FC Bayern Forum

MUTU wrote:

zozon wrote:

MUTU wrote:zozon, it's time for your pills again

Can you please explain to me how Coman's "chance" is rated 0.08 xG and Gnabry's post is rated at 0.07?

Just explain please. Or please never use the Understat "stats" again. Its easy to try to "calm" me down when you dont have the answer. Just diluting the conversation.

Easy, it's because you exaggerate all the chances.

You probably thought Mueller should have scored 3 goals from his chance.

Our goal was to create the most precise method for shot quality evaluation.

For this case, we trained neural network prediction algorithms with the large dataset (>100,000 shots, over 10 parameters for each).

On this site, you will find our detailed xG statistics for the top European leagues.

But yes, go with your gut feeling by all means, after all you probably did watch over 100,000 shots.

You still havent answered my question. If you rate football by neural network prediction algorithms - shame on you, really. Basically - you go with the shitty xG ratings just because its a computer thing. Says more about your perception of football than them.

You just need a pair of eyes to see which chance had a better probability to go in - and yet, Understat rates Coman's blocked shot 1m from him a better chance than Gnabry's post.

Hilarious.

Coutinho
2012/13: 1 scorer point every 311 mins played for Liverpool
2013/14: 1 scorer point every 467.2 mins played for Liverpool
2014/15: 1 scorer point every 607.4 (!!!) mins played for Liverpool
2015/16: 1 scorer point every 193.4 mins played for Liverpool
2016/17: 1 scorer point every 117.5 mins played for Liverpool
2017/18: 1 scorer point every 85.5 mins played for Barcelona & Liverpool
2018/19: 1 scorer point every 220.6 mins played for Barcelona
2019/20: 1 scorer point every 125 mins played for Bayern
Range: 85.5 to 607.4.

Mueller
2009/10: 1 scorer point every 117.9 mins played
2010/11: 1 scorer point every 113.3 mins played
2011/12: 1 scorer point every 147.1 mins played
2012/13: 1 scorer point every 87 mins played
2013/14: 1 scorer point every 80.9 mins played
2014/15: 1 scorer point every 89.2 mins played
2015/16: 1 scorer point every 83.6 mins played
2016/17: 1 scorer point every 137.5 mins played
2017/18: 1 scorer point every 103 mins played
2018/19: 1 scorer point every 146.6 mins played
2019/20: 1 scorer point every 78.6 mins played (but 1 scorer point every 32.2 mins played since Coutinho joined)
Range: 78.6 to 147.1.

zozon wrote:You just need a pair of eyes to see which chance had a better probability to go in - and yet, Understat rates Coman's blocked shot 1m from him a better chance than Gnabry's post.

Are you talking about this chance? It was difficult, but so was Gnabry's chance. I think you are misunderstanding what xG stands for. It has nothing to do with how close the player was to scoring, but rather the probability of a random player scoring from the point where the real player took the shot. Imagine your satellite/stream feed stops just as the player is about to shoot... and you're left wondering whether he scored or not... that's xG. And quit complaining about a difference of 0.01. That's a 1% chance of scoring more than the other; it's as negligible as they get.

My comparison of main characteristics (serious)

Müller pros: Vision, Positioning, Finishing, Mentality
Müller cons: Technique

Coutinho pros: Vision, Passing, Long Shots, Technique
Coutinho cons: Consistency

Of course, we had a very few games with both of them in this club, and these are not the only skills / weaknesses they have, only the most obvious ones

Hardrade wrote:My comparison of main characteristics (serious)

Müller pros: Vision, Positioning, Finishing, Mentality
Müller cons: Technique

Coutinho pros: Vision, Passing, Long Shots, Technique
Coutinho cons: Consistency

Of course, we had a very few games with both of them in this club, and these are not the only skills / weaknesses they have, only the most obvious ones

Fair enough. I'd add off-the-ball movement for Mueller unless that's what you mean by positioning.

MUTU wrote:

Hardrade wrote:My comparison of main characteristics (serious)

Müller pros: Vision, Positioning, Finishing, Mentality
Müller cons: Technique

Coutinho pros: Vision, Passing, Long Shots, Technique
Coutinho cons: Consistency

Of course, we had a very few games with both of them in this club, and these are not the only skills / weaknesses they have, only the most obvious ones

Fair enough. I'd add off-the-ball movement for Mueller unless that's what you mean by positioning.

Yeah, that’s it, FM uses positioning for def movement, and off the ball for offensive movement, I use positioning for both, depending on the context

MUTU wrote:

zozon wrote:You just need a pair of eyes to see which chance had a better probability to go in - and yet, Understat rates Coman's blocked shot 1m from him a better chance than Gnabry's post.

Are you talking about this chance? It was difficult, but so was Gnabry's chance. I think you are misunderstanding what xG stands for. It has nothing to do with how close the player was to scoring, but rather the probability of a random player scoring from the point where the real player took the shot. Imagine your satellite/stream feed stops just as the player is about to shoot... and you're left wondering whether he scored or not... that's xG. And quit complaining about a difference of 0.01. That's a 1% chance of scoring more than the other; it's as negligible as they get.

No, not that chance.

Its not about how much the difference in percentage, but that is a HIGHER rating than a definitely better chance (the Gnabry post). It could be a 0.00005 higher, but its still higher. And the Coman shot was in reality a way, WAY far from scoring.

I'll edit both chances side by side so you could see how ridiculous is xG.

PunkCapitalist wrote:

MUTU wrote:

zozon wrote:You just need a pair of eyes to see which chance had a better probability to go in - and yet, Understat rates Coman's blocked shot 1m from him a better chance than Gnabry's post.

Are you talking about this chance? It was difficult, but so was Gnabry's chance. I think you are misunderstanding what xG stands for. It has nothing to do with how close the player was to scoring, but rather the probability of a random player scoring from the point where the real player took the shot. Imagine your satellite/stream feed stops just as the player is about to shoot... and you're left wondering whether he scored or not... that's xG. And quit complaining about a difference of 0.01. That's a 1% chance of scoring more than the other; it's as negligible as they get.

Exactly. People who dispute the use of statistics such as xG and Goalimpact are people who just plainly do not understand statistics or have no clue what they're looking at.

Sent from my Moto G (5S) Plus using Tapatalk

Yeah yeah, just because people disagrew about stupid RANDOM "stats" like xG doesnt make them look ridiculous. Also xG is not a stat, its a PROBABILITY shit. Computer generated probability data are not real stats.

Get educated on what stats are and what random stuff is.

MUTU wrote:

IsiahRashad wrote:There are players with less minutes than him, but they remain up in the standings.
It's not only the goals and assits, there are other stats, also important for the game of football, and yall know that very well.

It's not about the total minutes but about the number of minutes per appearance.

Yes there's other stats, like winning possession and aerial duels. In the latter, as per your screenshot, Mueller wins 5x more aerial duels than Coutinho, and they're not even /90 stats.

This is an important distinction. Whoscored (and most other aggregate ratings) all start at a baseline rating (I wanna say for whoscored it's 6.0) and of course positive actions improve that rating whereas negative ones decrease it.

So it naturally follows that the longer a player is in a given match the more chances to perform a "positive actions" they'll have. Of course, some players are more involved than others and will accumulate these actions at a higher rate. That's fair. But we also understand that there's also a margin of time that most players require to "play into the game" and they're less involved (and by extension effective) prior to this time elapsing. That's not a hard-and-fast time limit, but I think in large part we accept this to be true for most players. So we have to keep this in mind. Mutu is right in saying minutes per appearance does have a big impact on overall whoscored rating: more minutes per match generally leads to a higher rating - or at least a greater margin of improvement over less minutes per match. And so then of course it follows that if less minutes per match (generally) leads to less positive actions and consequently a lower match rating - when you average this out over multiple matches (without taking into account minutes per match) it does stand to reason that the player who averages more minutes per match is simply more likely to earn a higher average rating than the player who averages less minutes per match.

That's not necessarily meant to be a defense of Müller nor meant to be a shot at Coutinho but just something to keep in mind when comparing players by whoscored ratings.

I think Coutinho is a very good player, but no way a 120m player. There can be better options for that price, not that Bayern would pay that anyway even if Coutinho scored hattricks in every round and won the CL .

However, I've never been a fan of Mueller and he is a little above an average player. All this rubbish about determination, making the most of your ability and being a role model . Well that may be the case in the BL but not at the very top and across the top CL teams. The others are filled with players who have all the non-technical qualities Muller has in spades and also world class skill.

Paphlagonian wrote:I need a "neither of them" choice.

Yup, same here. I'd go further and say a "Get rid of both of them" choice.

Müller had the perfect chance yesterday to add up one or two goals to his empty stats

He came in fresh for the last 10 minutes against a team that

1- was tired after 80 minutes of effort
2- was leaving plenty of space at the back as they desperatedly tried to equalize


If you compare that to Coutinho who had no physical advantage and played against a compact defense then you would expect Müller to do better.

But he didn't. He actually did it worse

For those of you which are interested in keeping an open mind to stats, I've taken the liberty of of pulling the Bundesliga expected goal information of everyone of our players who has played at least 100 minutes this season(which excludes only Goretzka and Phonzie):


I've circled the most pertinent stats for direct comparison, NPxG+xA per 90 and xGBuildup per 90. The former should be very familiar to everyone at this point, while the latter is an attempt to measure participation in buildup play(if you want a more detailed explanation of what it is, you can read this). To be clear, I'm personally a little skeptical of the value of xGBuildup, particularly because it seems to be so dependent on what style of play your team uses, but arguably it has at least a little value in making comparisons between teammates.

My own personal opinion is on stats is that they are a very useful analytic tool but not the end to all discussion. Team xG tells you one thing, individual xG + xA tells you another, individual goals scored tells you yet another, Goalimpact yet another, and whoscored ratings yet another(although I don't think these are very useful, for a variety of reasons).

As for Coutinho, I think I might rate him higher if I had ever seen any one of the many Bayern fans who claim that he is world class present a coherent, logical, and evidence-based argument explaining why he is WC. By "evidence-based" I don't necessarily mean "statistic-based." Instead, what I really mean is an argument that is based on a combination of careful observation of the various elements of Coutinho's play and analysis of what these elements of his play(and the strengths and weaknesses revealed thereupon) mean for Coutinho's value for the team.

Instead, what I see too often is arguments that amount to "Coutinho is WC because he is WC." I think you can understand why I find this frustrating. The most developed argument I think I've seen is basically "Coutinho is WC because he participates in buildup play," which for me is the equivalent of saying a winger is WC because he's fast. Coutinho's a 10, of course he participates in buildup play. Julian Brandt participates in build up play. T. Muller participates in build up play. James did a really good job participating in buildup play, yet many of the people who are such ardent partisans of Coutinho didn't think James was good enough to deserve buying. It is possible that Coutinho's buildup play is so exceptional that he provides WC value through it, but if people want to argue that, I think they need to produce a much more detailed and sophisticated analysis to support it. Otherwise, I don't think we can carry on a substantive discussion.

We can carry on with a substanstive discussion when any of you xG worshippers stop using that PREDICTION based, imaginative "what could have happened" shite.