The Wizard of Özil

[EliteKiller]

EK, I sourced player ratings from a variety of sources - none of which I don't believe were Whoscored. In any case, if you have a rating system that goes from 1-10, 10 being the highest score, mathematically there are 2 middle scores, 5 and 6, 5 being lower middle and 6 being upper middle. The middle scores are therefore the average, as in, the player was neither good or bad. Scores below 5 are below average, scores above 6 are above average.

Now, if you are taking an Arithmetic Mean average (which clearly none of the ratings I claimed are using and clearly I am not referencing), by totalling the ratings scored and dividing that total by the number of players, that is a totally different value that you are retrieving. Now, assuming you are quoting arithmetic mean averages (which I was not!), if the average in a game is 7, that tells me that by and large most players had higher than average performances. If Ozil scored a 6, but the average works out to be 7, Ozil performed lower than the team's average - but he still put in a performance that was upper average in the scale of 1-10. So even in this scenario, Ozil performed better than Aniym's claimed "He did nothing whatsoever in those games" statement !

You are way over complicating ....

An average is worked out by adding up a set of data and dividing it by the number of units in that set ... the range may be 1 > 10 but that's entirely irrelevant ... if every player gets a 9 and Ozil gets an 8 he is not above average and has not performed better than anyone, he's just been rated above the mid-point but is still below average for that game.

In a football match you have 22 players who are rated, thus to find the average you add those 22 ratings then divide by 22 to get a figure ...

The average in the entire EPL this season, that's every player's game rating added together and divided by the number of games they've all played, is 6.74 ... it's not 5 ... lower than 6.74 is thus a below average rating, higher than 6.74 is an above average rating.

Ozil is rated at 6.71 this season against an EPL average of 6.74 ... does that make sense now?

Erm no. It's whoscored that over complicates the overall rating. They use so many variables and apply them to all positions. It results in Adama Traore being ranked the 3rd best player in the league, well ahead of, for example, David Silva - who had more goals, more assists, more passes (3 times as many!), better completion rate of passes (by miles), more key passes. (by miles)
One of the primary reasons is that Traore got 6 MOTM awards given by Whoscored. But being decent in a game full of shit performances provides man of the match awards.

Use whoscored for inidividual stats, not their ridiculous rating system.

Erm Yes - that is how an average is worked out, did you not do basic math? ... doh

You may not like whoscored (or Opta or Sqwaka) but they all say the same thing ... Ozil was great he is now below average ... shooting the messenger doesn't change the message.

[gamechannel]

This board just can't resist the Ozio talk can it?

The guy hasn't played a single minute for a long time yet people are still finding ways to trash him. Lol

[LMAO]

This board just can't resist the Ozio talk can it?

The guy hasn't played a single minute for a long time yet people are still finding ways to trash him. Lol

I'd wager it's mainly down to his wages. If he were on £80k a week, there'd likely be less animosity towards him.

[theHotHead]

EK, I sourced player ratings from a variety of sources - none of which I don't believe were Whoscored. In any case, if you have a rating system that goes from 1-10, 10 being the highest score, mathematically there are 2 middle scores, 5 and 6, 5 being lower middle and 6 being upper middle. The middle scores are therefore the average, as in, the player was neither good or bad. Scores below 5 are below average, scores above 6 are above average.

Now, if you are taking an Arithmetic Mean average (which clearly none of the ratings I claimed are using and clearly I am not referencing), by totalling the ratings scored and dividing that total by the number of players, that is a totally different value that you are retrieving. Now, assuming you are quoting arithmetic mean averages (which I was not!), if the average in a game is 7, that tells me that by and large most players had higher than average performances. If Ozil scored a 6, but the average works out to be 7, Ozil performed lower than the team's average - but he still put in a performance that was upper average in the scale of 1-10. So even in this scenario, Ozil performed better than Aniym's claimed "He did nothing whatsoever in those games" statement !

You are way over complicating ....

An average is worked out by adding up a set of data and dividing it by the number of units in that set ... the range may be 1 > 10 but that's entirely irrelevant ... if every player gets a 9 and Ozil gets an 8 he is not above average and has not performed better than anyone, he's just been rated above the mid-point but is still below average for that game.

In a football match you have 22 players who are rated, thus to find the average you add those 22 ratings then divide by 22 to get a figure ...

The average in the entire EPL this season, that's every player's game rating added together and divided by the number of games they've all played, is 6.74 ... it's not 5 ... lower than 6.74 is thus a below average rating, higher than 6.74 is an above average rating.

Ozil is rated at 6.71 this season against an EPL average of 6.74 ... does that make sense now?

No EK it is basic maths, on a sliding scale of 1 to 10, 5.5 is slap bang in the middle, not using decimals 5 and 6 become the middle numbers. If I rated the looks of a woman who I thought was average she would get a 5 or a 6 out of 10, anything more than 6 suggests higher than average, anything less 5 suggests below average. Average CANNOT BE above 6 on a sliding scale of 1 - 10.

Now if you are taking "the average of a group of player ratings", thats an entirely different topic and entirely different argument that you and you alone are having, I am NOT talking about the Arithmetic Mean "average", as you stated at the top of this post. Based on a sliding scale rating system for performance, between 1 and 10, "5 & 6" represent an average performance. Its basic maths dude!

[EliteKiller]

No EK it is basic maths, on a sliding scale of 1 to 10, 5.5 is slap bang in the middle, not using decimals 5 and 6 become the middle numbers. If I rated the looks of a woman who I thought was average she would get a 5 or a 6 out of 10, anything more than 6 suggests higher than average, anything less 5 suggests below average. Average CANNOT BE above 6 on a sliding scale of 1 - 10.

Now if you are taking "the average of a group of player ratings", thats an entirely different topic and entirely different argument that you and you alone are having, I am NOT talking about the Arithmetic Mean "average", as you stated at the top of this post. Based on a sliding scale rating system for performance, between 1 and 10, "5 & 6" represent an average performance. Its basic maths dude!

You can't be that uneducated can you?

Ten girls in a line up five are eights and five are nines ... you say the average remains five? that's just not how an average works ... in this example of course the average can be above 5 in fact it's 8.5 .... doh !!!!!!!!!!!

Here's maybe an easier example for you to understand - your car can drive from between 0 to 100 miles an hour, is your average speed always 50? Of course not, in town it might be 30 on a motorway it might be 70 .... If there are ten cars nine doing 70 miles an hour and you doing 51 miles an hour are you doing above the 'average' speed?

Same applies with player ratings if ten players get a 7 and one player gets a 6 the one getting the six is BELOW the average .... get it?

Five is the MEDIAN number if you have each number between 0 and 10 .... it is not necessarily the AVERAGE ...

[LMAO]

Mean, median, and mode are all averages.

If there are five 8s and five 9s (the mean & median are both 8.5, and the mode is both 8 & 9), the 8s performed relatively worse than the 9s, but everyone performed above the midpoint (which is both the mean and median in this case) 0-10 scale (5) or on a 1-10 scale (5.5)

Assuming both scales (0-10 and 1-10) have normal distributions with the apex of the curves at the midpoints with a σ of 1, it's not incorrect to say the 8s and 9s performed above average.

Seems like HH is arguing absolute and EK is arguing relative.

[theHotHead]

The average in the entire EPL this season, that's every player's game rating added together and divided by the number of games they've all played, is 6.74 ... it's not 5 ... lower than 6.74 is thus a below average rating, higher than 6.74 is an above average rating.

Ozil is rated at 6.71 this season against an EPL average of 6.74 ... does that make sense now?

Ok, I will play Devil's Advocate. Lets say Ozil was rated at 6.71 and the EPL average is 6.74, are you trying to tell me that the 0.03 that Ozil is below the average equates to the claims that he did absolutely nothing in the games where he did not get assists ? Surely 0.03 below the average performance suggests that is not the case.

Thank you for making my point for me.

[theHotHead]

Mean, median, and mode are all averages.

If there are five 8s and five 9s (the mean & median are both 8.5, and the mode is both 8 & 9), the 8s performed relatively worse than the 9s, but everyone performed above the midpoint (which is both the mean and median in this case) 0-10 scale (5) or on a 1-10 scale (5.5)

Assuming both scales (0-10 and 1-10) have normal distributions with the apex of the curves at the midpoints with a σ of 1, it's not incorrect to say the 8s and 9s performed above average.

Seems like HH is arguing absolute and EK is arguing relative.

Thanks LMAO - we would be going round in circles !! I don't see how EK cannot see that, given the number range 1-10, 5.5 is the exact middle - or the point where something is neither good or bad - and so the nearest whole numbers in either direction - 5 and 6 must be in the centre because we are not using decimals.

[hs6bx]

The Ozil thread has been hijacked by aspiring mathematicians!!?

[EliteKiller]

Mean, median, and mode are all averages.

If there are five 8s and five 9s (the mean & median are both 8.5, and the mode is both 8 & 9), the 8s performed relatively worse than the 9s, but everyone performed above the midpoint (which is both the mean and median in this case) 0-10 scale (5) or on a 1-10 scale (5.5)

Assuming both scales (0-10 and 1-10) have normal distributions with the apex of the curves at the midpoints with a σ of 1, it's not incorrect to say the 8s and 9s performed above average.

Seems like HH is arguing absolute and EK is arguing relative.

It's simple isn't it ... guess basic maths is no longer part of the curriculum ...

So 5 is the mid point and we can all agree that above 5 is better than the mid point below 5 is worse .... however if the average player rating in a game is 7 then above 7 is better than average and below 7 is worse than average ... same but different ...

Now as for Ozil - Ozil has until this season always been well above the EPL average, always ranked in the top twenty players and always achieved a rating above 7 ... this season he has dropped to 6.71 he's now ranked 123rd and sits on just about the average player rating - for most players that's not a bad performance but for our 350k a week man it's not what we need ....

[Phil71]

88.2% of statistics are made up on the spot.

[jayramfootball]

EK, I sourced player ratings from a variety of sources - none of which I don't believe were Whoscored. In any case, if you have a rating system that goes from 1-10, 10 being the highest score, mathematically there are 2 middle scores, 5 and 6, 5 being lower middle and 6 being upper middle. The middle scores are therefore the average, as in, the player was neither good or bad. Scores below 5 are below average, scores above 6 are above average.

Now, if you are taking an Arithmetic Mean average (which clearly none of the ratings I claimed are using and clearly I am not referencing), by totalling the ratings scored and dividing that total by the number of players, that is a totally different value that you are retrieving. Now, assuming you are quoting arithmetic mean averages (which I was not!), if the average in a game is 7, that tells me that by and large most players had higher than average performances. If Ozil scored a 6, but the average works out to be 7, Ozil performed lower than the team's average - but he still put in a performance that was upper average in the scale of 1-10. So even in this scenario, Ozil performed better than Aniym's claimed "He did nothing whatsoever in those games" statement !

You are way over complicating ....

An average is worked out by adding up a set of data and dividing it by the number of units in that set ... the range may be 1 > 10 but that's entirely irrelevant ... if every player gets a 9 and Ozil gets an 8 he is not above average and has not performed better than anyone, he's just been rated above the mid-point but is still below average for that game.

In a football match you have 22 players who are rated, thus to find the average you add those 22 ratings then divide by 22 to get a figure ...

The average in the entire EPL this season, that's every player's game rating added together and divided by the number of games they've all played, is 6.74 ... it's not 5 ... lower than 6.74 is thus a below average rating, higher than 6.74 is an above average rating.

Ozil is rated at 6.71 this season against an EPL average of 6.74 ... does that make sense now?

Erm no. It's whoscored that over complicates the overall rating. They use so many variables and apply them to all positions. It results in Adama Traore being ranked the 3rd best player in the league, well ahead of, for example, David Silva - who had more goals, more assists, more passes (3 times as many!), better completion rate of passes (by miles), more key passes. (by miles)
One of the primary reasons is that Traore got 6 MOTM awards given by Whoscored. But being decent in a game full of shit performances provides man of the match awards.

Use whoscored for inidividual stats, not their ridiculous rating system.

Erm Yes - that is how an average is worked out, did you not do basic math? ... doh

You may not like whoscored (or Opta or Sqwaka) but they all say the same thing ... Ozil was great he is now below average ... shooting the messenger doesn't change the message.

I am not arguing about averages.
I am saying that the average is irrelevant because the Whoscored overall rating itself is complete nonsense.

I am going to flesh out the example i gave you:

Adama Traore - played across the midfield this year, but mainly as a wide midfielder. He made the team of the season based on his ratings and that rating also placed him as 3rd best player in the Premier League.

Scored 4 goals and got 9 assists.
He averaged 1.2 shots per game.
His passes per game were very low, just 19.2 per game and he didn't do a good job of finding his team mates, with just a 74.4% pass completion rate.
He averaged 1.3 key passes per game and put in 1.2 crosses per game.
Defensively he had 0.9 tackles per game and 0.4 interceptions per game, 0.4 clearances and 0.1 blocks per game
He was dispossessed 1.8 times per game on average and had 2.4 poor touches per game
He was given 6 MoTM awards

Let's looks at David Silva.

Scored 6 goals and got 10 assists.
He averaged 1.7 shots per game.
His passes per game were 54.9 and he had an 89.3% pass completion rate.
He averaged 1.8 key passes per game and put in 0.1 crosses per game.
Defensively he had 1.0 tackles per game and 0.5 interceptions per game, 0.2 clearances and 0.1 blocks per game
He was dispossessed 0.7 times per game on average and had 1.3 poor touches per game
He was given 3 MoTM awards

For Silva I have colour coded where he had better numbers than Troare in green and worse numbers in red.

Despite all of that Silva had a rating of 7.29 and Traore a rating of 7.49.

This is just 1 example, but there are many oddities. Saint-Maximillion is also rated higher than David Silva...Pepe is rated higher than Aubameyang... Ricardo Pereira was apparently the second best player in the entire league.

Tell me why - show that you understand these summary ratings for players first, before we get to discussions about averages.

[EliteKiller]

Tell me why - show that you understand these summary ratings for players first, before we get to discussions about averages.

If you're going to shoot the messengers the whole debate is pointless .... each site uses it's own ratings system, the point is that same system is applied equally to every player .... it's therefore about as level a playing field as you can get.

How would you compare footballers if you're going to ignore all the top statistical comparisons sites - Whoscored, Opta, Squawka - who all have Ozil ranked at somewhere between 120 and 150 from the EPL players last season, just what criteria do you want to use?

If you use the purely subjective player ratings given by posters on this site then Ozil is onr of our worst performers, certainly well below average ...

If you won't accept any of the recognised football analysis specialists, just what data do you want to use?

[jayramfootball]

Tell me why - show that you understand these summary ratings for players first, before we get to discussions about averages.

If you're going to shoot the messengers the whole debate is pointless .... each site uses it's own ratings system, the point is that same system is applied equally to every player .... it's therefore about as level a playing field as you can get.

How would you compare footballers if you're going to ignore all the top statistical comparisons sites - Whoscored, Opta, Squawka - who all have Ozil ranked at somewhere between 120 and 150 from the EPL players last season, just what criteria do you want to use?

If you use the purely subjective player ratings given by posters on this site then Ozil is onr of our worst performers, certainly well below average ...

If you won't accept any of the recognised football analysis specialists, just what data do you want to use?

I say use the individual statistics - for example it is easy for me to see that David Silva had a better season that Traore.
The summary score is nonsense - demonstrably so.
It's not a level playing field at all - because different weightings appear to be given to different stats.
That provides a bias depending on the position played.
More bias is introduced into the whoscored summary rating because they apply bonuses to players for things like MOTM. But if you are just OK in a match but happen to be the best of a bad bunch you end up with a higher score for the game than someone else who played a blinder in another game. This is especially true for players like Silva at big clubs - he might play excellent (far better than some other guy who got a MOTM award in a dire game) but De Bryne was better.

The kind of work whoscored to is excellent - they keep a great log of all the key stats, but the model to combine those stats is just silly.
At the very least - if you are going to use it in a debate you need to understand it. I don;t think you do, which is not a criticism, because I doubt anyone outside of whoscored has actually seen the detail. It's about as valuable as a sun reporter giving his 1-10 opinion.

Here is one example from the site itself to underline how nonsensical it is to use whoscored player ratings:

Ratings are also subject to an increase or decrease based on the number of goals a team has scored or conceded, and whether the team has kept a clean sheet at 90 minutes. Such boosts in rating are affected by an individual's playing position and their time on the pitch and will be calculated at full time.

it's not a player rating at all - it's a combination of player stats and team stats.
This also explains your point about Ozils rating dropping. Arsenal have scored less goals overall and had a worse season, so his rating has been adjusted down. Same for all Arsenal players.

It makes even calculating an average across all players a futile exercise.

[gooney]

omg why are we still discussing this bug eyed greedy c***. Just close his thread. He is not part of the team and he has cult like following.