The Numbers Game

[PowerSpurs]

I bought this book earlier in the week and have had a good look at it.

Its premise is that what most people 'know' about football is wrong if you actually analyse the data. Some of the statistics are a bit of a matter of opinion as the analysis is quite difficult (although this is being improved by the things Opta and others are doing) but some things seem to be clear beyond reasonable doubt:

1) Corners are worse than useless as goalscoring opportunities. It seems that statistically you are slightly less likely to score from a corner than having possession in the opposition half in open play. When I thought this through it's not that surprising as the defence is organised at that point, but it actually makes a short corner the best tactic.
2) Based on player ratings it is much better to improve your worst player than your best one if you want success. Again, perhaps not surprising in a way as it is a team game and the weakest link matters
3) Players' pass success rate doesn't depend on how high their rating is at all, but only on how difficult the pass is. That was a surprise to me but again, after some thought I'm not sure why. Taking away the geniuses like Messi, the difference between players technical abilities at the top level is likely to be very small. The better players do make more of the difficult passes though.
4) About three quarters of the results in the PL can be explained by randomness. Keeping a clean sheet is slightly better than scoring 2 goals in the number of points obtained
5) Possession is very important in winning a match.

One other point with less evidence from the data - teams who pass to the wings do less well on average than those who keep the ball central!

I suspect AVB knows this stuff and it gives some insight as to why he uses these tactics.

And just to show that you can prove anything with statistics - over the last two seasons analysed (up to 2012 I think) the most successful striker in the PL measured by how many points were won by his goals (so for example the 3rd goal in a 3-1 doesn't count) was Darren Bent!

 

[The RDBD]

Statistics is basically about discovering patterns in numerical data.
And then trying to confirm/repute hypotheses about those patterns using the data.

As you have said, often you can gather statistics that are totally pointless (no value
can be derived from them) . I know this too well from my day job.

Other times you have obvious patterns, yet cannot construct a hypothesis to explain
the patterns (applying them becomes an act of faith) .

 

[Boss]

Power, Nice try at using some random book to back up Avb and his tactics? Did you get it on sale? Lol

That point about teams who pass it centrally getting more success is flawed, as does it mention the stats of teams who pass it centrally with the wrong players? As I suspect the teams which get success from more central play, have the squad to suit that system.

Man Utd, Ferguson arguably the best British manager used wingers and stretched defences, but also played through the centre, does your book talk about mixing it up?

Avb is using a system with the wrong type of squad, that's why we only perform in short periods and AVB will fail to have us finishing in the top four.

 

[deedub93]

I think Graham Taylor once said that statistics showed at least 50% of goals came from route one long balls. I guess that's why he had all of the success that he did.

 

[PowerSpurs]

Power, Nice try at using some random book to back up Avb and his tactics? Did you get it on sale? Lol

That point about teams who pass it centrally getting more success is flawed, as does it mention the stats of teams who pass it centrally with the wrong players? As I suspect the teams which get success from more central play, have the squad to suit that system.

Man Utd, Ferguson arguably the best British manager used wingers and stretched defences, but also played through the centre, does your book talk about mixing it up?

Avb is using a system with the wrong type of squad, that's why we only perform in short periods and AVB will fail to have us finishing in the top four.

It's not a 'random book' as it is the only one I can find that actually tries to use data to see if common practices are sensible. I use statistics a lot in my job and I know their limitations but if there is something interesting in the data it is daft to ignore it. I got interested in this by watching tennis where the 'expert' commentators say things that are clearly untrue. eg serving first in a tennis set is an advantage: no it isn't and the data proves it. The one I really liked was a match I watched where a player was getting 60% of his first serves in and winning 90% of those points, whereas his second serves were 100% in (ie no double faults) but he was only winning 30%of the points. The commentators insisted that he needed to get more first serves in as his second serve was so weak. What he actually needed to do was to go with his first serve action all the time as while that would have meant a sixth of his serves were double faults he would still have won more points.

I really don't understand your point about the 'wrong players'. Which of the following statements is more likely to be true: 1) The Spanish team's success is down to their being exceptional passers of the ball so they are very suited to the game they play. 2) The Spanish team's success is down to their practicing passing and using the same system all the time in training and matches so that practice makes perfect. Every player in the PL is a well above average passer of the ball, otherwise they wouldn't be at that level. The difference between players' abilities is actually quite subtle except for the ones right at the end of the distribution like Messi and Ronaldo. That's why its very risky to spend £30m on anyone and why buying 7 players for £100m is less risky than buying three.

 

[Spurf]

PS I am not a fan of stats at the best of times but combine that with Darren Bent.

It's almost as if you did not want your argument to work.

In the last 3 games there has been some improvement from AVB as far as I am concerned but I would like more attention paid to some 'obvious' truths in this 'simple' game.

For example:

Better players will do similar things to lesser players but they will have the ability to do them faster, so as we have spent 100 million on some of these better players let's up our speed and beat the likes of West Ham.

Having discovered the best position for a player it might be wise to play him there and not play him in a postion that forces some kind of enforcment to a system.

It does not matter how bloody clever you are if you ignore the basics you will be found out in the end.

 

[The RDBD]

"It's almost as if you did not want your argument to work."

Thought that was provided as more an example of "how to lie with statistics" .

 

[Spurf]

"It's almost as if you did not want your argument to work."

Thought that was provided as more an example of "how to lie with statistics" .

I would be more interested in how to tell the truth with stats. if that were possible, but of course it's in the nature of stats that they are used to support whatever your paticular angle is.

 

[The RDBD]

"I would be more interested in how to tell the truth with stats."

You don't "tell the truth" so much as try to give reasons as why the patterns exist.
And then use the same +/- different data to support/refute your reasons.

Sometimes you accept that no good reasons can be given, but you don't ignore
the patterns regardless.

 

[Boss]

It's not a 'random book' as it is the only one I can find that actually tries to use data to see if common practices are sensible. I use statistics a lot in my job and I know their limitations but if there is something interesting in the data it is daft to ignore it. I got interested in this by watching tennis where the 'expert' commentators say things that are clearly untrue. eg serving first in a tennis set is an advantage: no it isn't and the data proves it. The one I really liked was a match I watched where a player was getting 60% of his first serves in and winning 90% of those points, whereas his second serves were 100% in (ie no double faults) but he was only winning 30%of the points. The commentators insisted that he needed to get more first serves in as his second serve was so weak. What he actually needed to do was to go with his first serve action all the time as while that would have meant a quarter of his serves were double faults he would still have won more points.

I really don't understand your point about the 'wrong players'. Which of the following statements is more likely to be true: 1) The Spanish team's success is down to their being exceptional passers of the ball so they are very suited to the game they play. 2) The Spanish team's success is down to their practicing passing and using the same system all the time in training and matches so that practice makes perfect. Every player in the PL is a well above average passer of the ball, otherwise they wouldn't be at that level. The difference between players' abilities is actually quite subtle except for the ones right at the end of the distribution like Messi and Ronaldo. That's why its very risky to spend £30m on anyone and why buying 7 players for £100m is less risky than buying three.

I was jesting about the random book, I'm sure it's a good read.

The point about wrong players or players naturally suited to a system is that on paper Avb and his system makes sense. If I sat down and had a conversation with Avb and asked him to explain his system, it would make complete sense in terms of producing results.

Then there is a but, as any system is only as good as the players to implement it. AVB is using a system which relies on having players with short passing ability, ball retention, players ability to release the ball and create chances in tight areas . Why do you think we have such high possession and low goals to chances scored as the majority of our chances stem from difficult positions, created from poor passers, or we don't have the players to make use of our possession unless we're in safe areas.

When we pass quick, use the wings and mix it up, we look much better. When we play slow and possession based we look short on ideas.

I understand you believe by practicing the squad will adapt but I disagree and can't see us ever completely being comfortable with AVB's system.

 

[O.Spurcat]

I think Graham Taylor once said that statistics showed at least 50% of goals came from route one long balls. I guess that's why he had all of the success that he did.

Taylor did indeed say that and it was based on stats showing the more frequently a team got the ball into the opposition penalty area, the more goal scoring opportunities were made. So, Taylor's idea was, why faff about playing attractive football and taking 4 or 5 passes to get into this "area of maximum opportunity" as he called it, when one simple hoofball will get it there quicker. It was pointed out to him that playing beautiful football and taking 4 or 5 passes to create a chance, never did Brazil any harm, but to no avail and to the detriment of the English game, he persevered with his long ball style.

 

[notsosmartspur]

Have you tried reading it the right way up?

 

[remembercolinlee]

Stats can and are being manipulated.
Look at assists...
a 2 yard pass to a player who smacks it in from 20 yards is an assist...
A 20 yard mazey run beating 3 players ending with a pull back only for the striker to do an impersonation of jason dozzel counts for nothing...
One was a great piece of play which is not counted on stats and the other played no part in the end product but is counted on the stats

 

[notsosmartspur]

Why do you think we have such high possession and low goals to chances scored as the majority of our chances stem from difficult positions, created from poor passers, or we don't have the players to make use of our possession unless we're in safe areas.

We've been given a new term, 'passive possession', heard it over the weekend, its a phrase commentators can broadcast rather than the more common 'do **** all with it'.

 

[The RDBD]

"Taylor did indeed say that and it was based on stats showing the more frequently a team got the ball into the opposition penalty area,
the more goal scoring opportunities were made. So, Taylor's idea was, why faff about playing attractive football and taking 4 or 5 passes
to get into this "area of maximum opportunity" as he called it, when one simple hoofball will get it there quicker."

Classic fallacious reasoning using the raw data. ***
Basic statistics would go thus ...

For every 100 goals scored from inside the box, you would need to get the distribution of how many passes
it took from a reference point (say the halfway line) to get there, So hoof would be passes = 1.

If the distribution shows a significant number of goals come from passes = 1, then the claim has merit.

The next statistics hurdle would be to consider population size.
If the number of goals analysed goes from 100 to 1000 goals etc, does the distribution still hold.

And so on.


*** The fallacy being that far more goals are scored from within the box than outside.
With the speed of delivery into the box being another variable that would have to be assessed.

 

[The RDBD]

Perhaps the phrase should be :

Possession is nine tenths of the law.
Possession in the final third is nine tenths of the score.

 

[PowerSpurs]

PS I am not a fan of stats at the best of times but combine that with Darren Bent.

It's almost as if you did not want your argument to work.

In the last 3 games there has been some improvement from AVB as far as I am concerned but I would like more attention paid to some 'obvious' truths in this 'simple' game.

For example:

Better players will do similar things to lesser players but they will have the ability to do them faster, so as we have spent 100 million on some of these better players let's up our speed and beat the likes of West Ham.

Having discovered the best position for a player it might be wise to play him there and not play him in a postion that forces some kind of enforcment to a system.

It does not matter how bloody clever you are if you ignore the basics you will be found out in the end.

You've made all sorts of deductions from the data in your head (you know the best position for the players and that we are deliberately doing a slow build up). So you are using statistics yourself. A set of statistics collated by an independent observer is almost bound to lead to more valuable conclusions.

 

[Boss]

We've been given a new term, 'passive possession', heard it over the weekend, its a phrase commentators can broadcast rather than the more common 'do **** all with it'.

Lol that's a good one. Although Talksport will still use the' do **** all with it '. Allows their listeners to understand!

 

[Spurf]

You've made all sorts of deductions from the data in your head (you know the best position for the players and that we are deliberately doing a slow build up). So you are using statistics yourself. A set of statistics collated by an independent observer is almost bound to lead to more valuable conclusions.

You call it 'data' I call it observation and experience. Statistics are the collection of numerical data, as you know, this I do not use in my observations. You can play with words but the simple truths remain; in football slow is of use to delay or survive in the heat of a world cup but one of the major reasons England rarely perform well in the WC is that the heat often prevents use of our high speed game. AVB is used to heat, he comes from Portugal but he's in England now he needs to speed up IMO.

 

[The RDBD]

"You've made all sorts of deductions from the data in your head (you know the best position for the players and that
we are deliberately doing a slow build up). So you are using statistics yourself."

I suspect Spurf was making a qualitative assessment, not a quantatitive one.


"A set of statistics collated by an independent observer is almost bound to lead to more valuable conclusions."

And it is often very nice when the (independent) statistics quantatively corroborate your
qualitative thoughts.