Defence or Attack - which is most important

[DoomMerchant]

Playing in the hardest of the top leagues in Europe means that we destroy each other domestically while other European teams can chill out and focus on CL.

We may have been naive in terms of setup at times but if we played in Spain or Italy we'd have enough of a sleepwalk that our energy and prep for CL would take us up another level.

[Mikhail Chigorin]

I must say that I'm absolutely loving this thread and I hope it continues for a little while longer to extend the enjoyment further......so keep on posting everyone.

However, I'd just like to ask a question, or two, of Sparty if I may :-

On the basis of the statistics you have produced on here, the posts you have made in this thread seem to suggest that you, yourself, believe that defence is more critically important than the attack.

Is this actually the case and, if so, have these studies been instrumental in bringing you to this conclusion, or have you always felt this way ??

On the other hand, in spite of everything these statistics seem to be telling you, does your heart still rule your head and do you believe that all out attack is still the (only) way forward, if you'll pardon the weak pun ??

[Ted Hughes]

The stats absolutely prove nothing & suggest the complete opposite of the way they are being interpreted imo.

The idea that Mourinho's style is a new thing is complete nonsense. There have always been loads of people trying to do what he did. The reason he has been successful is because he has access to loads of really good players.

If you gave us Hazard & him Nasri, we would be Champions this season. His defensive mentality would mean nothing.

[Im_Spartacus]

I must say that I'm absolutely loving this thread and I hope it continues for a little while longer to extend the enjoyment further......so keep on posting everyone.

However, I'd just like to ask a question, or two, of Sparty if I may :-

On the basis of the statistics you have produced on here, the posts you have made in this thread seem to suggest that you, yourself, believe that defence is more critically important than the attack.

Is this actually the case and, if so, have these studies been instrumental in bringing you to this conclusion, or have you always felt this way ??

On the other hand, in spite of everything these statistics seem to be telling you, does your heart still rule your head and do you believe that all out attack is still the (only) way forward, if you'll pardon the weak pun ??

I actually don't (or I certainly didn't) believe that defence is more important than attack. I started doing the numbers because of a post by PBL in the Chelsea thread, and I've been really surprised by what I've found. I didn't expect this at all.

I think what I've found is a key difference between the quality in depth of the PL and other European leagues, and pinpoint increased competition as the reason for the change from the first 12 years when it was effectively a 2 team league, just like Spain is, to the coming of Abramovich and the increasing sky tv money which seems to have been the catalyst for a complete change in the way British football is approached. I don't think the change is as simple as some have said, that its down to having negative managers like Mourinho or Benitez - Benitez after all, never won a title and I'm only looking at champions here. I think the money has caused a considerable increase in quality in the league, which other leagues simply can't replicate, and meaning that teams which win the title now, are more reliant on low goals conceded, than they were historically when it was a 2 team league when it was decided by who scored the most.

I think beyond Mourinho and perhaps Mancini, managers just go out to try and win by outscoring the opposition - it goes no deeper than that, Ferguson certainly did, but what has happened without design (I think), is that as the opposition has got better, the number of goals conceded has become more important than the number scored for title winners. I think it's kinda just happened and sneaked up on us all.

If I can compare Ted to a manager for a minute, some managers will simply refute such findings as a passing fad as Ted perhaps has done earlier in the thread - indeed the soccernomics book gives lots of examples of where managers' beliefs (based on perceptions, not reality) are fundamentally flawed in the numbers. Other managers will look to address the issue, build the defence first, then try to attack - exactly as Mancini did - I'm fairly sure his approach is grounded in research like this.

My personal opinion is that pragmatism has its place and may in fact be more successful title winners (Mourinho's record bears this out), but managers like Pellegrini NEED to exist to add the X-Factor to games and provide entertainment, or we would all have stopped watching football a long time ago, as I came very close to doing in Mancini's 2nd season.

[Ted Hughes]

I'm not saying it's a passing fad at all.

Having a solid defence has always historically been a way to win trophies & still is today.

What I'm saying is that there had never been a Barca or Real.Madrid type team in the Premier League, apart from Pellegrini's City. So stats comparing styles are meaningless.

If the stats are interpreted as they actually are, the best atfacking teams traditionally finish at the top. This is being ignored & discounted, but not by me.

[Im_Spartacus]

The stats absolutely prove nothing & suggest the complete opposite of the way they are being interpreted imo.

The idea that Mourinho's style is a new thing is complete nonsense. There have always been loads of people trying to do what he did. The reason he has been successful is because he has access to loads of really good players.

If you gave us Hazard & him Nasri, we would be Champions this season. His defensive mentality would mean nothing.

It's got nothing to do with Mourinho. With respect, if you try to look at a specific team, manager, or individual season, you will always find exceptions, but what you can't do is argue that the long term trend doesn't exist, because it clearly does, and the statistical approach taken is designed to identify causation, eg, it rules out the possibility of the relationship between goals and points being random or unduly affected by 1 or 2 outcomes.

It's got everything to do with a shift in the depth of quality in the premier league which has increased the importance of goals conceded over time. Some managers will take that on board and build their defence as a priority (Mancini), or deploy stifling tactics (Mourinho), others will just go for it (Pellegrini).

There's clearly not a right or wrong way to win a title, but over the last 10 years, conceding less goals has had a significantly bigger factor in the final points total of the champions than how many they scored - however you want to dress that up, that's a fact.

What managers choose to do with that fact is down to personal preference

[Im_Spartacus]

If the stats are interpreted as they actually are, the best atfacking teams traditionally finish at the top. This is being ignored & discounted, but not by me.

Regression analysis evaluates the correlation between a dependent variable (points), and an independent variable (goals scored or conceded).

There is no ignoring and discounting going on, goals conceded had a far higher correlation with points total, than goals scored. Putting a Barcelona type team in the premier league would make no automatic difference to the outcome, just because they were an attacking team, it doesn't automatically follow that goals for will become a more valuable commodity than goals against.

You're basing your views on opinions, which I respect (and to a large extent my gut feeling would be to agree with you before seeing the data), I'm basing my posts on facts though, that conceding less goals predicts champions better than goals scored.

[Wonderwall]

Nice thread, I like the debate.
Emphasis on Attack = Happy crowd and TV neutrals
Emphasis on Defence = Bored crowd and TV's turned off

In 1937/38 we scored the most goals in the league and got relegated, that's some stat, especially when we scored 3 more than the champions that year, Arsenal. However, we finished 20 places below them.

Another interesting point is, there are several posts identifying Mourinho as defensive and Pellegrini as attacking. Quick question....When playing away from home this season, which team do you think has scored the most goals, City or Chelsea?
Answer = Chelsea, they have scored 37 (averaging just over 2 goals an away game). We have scored 35 averaging just under 2 goals an away game.

I remember Vinny saying how it is more difficult to defend under Pellegrini due to the difference in philosophies between him and Mancini.
I think Kompany has been poor under Pellegrini, however, I dont think any centre back would excel when constantly exposed by the system. Its no coincidence that John Terry is all of a sudden being lauded as one of the best ever as soon as Mourinho comes back. He knows how to protect his Centre Backs.

[Ted Hughes]

If the stats are interpreted as they actually are, the best atfacking teams traditionally finish at the top. This is being ignored & discounted, but not by me.

Regression analysis evaluates the correlation between a dependent variable (points), and an independent variable (goals scored or conceded).

There is no ignoring and discounting going on, goals conceded had a far higher correlation with points total, than goals scored. Putting a Barcelona type team in the premier league would make no automatic difference to the outcome, just because they were an attacking team, it doesn't automatically follow that goals for will become a more valuable commodity than goals against.

You're basing your views on opinions, which I respect (and to a large extent my gut feeling would be to agree with you before seeing the data), I'm basing my posts on facts though, that conceding less goals predicts champions better than goals scored.

No you are giving an opinion & disguising it as fact, by interpreting statistics to suit your own argument rather than considering the reality of it.

Mourinho tends to concede less goals, so when HE wins the title, it is usually with the lowest goals against record, occasionally spectacularly so. When others win the title, your stats become complete & utter bollocks as there is no general rule at all. Sometimes the Champions concede fewest, sometimes the runners up, or the 3rd place team.

You are unconciously manipulating the stats, incorrectly, to suit your argument.

Most of the time, the side which wins the title are the top scorers & often not the tightest defence. This is a fact, not an opinion.

Please just discus it as an opinion. Your statistical analysis is wrong.

[Spurge]

Great thread Sparty, very interesting.

The millions which Sky have invested into the game has I believe enabled them to carry sway where modifying rules the with governing bodies in order to favour attacking teams, which has in turn meant that the art of defending has become increasingly more difficult. Strikers are given the benefit of any doubt (or are meant to be) on offside decisions, whilst making a tackle and winning the ball is no longer necessarily seen as a fair challenge. Gone are the days when a defender can take the ball cleanly and their momentum means they take the man is still seen a good challenge (sadly). More often than not a free kick is awarded to the attacking side in such circumstances.

This should encourage teams to go on the offensive in the modern game meaning in essence that goals for becomes more important. So will the stats in 10 years time still send out the same message as the last 10 which you've looked at? Or will the team who outscores the rest invariably prevail? Well if this season is likely to be a fair indication with Chelsea taking the title then once again quality defending will remain key.

Maybe with defending becoming more difficult the team(s) that masters this element of the game are likely to be those that reap the biggest rewards.

[Slim]

Excellent research, however as pointed out there is an old saying and I think we'd all agree that it's been the case in the past.

Last season surely has to be outside the SD.

[Im_Spartacus]

If the stats are interpreted as they actually are, the best atfacking teams traditionally finish at the top. This is being ignored & discounted, but not by me.

Regression analysis evaluates the correlation between a dependent variable (points), and an independent variable (goals scored or conceded).

There is no ignoring and discounting going on, goals conceded had a far higher correlation with points total, than goals scored. Putting a Barcelona type team in the premier league would make no automatic difference to the outcome, just because they were an attacking team, it doesn't automatically follow that goals for will become a more valuable commodity than goals against.

You're basing your views on opinions, which I respect (and to a large extent my gut feeling would be to agree with you before seeing the data), I'm basing my posts on facts though, that conceding less goals predicts champions better than goals scored.

No you are giving an opinion & disguising it as fact, by interpreting statistics to suit your own argument rather than considering the reality of it.

Mourinho tends to concede less goals, so when HE wins the title, it is usually with the lowest goals against record, occasionally spectacularly so. When others win the title, your stats become complete & utter bollocks as there is no general rule at all. Sometimes the Champions concede fewest, sometimes the runners up, or the 3rd place team.

You are unconciously manipulating the stats, incorrectly, to suit your argument.

Most of the time, the side which wins the title are the top scorers & often not the tightest defence. This is a fact, not an opinion.

Please just discus it as an opinion. Your statistical analysis is wrong.

Ted, you know there are some times when you are just wrong, and this is one - you're pissing in the wind with this one, however I welcome your opinion as an alternative side to the debate.

The entire premise of the thread is that facts don't support commonly held opinions and perceptions, and clearly you're in the group with the managers who chose not to listen to sports science and statistics because they 'trust their instinct' despite all evidence being that they could learn from the numbers. That's fine, but your opinion, however strongly held, doesn't make the facts any different.

Seeing as you seem to have a fixation on taking Mourinho out of the mix, I've taken out all 3 Mourinho titles, and these are the outcomes.

Correlation of points to goals conceded: 33%
Correlation of points to goals scored: 2%

So still, the number of goals conceded is more correlated with the points tally of the champions, and now the correlation between points and goals scored is virtually random. Let me explain why that's happened, mathematically for you, step by step.

The points totals in those years (excluding Mourinho) were 86,89,89,80,86,90,87,89 (A range of 9 points)
The goals for were 102,86,83,78,103,68,80,83. (A huge range of 35, 4 x the range of points)
The goals conceded were 37,43,29,37,32,24,27 (A moderate range of 19, 2x the range of points....43 is an obvious anomaly, without this the range would be 13)

If you set a big range of goals scored, against a small range of points, it's obvious that goals scored, which varied by upto 35, have little relevance to the number of points which varied by only 9. What's more consistent however, is the range of goals conceded.

Another way of proving this perhaps in terms most people will find more accessible than regression analysis (although far less scientifically), is to look at averages.
** The average number of goals scored in the first 5 years was 75, but since then its 89.
** The average number of points in the first 5 seasons was 90, but since then its 87
So the number of goals have gone up by an average of 14 a season, but the number of points has dropped by 3. Odd? No, because the avg number of goals conceded has also increased from 22 to 34, a fact which obviously had a bigger impact on points than the increase in goals.

So we have regression analysis supporting disproving your opinion that scoring the most goals = the most points, and now averages. Need I go on?

Whilst picking one example isn't good practice, this one is significant because it ilustrates the point perfectly. The highest number of points in the period exlcuding mourinho (90 in 2008/9) were achieved by the rags with the lowest number of goals (68). So how do you explain that if a major contributor to those points wasn't by conceding less (24)

Essentially, all the evidence here is telling us, is that it doesn't particularly matter whether you score 68 or 103 goals in any given year, what matters to be a champion, is that you keep your goals conceded low

[Im_Spartacus]

Excellent research, however as pointed out there is an old saying and I think we'd all agree that it's been the case in the past.

Last season surely has to be outside the SD.

If you have a nosey at the post above in response to Ted about the ranges of goals conceded and the ranges of goals scored, it appears that it doesn't particularly matter how many goals you score to win the title. Provided the goals conceded are within a fairly narrow range (which last season was, just), this is a better predictor of champions.

In the last 10 years, the only big anomaly is the rags title in 2013, where they conceded 43 goals achieving 89 points - an odd season characterised by the total lack of a title challenge, however, they still picked up an impressive points total, so we can hardly write that off as a fluke. The other minor anomaly is City in 2014 conceding 37 and achieving 86 points. United in 2011 also conceded 37 but only achieved 80 points which is consistent with 'more conceded = less points'

But what I'm not sure some people get, is that there will always be anomalies which aren't explained by a trend model. Anomalies don't disprove a model, they simply show the possibility that other outcomes can occur.

[bayblue]

This thread is bringing back nightmares about school !!
Thanks for your patience in explaining !!

[Ted Hughes]

Excellent research, however as pointed out there is an old saying and I think we'd all agree that it's been the case in the past.

Last season surely has to be outside the SD.

If you have a nosey at the post above in response to Ted about the ranges of goals conceded and the ranges of goals scored, it appears that it doesn't particularly matter how many goals you score to win the title. Provided the goals conceded are within a fairly narrow range (which last season was, just), this is a better predictor of champions.

In the last 10 years, the only big anomaly is the rags title in 2013, where they conceded 43 goals achieving 89 points - an odd season characterised by the total lack of a title challenge, however, they still picked up an impressive points total, so we can hardly write that off as a fluke. The other minor anomaly is City in 2014 conceding 37 and achieving 86 points. United in 2011 also conceded 37 but only achieved 80 points.

But what I'm not sure some people get, is that there will always be anomalies which aren't explained by a trend model. Anomalies don't disprove a model, they simply show the possibility that other outcomes can occur.

What you are doing is proving how dangerous this kind of information can be when used in a misleading fashion.

Most of the time, the side who wins the league is the one who scores the most goals. If not, on the few occasions this general rule is broken, it usually finishes 2nd. You are basically saying 'let's ignore that & concentrate on something else'.

If you sit back for a minute & think about it, you may just start to realise how ridiculous that is.

So when your team finishes the season relegated having scored 30 goaks but only conceded 15, you can go to the chairman & explain how clever your sports science department is, at analysing data.

OF COURSE it is important not to concede goals. But most of the time, the team which wins the title is the team which scores the most goals.

How does a team score goals ?

What was the original question ?

Perhaps 'I have a theory about the percentage of goals scored vs goals conceded' would have been a better title rather than asking people about the merits of attack v defence.

[Im_Spartacus]

Sorry Ted, but you're missing the point so spectacularly, I'm not wasting my time trying to explain it to you again.

As to your assertion that 'most of the time the team who scores the most wins', the percentage of times in the last 10 seasons that a title winner scores the most goals, is identical to the percentage of times a team concedes the least. (around 55%)

At least if you're going to accuse someone of bringing only 1 side of a debate to the table, don't do the fucking same yourself.

[Ted Hughes]

For any of these stats to have any meaning at all in relati9n to the original question, you would hage to analyse each season as a seperate entity, how the opposition lined up, whether goals were conceded due to a team over attackinv or indeed being unduly negative, etc etc.

It really is such utter nonsense to try & use this as proof of anything one way or the other.

There is no proven answer to the o.p. it is purely down to opinion.

That shouod he what we are discussing, not this nonsense.

[Ted Hughes]

Sorry Ted, but you're missing the point so spectacularly, I'm not wasting my time trying to explain it to you again.

As to your assertion that 'most of the time the team who scores the most wins', the percentage of times in the last 10 seasons that a title winner scores the most goals, is identical to the percentage of times a team concedes the least. (around 55%)

At least if you're going to accuse someone of bringing only 1 side of a debate to the table, don't do the fucking same yourself.

You can't win a fucking game of football, without scoring ffs ! It's that simple.

The team who wins tends to be the one who scores the most. So therefore any attempt to PROVE this wrong is bollocks.

There is no proof, your stats are nonsense. Just stick to opinion instead of thinking you have found the holy fucking grail, that's what I'm pointing out. I'm not claiming that it proves attacking football is the answer, I'm claiming it proves there is insufficient data.

It is all down to OPINION not a proven science.

[Im_Spartacus]

Sorry Ted, but you're missing the point so spectacularly, I'm not wasting my time trying to explain it to you again.

As to your assertion that 'most of the time the team who scores the most wins', the percentage of times in the last 10 seasons that a title winner scores the most goals, is identical to the percentage of times a team concedes the least. (around 55%)

At least if you're going to accuse someone of bringing only 1 side of a debate to the table, don't do the fucking same yourself.

You can't win a fucking game of football, without scoring ffs ! It's that simple.

The team who wins tends to be the one who scores the most. So therefore any attempt to PROVE this wrong is bollocks.

There is no proof, your stats are nonsense. Just stick to opinion instead of thinking you have found the holy fucking grail, that's what I'm pointing out. I'm not claiming that it proves attacking football is the answer, I'm claiming it proves there is insufficient data.

It is all down to OPINION not a proven science.

But you don't have to score 10 to fucking win you thick cunt, 10 goals in a game gets you no more points than 1 goal can. It might increase the probability of winning a game, but we're not looking at games, we're looking at 10 fucking seasons worth of data.

No holy grail, nothing complex, just the relationship between goals and points for the champions.

[Slim]

Sorry Ted, but you're missing the point so spectacularly, I'm not wasting my time trying to explain it to you again.

As to your assertion that 'most of the time the team who scores the most wins', the percentage of times in the last 10 seasons that a title winner scores the most goals, is identical to the percentage of times a team concedes the least. (around 55%)

At least if you're going to accuse someone of bringing only 1 side of a debate to the table, don't do the fucking same yourself.

You can't win a fucking game of football, without scoring ffs ! It's that simple.

The team who wins tends to be the one who scores the most. So therefore any attempt to PROVE this wrong is bollocks.

There is no proof, your stats are nonsense. Just stick to opinion instead of thinking you have found the holy fucking grail, that's what I'm pointing out. I'm not claiming that it proves attacking football is the answer, I'm claiming it proves there is insufficient data.

It is all down to OPINION not a proven science.

Did Ted just say statistical analysis isn't a proven science?

Well end of, lock the thread, we have a winner.