Both had Elliot Anderson as number 3, they ranked Tonali as 2nd and 4th respectively and Wharton as 5th and 2nd respectively. Baleba and Joao Gomes were the 2 lowest ranked combined. But both agree that all players would in some way strengthen United.
What kind of matrix that City used when tracking Rodri in the past? It’s worth to look at, if anyone knows. He’s the closest to young Casemiro that i can think of.
Who are the experts?
Rodri's stats have been absolutely ridiculous from the get-go.
Sadly I haven't got some of the same aggregated data scores going back to when Rodri was playing in Spain, but this is him from Atletico aged 23, the full season before Man City bought him:
And this is him already in his first full season in La Liga at 21 for Villareal:
Even just looking at his most stand-out key top stats from the top of the chart, at 21 he was 65th+ percentile for all of:
Long Pass Completion %; Short&Med Pass Completion %; Aerial Win %; Poss-Adj Tackles+Interceptions; Defensive Actions; Progressive Passes.
Many of them are >80th percentile.
Even if you reduce Rodri's percentiles aged 21 for the above 6 metrics by 10-15 points (actually his super high long pass completion and defensive actions by >20 points), and run those filters in the Big 5 Leagues for this season, you only get:
Pedri, 23
Nico Gonzalez, 24
Boubacar Kamara, 26
Vitaly Janelt, 27
Ibrahim Sangare, 28
Rodri, 29
Nicolas Seiwald, 24
Locatelli, 28
Modric, 40
if you add Portugal and Eredivisie there's also
Barrenechea, 24
Regeer, 22
Slim pickings. Most of them don't really have Rodri's creativity on the ball if you cross-reference with my XY chart above. City do already have a pretty good Rodri replacement on their books... greedy sods.
This isn't the ideal set of filters I would use btw - mainly doing it for illustration. I did a more customised "Casemiro replacement research on the same database below (a number of the same names came up though).
You can see that here:
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-34033035
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-33997872
What Chema Andres is doing in the Bundesliga at 20 definitely seems the closest thing by a sub-23 player in the big 5 leagues. He doesn't have the passing range and creativity Rodri already showed at a young age yet though.
Last edited:
Kasper Wikestad and Lars Tjernås
Think Madrid would rather dump Camavinga and his injuries on us to fund any potential Enzo move.
There're no links, and player might not even be interested .But...would anyone take Jude Bellingham for £75m?
He plays in Bruno’s position. He’s OK in the #8, but not spectacular. Only way would be with Bruno leaving.
Really isn't what we need.
Actually no - he is exactly the kind of player we dont need
Absolutely not.
What I'm reading from this is that we should test Liverpool's resolve for Curtis Jones as our long lost Frenkie alternative, and Bentaleb as our Rodri.I'm fairly happy with this as an overarching summary of midfielders' Offensive vs Defensive output. It seems a fairly decent amalgamation of a number of different key metrics, and shows a fairly comprehensive view of players. Top 5 leagues for now.
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented (Padj)' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points for transition value. I like this measure overall, as unlike say just PAdj tackles+interceptions, this seems to give a good weight to defending the penalty box; this seems valuable type of information for United given the loss of Casemiro who is excellent in these areas.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
Problem is they would have to make changes to the doors at the training ground as his ego wouldn't fit through them.
Bentaleb has reached his very peak at 31, and is playing in Ligue 1.
Curtis Jones' position is partly due to most of his appearances coming as a sub. But lots of Liverpool fans think he's their best passer of the ball and should be starting ahead of MacAllister. There is some speculation he's fallen out with Slot.
Nice attempt at cynicism though. Got anything a bit more constructive to add?
Sure, but I'd be surprised if United touched that after experiences with Mount.
It was a joke more than cynicism.
All I'll add is yes, these stats can be helpful and should be part of our decision process, but ultimately there's always missing context (league, quality of team, tactical set-up/instruction) and when Curtis Jones comes out alongside Frenkie De Jong on a data map I think it is fair to question it.
Sure, I've mentioned the impact of the league elsewhere - though that the different leagues slightly differ in quality should already be knowledge amongst most football observers.
I have already discussed and detailed here in the Joao Gomes thread, that the quality of the team does not actually seem to make that significant of an impact, e.g.
https://www.redcafe.net/threads/joao-gomes.481957/post-34158820
Of course such a simplification and streamlining of so many variables will never fully encapsulate everything, but I think it serves as a fairly reasonable and handy point of comparison.
Well last season he did start 19 Premier League matches for the team which won the league...
It's fair to question, but again as above there is also additional context. It probably does help his stats a bit that he is unusual in having had so many of his minutes as a sub - similar to the benefit Sesko can gain from being a super-sub at Utd.
Here's his more detailed stats breakdown for further context though. The problem is not really with the metrics themselves. He has one of the highest xAssists in the league, and his passing accuracy is superb:
Some interesting posts here @goldenboy
Wow, my man. Are you a blogger or some sort? Never seen this type of chart.Rodri's stats have been absolutely ridiculous from the get-go.
Sadly I haven't got some of the same aggregated data scores going back to when Rodri was playing in Spain, but this is him from Atletico aged 23, the full season before Man City bought him:
And this is him already in his first full season in La Liga at 21 for Villareal:
Even just looking at his most stand-out key top stats from the top of the chart, at 21 he was 65th+ percentile for all of:
Long Pass Completion %; Short&Med Pass Completion %; Aerial Win %; Poss-Adj Tackles+Interceptions; Defensive Actions; Progressive Passes.
Many of them are >80th percentile.
Even if you reduce Rodri's percentiles aged 21 for the above 6 metrics by 10-15 points (actually his super high long pass completion and defensive actions by >20 points), and run those filters in the Big 5 Leagues for this season, you only get:
Pedri, 23
Nico Gonzalez, 24
Boubacar Kamara, 26
Vitaly Janelt, 27
Ibrahim Sangare, 28
Rodri, 29
Nicolas Seiwald, 24
Locatelli, 28
Modric, 40
if you add Portugal and Eredivisie there's also
Barrenechea, 24
Regeer, 22
Slim pickings. Most of them don't really have Rodri's creativity on the ball if you cross-reference with my XY chart above. City do already have a pretty good Rodri replacement on their books... greedy sods.
This isn't the ideal set of filters I would use btw - mainly doing it for illustration. I did a more customised "Casemiro replacement research on the same database below (a number of the same names came up though).
You can see that here:
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-34033035
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-33997872
What Chema Andres is doing in the Bundesliga at 20 definitely seems the closest thing by a sub-23 player in the big 5 leagues. He doesn't have the passing range and creativity Rodri already showed at a young age yet though.
And now I want that kid Chema Andres, after watching his clips. Albeit I'd prefer a bit older player at the age of 24-ish, but he looks so robust and suitable for PL hustle.
I think Chema as a third midfield signing would be a fantastic idea but I’d also bring in Anderson and Tonali as first choice then have Mainoo and Chema as back up and to progress.
Baleba, Guimaraes and Chema Andres. Those would be my 3 midfield signings.
Baleba and Chema Andres for CDM
Guimaraes and Mainoo for CM
Baleba and Chema Andres for CDM
Guimaraes and Mainoo for CM
A new name to add to the mix.
Danilo. Yes the one that was at Nottingham Forest. 5 goals and 1 assist so far this season for Botafogo(9 games played).
Not going to pretend I watch the Brazilian league but those kind of stats are impressive.
Danilo. Yes the one that was at Nottingham Forest. 5 goals and 1 assist so far this season for Botafogo(9 games played).
Not going to pretend I watch the Brazilian league but those kind of stats are impressive.
What is the credible link?
There's reports that Real would sell Camavinga. I'm uncertain whether he would be a good signing or not. I don’t watch RM often but when I do it's a mixed bag of a performance from him. He would do something good and follow it up with giving the ball away cheaply. My sample size isn't big enough to say for certain so I would bow down to someone else's knowledge if they've seen him regularly
I haven't watched him enough, just spoke to a couple of Real Madrid followers at work who claim he's been shite. Still very young though.
Are we still convinced on Baleba? He’s had too much of a bad season to give me confidence he’d be a good signing
He gets shunted around by them a lot, which probably hasn’t helped his development. I like him, good physicality and technical ability but injury issues ring alarm bells.
I think Anderson is the only midfielder we should push the boat out for should he be willing to come. If he chooses City then we have to go after an assortment of midfielders whose cost in in the £45m - £70m bracket. The only other player I'd pay a hundred million for is Bruno G.
The likes of Gomes, Sangare, Chema and other names being touted should come before Baleba. He is too big a risk and especially after the season he is having. I'd rather have a couple, or three, of CMs in the under £50m bracket around the 23-25 age bracket that we can look to mould than risk it all on Baleba, if that transfer goes tits up, it will be years before we can move on from such deal.
Hope we stay away. Away from his injury concerns, everything I see about him online suggests he basically hasn’t developed much at all in the past 2-3 years.
If we’re serious about fixing the midfield then we go all out for Anderson and Wharton who’d be our first choice pairing then bring in a young high energy ball winner who’d be back up along with Mainoo, if it was set in stone Anderson was going to City we wouldn’t still have credible links to him.
If Anderson is our number one priority then treat him as such and blow City out the water and offer Anderson the vice captaincy whilst telling him he’ll be our midfield leader for the next decade, Wharton would be perfect alongsude him as they offer different qualities whilst covering each others flaws.
Someone like Chema, Sangare, Gomes or Bouaddi as a young ball winner who could slowly progress as back up to Anderson would be perfect whilst Mainoo could be allowed to progress as back up to Wharton instead of being expected to boss games at his current age, we’ve neglected midfield for so many years so can’t make the same mistakes again.
If Anderson is our number one priority then treat him as such and blow City out the water and offer Anderson the vice captaincy whilst telling him he’ll be our midfield leader for the next decade, Wharton would be perfect alongsude him as they offer different qualities whilst covering each others flaws.
Someone like Chema, Sangare, Gomes or Bouaddi as a young ball winner who could slowly progress as back up to Anderson would be perfect whilst Mainoo could be allowed to progress as back up to Wharton instead of being expected to boss games at his current age, we’ve neglected midfield for so many years so can’t make the same mistakes again.
Last edited:
I'm fairly happy with this as an overarching summary of midfielders' Offensive vs Defensive output. It seems a fairly decent amalgamation of a number of different key metrics, and shows a fairly comprehensive view of players. Top 5 leagues for now.
I have been looking into a very large range of metrics and sources, and these seem near enough the most representative to me based on my observations - that is, the closest I could find without having access to the full range of data and tools professionals have.
I opted for 'Offensive Action VAEP' because I didn't have access to an xThreat value which includes goals (and VAEP gives slightly more weight to goals than xThreat anyway, which makes some sense to me). Conversely, I had some unanswered questions on how defensive VAEP works, while xThreat Prevented felt more intuitive.
Definitions (some can TLDR skip this):
- 'Offensive Action VAEP (Valuing Actions by Estimating Probabilities')' - "VAEP from offensive actions (passes, crosses, take-ons, shots, dribbles) - net impact on the team's probability of scoring from actions that progress or finish attacks. Trained with xG-based labels (socceraction framework).
Pretty self-explanatory, but here's a more detailed description:
"VAEP (Valuing Actions by Estimating Probabilities) is a framework for valuing player actions in soccer. It assigns a value to each on-the-ball action in event stream data based on its impact on the game outcome while accounting for the context in which the action happened. Intuitively, an action value reflects the action’s expected influence on the scoreline. That is, an action valued at +0.05 is expected to contribute 0.05 goals in favor of the team performing the action, whereas an action valued at -0.05 is expected to yield 0.05 goals for their opponent."
Explore it more here:
https://dtai.cs.kuleuven.be/sports/vaep/
- 'xThreat Prevented (Padj)' measures threat eliminated by defensive actions (tackles, interceptions, clearances, recoveries, shot blocks). Uses an inverted xThreat grid. If possession is recovered, credits the threat removed plus transition value; if not, credits how far the opponent was pushed back. (Padj: adjustment for team possession).
To help give a bit more of an explanation for this: defensive actions in more dangerous situations, i.e. nearer one's goal, get more weight applied to them, so for instance a shot blocked from the 6-yard box is seen as more valuable than one from 25 yards out. On its own, a tackle on the half-way line is worth less than one on the edge of your own box, but if that tackle is in an advantageous situation for a counter-attack it gains additional points for transition value. I like this measure overall, as unlike say just PAdj tackles+interceptions, this seems to give a good weight to defending the penalty box; this seems valuable type of information for United given the loss of Casemiro who is excellent in these areas.
--------------------------
For those struggling to find E. Anderson, he's at 0.26 on the X-Axis (left on Manzambi). For those struggling to find S. Tonali, he's at 0.2 on the X-Axis (just to the right of Stach).
Awesome chart. Funny you mention Johan Manzambi, he's a young player I've been keeping an eye on. Insane dribbling and carrying stats as a CM. He's almost winger like in his dribbling ability. His defensive game is decent and his passing needs more work but wow, I think he could be a very press resistant 8 or even 6.
I think «mine» would be Anderson, Scott and Manzambi. If Anderson is not possible I would go for Garner or Wharton.
Goldenboy in here illustrated with a nice graph the other day what we lose when Casemiro. His defensive contribution is elite/very good while he actually also has a good attacking contribution. Probably down to both goals and and the fact that he is a proactive passer, good at moving the ball forward. It is not easy replacing that with a single player.
But I think the combination of, say, Garner and Scott in combination would replace Casemiro defensively if one of these replace Mainoo. Both of them is really agile with significant running power compared to a combo of Mainoo and Casemiro.
What Garner and Scott cant replace is the goal contribution and the incisive passing Casemiro has added to his game the past few seasons. I think he solution to this problem is a new CB in the mould of van Hecke or Schlotterbeck and a really good attacking left back (David Raum?!). Probably van Hecke crossing our fingers Martinez can stay fit. Considering how much more fluid and fun Man Utd are watching with Martinez in the line up, someone with similar qualities, like van Hecke, would be very tempting.
Rodri's stats have been absolutely ridiculous from the get-go.
Sadly I haven't got some of the same aggregated data scores going back to when Rodri was playing in Spain, but this is him from Atletico aged 23, the full season before Man City bought him:
And this is him already in his first full season in La Liga at 21 for Villareal:
Even just looking at his most stand-out key top stats from the top of the chart, at 21 he was 65th+ percentile for all of:
Long Pass Completion %; Short&Med Pass Completion %; Aerial Win %; Poss-Adj Tackles+Interceptions; Defensive Actions; Progressive Passes.
Many of them are >80th percentile.
Even if you reduce Rodri's percentiles aged 21 for the above 6 metrics by 10-15 points (actually his super high long pass completion and defensive actions by >20 points), and run those filters in the Big 5 Leagues for this season, you only get:
Pedri, 23
Nico Gonzalez, 24
Boubacar Kamara, 26
Vitaly Janelt, 27
Ibrahim Sangare, 28
Rodri, 29
Nicolas Seiwald, 24
Locatelli, 28
Modric, 40
if you add Portugal and Eredivisie there's also
Barrenechea, 24
Regeer, 22
Slim pickings. Most of them don't really have Rodri's creativity on the ball if you cross-reference with my XY chart above. City do already have a pretty good Rodri replacement on their books... greedy sods.
This isn't the ideal set of filters I would use btw - mainly doing it for illustration. I did a more customised "Casemiro replacement research on the same database below (a number of the same names came up though).
You can see that here:
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-34033035
https://www.redcafe.net/threads/centre-midfielder-transfer-links.488776/post-33997872
What Chema Andres is doing in the Bundesliga at 20 definitely seems the closest thing by a sub-23 player in the big 5 leagues. He doesn't have the passing range and creativity Rodri already showed at a young age yet though.
Love these stats you provide.
Seiwald is totally slept on but the guy is really good. Austria's best player alongside Laimer.
The main reason I haven't really watched and considered him more is that, considering United have just bought Sesko and seem properly in for Diomande, it might be surprising if they added a third Leipzig player to their shopping list - it could spark accusations that Vivell is just being lazy in his recruiting and Leipzig are our feeder club...