Vincent Janssen

[Doom]

I am not 100% into the different statistical football measures, and I recall Wenger saying something recently that got me thinking.

He said: "Last season we scored 114 per cent of our expected goals. This year we've scored 50 per cent, or 55 per cent."

Someone explain how it is possible to obtain a ratio above 1 in a such measure? In my mind, it does not make sense. You have X number of expected goal chances, either you score or you miss. If we score a goal which is not an expected goal chance, shouldn't it be omitted from the goal statistics? On other words, the ratio is between 0 and 1.

I would guess it was a comment on the shot conversion rate.

Arsenal have decided somehow that a certain number of shots on average equals a certain number of goals. For convenience say 10 shots equals 5 goals - a conversion rate of 0.5.

5 goals from 10 shots or 20 goals from 40 shots = Conversion Rate of 0.5 = 100% of expected conversion rate.

6 goals from 10 shots or 24 goals from 40 shots = Conversion Rate of 0.6 = 120% of expected conversion rate.

Conversely

3 goals from 10 shots or 12 goals from 40 shots = Conversion Rate of 0.3 = 60% of expected conversion rate.

From his comments over the years it sounds to me that Wenger concentrates on
1 - increasing the number of overall shots
2 - from places on the pitch with a high probability of conversion.

The flap a fortnight ago over Wenger telling Elneny not to shot from outside the box was a case in point. Shots from outside the box have an atrocious conversion rate of about something like 2.5%. View purely from the point of view of expected goals they're basically a waste, might as well have Cech shot from the half way line.

Of course shots from outside the box have the advantage of signalling that the player is trying hard and when someone does produce a Tony Yeboah Special it looks amazing. A lot more fun than pouring over a spreadsheet of goal conversion rates.

An example of a Tony Yeboah Special

 

[scytheavatar]

Of course shots from outside the box have the advantage of signalling that the player is trying hard and when someone does produce a Tony Yeboah Special it looks amazing. A lot more fun than pouring over a spreadsheet of goal conversion rates.

Shots from outside the box also have the advantage of forcing our opponent's defense to close down on our players more and not to defend too deep. Wenger complains about our opponents defending too deep, so why hasn't he been putting these 2 together?

 

[Sanchez11]

I would guess it was a comment on the shot conversion rate.

Arsenal have decided somehow that a certain number of shots on average equals a certain number of goals. For convenience say 10 shots equals 5 goals - a conversion rate of 0.5.

5 goals from 10 shots or 20 goals from 40 shots = Conversion Rate of 0.5 = 100% of expected conversion rate.

6 goals from 10 shots or 24 goals from 40 shots = Conversion Rate of 0.6 = 120% of expected conversion rate.

Conversely

3 goals from 10 shots or 12 goals from 40 shots = Conversion Rate of 0.3 = 60% of expected conversion rate.

From his comments over the years it sounds to me that Wenger concentrates on
1 - increasing the number of overall shots
2 - from places on the pitch with a high probability of conversion.

The flap a fortnight ago over Wenger telling Elneny not to shot from outside the box was a case in point. Shots from outside the box have an atrocious conversion rate of about something like 2.5%. View purely from the point of view of expected goals they're basically a waste, might as well have Cech shot from the half way line.

Of course shots from outside the box have the advantage of signalling that the player is trying hard and when someone does produce a Tony Yeboah Special it looks amazing. A lot more fun than pouring over a spreadsheet of goal conversion rates.

An example of a Tony Yeboah Special

awsome goal by Yeboah!!

 

[razörist]

I would guess it was a comment on the shot conversion rate.

Arsenal have decided somehow that a certain number of shots on average equals a certain number of goals. For convenience say 10 shots equals 5 goals - a conversion rate of 0.5.

5 goals from 10 shots or 20 goals from 40 shots = Conversion Rate of 0.5 = 100% of expected conversion rate.

6 goals from 10 shots or 24 goals from 40 shots = Conversion Rate of 0.6 = 120% of expected conversion rate.

Conversely

3 goals from 10 shots or 12 goals from 40 shots = Conversion Rate of 0.3 = 60% of expected conversion rate.

From his comments over the years it sounds to me that Wenger concentrates on
1 - increasing the number of overall shots
2 - from places on the pitch with a high probability of conversion.

The flap a fortnight ago over Wenger telling Elneny not to shot from outside the box was a case in point. Shots from outside the box have an atrocious conversion rate of about something like 2.5%. View purely from the point of view of expected goals they're basically a waste, might as well have Cech shot from the half way line.

Of course shots from outside the box have the advantage of signalling that the player is trying hard and when someone does produce a Tony Yeboah Special it looks amazing. A lot more fun than pouring over a spreadsheet of goal conversion rates.

An example of a Tony Yeboah Special

Does not take into account:
1) rebounds
2) subsequent corners
3) variation in play
4) close shots mostly are converted by strikers, ours are ****.

 

[Doom]

Shots from outside the box also have the advantage of forcing our opponent's defense to close down on our players more and not to defend too deep. Wenger complains about our opponents defending too deep, so why hasn't he been putting these 2 together?

This might sound like I'm being purposely annoying, I'm really not, do shots from outside the box really force the other team to come out?

Like I've often thought that too but I do wonder if the theory:
shots from outside box = draw defenders out = more space = get in behind and score is just wrong.

I watch nearly every Arsenal game and when a team is sitting deep they usually still have a midfielder or two outside the box closing down our players, obviating the time to line up and take a shot, without the need for their whole team to step up say 10 meters.

As far as I can see the real problem is slow passing and lack of runs on diagonals into the box.

 

[Heisenberg]

I would guess it was a comment on the shot conversion rate.

Arsenal have decided somehow that a certain number of shots on average equals a certain number of goals. For convenience say 10 shots equals 5 goals - a conversion rate of 0.5.

5 goals from 10 shots or 20 goals from 40 shots = Conversion Rate of 0.5 = 100% of expected conversion rate.

6 goals from 10 shots or 24 goals from 40 shots = Conversion Rate of 0.6 = 120% of expected conversion rate.

Conversely

3 goals from 10 shots or 12 goals from 40 shots = Conversion Rate of 0.3 = 60% of expected conversion rate.

From his comments over the years it sounds to me that Wenger concentrates on
1 - increasing the number of overall shots
2 - from places on the pitch with a high probability of conversion.

The flap a fortnight ago over Wenger telling Elneny not to shot from outside the box was a case in point. Shots from outside the box have an atrocious conversion rate of about something like 2.5%. View purely from the point of view of expected goals they're basically a waste, might as well have Cech shot from the half way line.

Of course shots from outside the box have the advantage of signalling that the player is trying hard and when someone does produce a Tony Yeboah Special it looks amazing. A lot more fun than pouring over a spreadsheet of goal conversion rates.

You are probably right that it was about conversion. However, I though we used more sophisticated metrics to evaluate our performance. And I hope we do.

The discussion about shots from outside the box is interesting. It is clear that on average it is not worth to take the shot.
As someone mentioned shots from outside the box might force the opponent to move out of position, and then make other possibilities. However, if the defensive side also are well aware that shots from outside the box does not give many goals, why should they move away from their position?

A similar case i corner kicks. The chances for scoring a goal is supposed to be higher if you just take a short pass from the flag and then try to open up the defence from just playing. But still you see an entire stadium go crazy if the home side gets a corner kick after 93 minutes.

Funny old game.

 

[Doom]

Does not take into account:
1) rebounds
2) subsequent corners
3) variation in play
4) close shots mostly are converted by strikers, ours are ****.

All of that is captured by the expected conversion rate.

Of course the expected conversion rate is not a constant. It can be changed by signing better players like swapping Zlatan for Giroud.

 

[Trilly]

FWIW - Arseblog had a stats guy on Arsecast...damn I can't remember his name but heavily involved in football statistics and was bigging up StatsDNA as one of the best in the business.

Not that it means much, as if he was going to trash them on a heavily pro Arsenal podcast.

Arsecast are quite critical of the team when they need to be.

 

[Gooner Zig]

Arsecast are quite critical of the team when they need to be.

Agreed - I was referring more to the guest who I'm sure would have been a bit more guarded in his appraisal of StatsDNA, given he was on a pro-Arsenal pod.

 

[Yousif Arsenal]

I don't trust dutch league these days. but if he is will let us get rid of Giroud then he is welcome.

 

[Trilly]

Harry Kane's after his first half season were world class already... Just like anyone can look at Kelechi Iheanacho's production this season and say that while there will be some drop off he is as close to guaranteed world class as it's likely to get.

Dimitri Payet's numbers *before* joining west ham were among the best as a creative player in Europe for 2 years before the move. His injuries, age and risk put teams off. He was far from a no namer.

Those no namers they signed from France you speak of?? Mahrez for example has performed at the exact same level bar conversion for the last 2 years at Leicester and quite possibly before then too.

I don't know enough about Kante to make any comments there but it wouldn't shock me if the advanced stats didn't back up his current performance levels.

Clubs keep all the advanced analytics stuff behind closed doors because it's a competitive advantage but don't fool yourself into believing that it's not in use. Shoot Fat Sam built his career on analytics use but it's not exactly a popular story I guess.

I think Kante's interception and tackle numbers stood out significantly when he was in France. Remember reading it somewhere.

 

[spartandre217]

I am not 100% into the different statistical football measures, and I recall Wenger saying something recently that got me thinking.

He said: "Last season we scored 114 per cent of our expected goals. This year we've scored 50 per cent, or 55 per cent."

Someone explain how it is possible to obtain a ratio above 1 in a such measure? In my mind, it does not make sense. You have X number of expected goal chances, either you score or you miss. If we score a goal which is not an expected goal chance, shouldn't it be omitted from the goal statistics? On other words, the ratio is between 0 and 1.

It's actually really really simple.

xG is a metric between 0&1 of the quality of any single shot event.

A shot from 2 yards from the middle of the goal with no nearby defenders with the foot is likely to be pretty damn close to 1 while a shot with the head from 20 yards out and surrounded by 5 defenders is likely to be close to zero.

A team can score more goals than their xG total over the course of a single game or a single season of a) they have a team of Messi's,b) got super lucky or the other team's GK made a lot of blunders.

Over performing xG by 114% is over performance but not riding our luck too much. Under performing and scoring 50% of xG indicates absolutely woeful finishing, luck or lack of talent.

By and large though teams and leagues perform in line with xG. Correlation coefficient for xG difference with actual goals is something like 0.7 which is much higher than any other metric such as pure shots, possession, passes or team wages

 

[razörist]

Bored out of my mind, can't you guys make a thread for this in football talk or something?

 

[Doom]

I think Kante's interception and tackle numbers stood out significantly when he was in France. Remember reading it somewhere.

Kanté's standout stat was that for two seasons running he was Ligue 1's top interceptor. Presumably his distance covered was off the chart too.

Leicester's transfer policy (like a lot of clubs) is highly reliant on the stats. Their head of transfers is Steve Walsh who has worked at Chelsea and Newcastle is considered to be among the best in the league.

Arsenal signed one of Leicester's data analysts, Ben Wrigglesworth, earlier this year.

http://www.telegraph.co.uk/football...o-helped-spot-mahrez-and-kante-joins-arsenal/

A quote from that article show's what direction Arsenal are heading

"Arsenal are investing millions at the club’s Hertfordshire training base on a new player performance centre that should open next year and will specifically include data analysis and scouting suites. It also follows the purchase of the US-based football data analytics company, statDNA, for just over £2 million."

 

[spartandre217]

Was ted knutson who worked for Brentford

 

[Heisenberg]

It's actually really really simple.

xG is a metric between 0&1 of the quality of any single shot event.

A shot from 2 yards from the middle of the goal with no nearby defenders with the foot is likely to be pretty damn close to 1 while a shot with the head from 20 yards out and surrounded by 5 defenders is likely to be close to zero.

A team can score more goals than their xG total over the course of a single game or a single season of a) they have a team of Messi's,b) got super lucky or the other team's GK made a lot of blunders.

Over performing xG by 114% is over performance but not riding our luck too much. Under performing and scoring 50% of xG indicates absolutely woeful finishing, luck or lack of talent.

By and large though teams and leagues perform in line with xG. Correlation coefficient for xG difference with actual goals is something like 0.7 which is much higher than any other metric such as pure shots, possession, passes or team wages

Great explanation, and a correlation of 0.7 is pretty damn good.

It seems that we are one of the best teams - if not the best - to produce chances with a high xG, but we simply are not able to transform those into goals.
Of course, this measure cannot stand alone, but it really indicates lack of proper attacking options.

 

[spartandre217]

Great explanation, and a correlation of 0.7 is pretty damn good.

It seems that we are one of the best teams - if not the best - to produce chances with a high xG, but we simply are not able to transform those into goals.
Of course, this measure cannot stand alone, but it really indicates lack of proper attacking options.

I actually undersold xG to a degree. The R-squared value for xG difference is close to 0.65 but the straight up Pearson correlation is above 80%.
Halfway through this season it was predicting league position with a correlation of 0.92.

We've been absolute gash at converting our chances into goals this season. Part of it might be that we don't create "enough" opportunities in the sense that we create a lot more high value chances but less chances overall than other teams.

i.e. create 10 chances for a total xG of 2.5 goals vs creating 20 chances for a total xG of 2.0. xG isn't a perfect metric over single matches and some of that might simply be missed covariates or requiring a high number of chances to be "accurate".


The other thing with us that I've noticed is that while we get into great positions, we either don't hit the target or put the ball far too close to the GK which is something that most (though not all) xG models kinda miss because ball position data isn't all that reliable as things stand. So we shoot from great positions but don't place the ball well from those positions for whatever reason which plays into our underperformance.


Anywho. On topic, Knutson is a fan supposedly, so I'm ok with this

 

[eye4goal]

I think we lack composure in dangerous areas, and are not anywhere near as good as some of our previous sides when it comes to walking it into the net. The ability to assist is not shared in this team as it was in some of our best sides when you'd have Henry, Pires and Bergkamp constantly in assist double figures(some of the Emirates sides came close to that too). Someone like Bellerin could be on 10 PL assists with abit of composure and a better final ball. Sanchez can also be erratic with his decision making. That coupled with our lack of clinical finishers, and teams defending deep against us sees us struggling for goals.

This article presents a good counter argument of sorts about the use of stats

I think too often in football analytics event data gets treated as if it’s on some sort of different plane than the lived, ninety minute games themselves. If football was a little more like hockey, perhaps, with eighty game seasons and loads and loads of shots per 60 minutes, I might understand that approach. But both the shot counts and, obviously, the goals in football are not so high as to dissuade us from taking a look, no matter how time consuming (though boy it would be nice to see something like the NBA’s ‘video boxscore’ tool).

http://frontoffice.report/arsenal-and-what-we-talk-about-when-we-talk-about-expected-goals/

 

[Porkinson]

the straight up Pearson correlation is above 80%

 

[nicky47]

http://www.squawka.com/news/vincent...-ready-to-feed-off-arsenals-creativity/663104

Some information regarding this chap. Thought I'd share.