Tim Sherwood…gone \o/

[Trotter]

Re: Tim Sherwood - Head Coach

Like I asked in the last post, can you please post the details/sources of these data?

I'd also be interested in why you think PDO regressing to mean doesn't happen.

Source of data is from Opta. Basically just taken the Total Shots, Shots on Target, Shots Conceded and SOTC data and done the calculation.
Happy to send you them, PM me your e-mail address

Why I think regression to mean doesn't happen in football is because data shows that the teams that finish top of the League always have the best PDO. and the numbers for the top teams are pretty static between the seasons. Individual teams may go up and down between seasons, based on their own performances, but in general the PDO for the team winning the league is very similar between seasons, similarly with the team finishing runner-up and all the way down the league. This is because of the difference in quality between the teams. What you are saying is that Emirates Marketing Project would have the same PDO as say Crystal Palace if not for luck. That just is not the case. As said above, PDO really only works in equal strength leagues (e.g. where there is a salary cap as in NHL where the author takes his inspiration from, would be a good example)

 

[Rossi22]

Re: Tim Sherwood - Head Coach

Is this from that James guy? I've read his NHL stuff before.

 

[scaramanga]

Re: Tim Sherwood - Head Coach

Source of data is from Opta. Basically just taken the Total Shots, Shots on Target, Shots Conceded and SOTC data and done the calculation.
Happy to send you them, PM me your e-mail address

Why I think regression to mean doesn't happen in football is because data shows that the teams that finish top of the League always have the best PDO. and the numbers for the top teams are pretty static between the seasons. Individual teams may go up and down between seasons, based on their own performances, but in general the PDO for the team winning the league is very similar between seasons, similarly with the team finishing runner-up and all the way down the league. This is because of the difference in quality between the teams. What you are saying is that Emirates Marketing Project would have the same PDO as say Crystal Palace if not for luck. That just is not the case. As said above, PDO really only works in equal strength leagues (e.g. where there is a salary cap as in NHL where the author takes his inspiration from, would be a good example)

Did you see the data on the page I linked? PDO does regress to mean in the PL at least (if not all football).

As for PDO varying between teams, the variation in quality of teams is taken care of in the SOTR. The PDO shows how the team's results over or under perform based on that difference between the teams. You can't use PDO in this case as a measure of quality because you're counting that ability twice if you do - once in the SOTR and once in the PDO.

Oh, and I think it's far more interesting if we can all discuss the data on here - otherwise we just end up sending it on to anyone who wants to contribute.

 

[Bullet]

Re: Tim Sherwood - Head Coach

Interesting stuff this PDO debate, I hadn't heard of PDO until today

On the StatsBomb site he defines abbrevations as follows:

ShAcc, Sh% – Shooting Accuracy
Percentage of total shots placed on target.

PDO – I’m just going to copy the definition from James Grayson.
“PDO is the sum of a teams shooting percentage (shots/shots on target) and its save percentage (saves/shots on target against). It treats each shot as having an equal chance of being scored – regardless of location, the shooter, or the identity or position of the ‘keeper and any defenders. Despite this obvious shortcoming it regresses heavily towards the mean – meaning that it has a large luck component. In fact, over the course of a Premiership season, the distance a teams PDO is from 1000 is ~60% luck.”

And the link to James Grayson says http://jameswgrayson.wordpress.com/2011/05/14/sh-sv-pdo-part-n/

The aim is to show graphically how regression to the mean occurs. The method I’ve used is as follows. First I’ve determined each teams sh% for games 1-19 of the season. I’ve then taken various samples such as ‘top 25%’ (the 50 teams that had the highest sh% out of the sample of 200) and determined their sh% for each successive 19 game sample (ie 2-20, 3-21.. ..20-38). I then repeated the whole process for sv% and PDO ... As you can see each and every line regresses towards the mean to some extent.




OK, so the 'PDO' is derived from Sh% and Sv% so I am looking at Sh% because that is essentially the same as looking at PDO i.e. if we can agree Sh% is rubbish we can agree PDO is rubbish but if we think it makes sense we can discuss Sv% as well... it is confusing to talk about the abstract PDO so let's talk about Sh%

It immediately strikes me that Sh% (the percentage of total shots on target) is a bit of a strange ratio because you could have 2 shots in a match and get them both on target and get 100%... or you could have 20 shots and get 10 on target and get 50%... but you would obviously prefer to have 20 shots rather than 2. As such it seems a fundamentally crude and flawed stat.

I mean... what we are really talking about here?

I don't honestly know whether we had more chances with one manager or the other. Let me have a look:

http://www.squawka.com/teams/totten...season-2013/2014#64#all-matches#1-16#by-match
AVB: Matches 1-16 Premier League Shot Accuracy 35%, chances created 210 and 210/16=13 per game

http://www.squawka.com/teams/totten...eason-2013/2014#64#all-matches#17-22#by-match
Sherwood: Matches 17-22 Premier League Shot Accuracy 34%, chances created 59 and 59/6=10 per game

So what does it mean? The same Sh% but COMPLETELY different ways of playing the game. AVB made more chances per game according to these stats, but almost nobody would agree that meant a great deal, as it is all about the quality of the chances/attempts at goal.


I suppose if the best teams have the best Sh% then there must be some merit in the stat... but it does seem very flawed at first glance.


[I'm not even mentioning that you could have 5 soft shots from distance on target which are easily saved, or you could have 5 good chances from close in that go over the bar... the former indicates you can't get into dangerous positions, the latter indicates you are getting into the right places but signed DJ Campbell rather than Gary Lineker]

 

[Modric THFC]

Re: Tim Sherwood - Head Coach

This is what happens when Tottenham don't play for a week. Somw one invents a new stat in an attempt to show that one of the richest managers in football after completing two full seasons is 'unlucky'.

 

[Cochise]

Re: Tim Sherwood - Head Coach

I'm not sure how best to word it, but simply imo we have been creating better chances under TS. It's common sense to say that the closer a player to the goal, the more likely they are to score, which is pretty much sums us up right now. In the six league games since TS took over we have scored 14 goals. 10 have been scored from within the 18 yrd box, 2 were own goals and the last 2 were from outside the box. Not only that but those goals from inside the box came from situations where play was stretched, where we had isolated the oppositions defence into situations of 3v3/4v4 where the right pass is easier to pick out. Under AVB we were regularly getting to the box and finding 8-9 men between us and the goal. The passes were harder to make and the shots taken had less chance of going in due to the distance and bodies in the way.

I think I understand where you are going with the "PDO regresses to the mean" tack, as over the course of a season every team will go through periods where their star player may not perfom (imagine Suarez going 5 games without a goal). Then you can have games where the keeper can turn in a world class performance (Krul and Marshall spring to mind) and games where you come up against the most well drilled defences in the league who won't concede. Right now Sherwood has taken us through six games where there have been none of those problems and our star man (Ade) has been on form. It is unsustainable because there is less chance of us creating these high % conversion rate chances against the better teams.

 

[Rossi22]

Re: Tim Sherwood - Head Coach

Chances and attempts/shots shouldn't be used interchangeably. A 30 yard shot isn't the same as Soldado putting one on a plate for Lennon from 6 yards out, or even from the same game Lennon putting one on a plate for Soldado 4 yards out. The PDO stuff doesn't want to recognise this, hence why it is flawed. We had lots of shots under AVB, now we're making chances instead.

 

[Bullet]

Re: Tim Sherwood - Head Coach

I suppose if the best teams have the best Sh% then there must be some merit in the stat... but it does seem very flawed at first glance.

I've looked at Sh% for each Prem team on Squawka. It does tend to be higher for the teams at the top and lower as you go down, but only slightly and also with outliers like Stoke and Fulham.

It seems like a byproduct rather than a tight correlation... as I said having 2 out of 2 shots on target is much worse than having 10 out of 20 shots on target, so I don't see how a ratio like that can mean much.

He selects the first 19 games of a season as his start point for the high/low scoring Sh% teams and then goes on to show their Sh% reverts to mean... maybe it is just his way of selecting the high/low Sh% scores which is flawed? Why should the first 19 games show something special? If he is just selecting a random grouping then of course over time this reverts back to mean... it would make more sense to choose the high/low Sh% scores based on final finishing position in the league or something tangible?

 

[thfcsteff]

Re: Tim Sherwood - Head Coach

You'll have to forgive me then for believing in fact posted by a neutral over your opinion.

If your line of reasoning is going to continue along the lines of "This is my opinion, your facts are clearly wrong" then I'm not sure what else I can say. It's like telling a religious person about dinosaurs and being told the devil put them there.

…He didn't?

 

[scaramanga]

Re: Tim Sherwood - Head Coach

Chances and attempts/shots shouldn't be used interchangeably. A 30 yard shot isn't the same as Soldado putting one on a plate for Lennon from 6 yards out, or even from the same game Lennon putting one on a plate for Soldado 4 yards out. The PDO stuff doesn't want to recognise this, hence why it is flawed. We had lots of shots under AVB, now we're making chances instead.

Yet it still regressed to mean. So it is (at least 60%) measuring 'luck'.

It doesn't matter whether you agree or disagree with how accurately it represent chances vs shots, it has to have some luck proportion as for all teams it regresses to mean. This means that you can take the 40% of it that you don't like out and it still shows that AVB was a little 'unlucky' when in charge and Timmeh is very, very 'lucky'

 

[SpurMeUp]

Re: Tim Sherwood - Head Coach

Mr Scara if your head is in the oven and your feet are in the freezer is your mean temperate comfortable?

I admire the loyalty of your arguments, and this PDO bolocks is interesting even if it is ultimately limited (imo), but you have pulled out all the stops here - focusing on Sherwood's intelligence, his class, to a stat that can apparently show he has simply been lucky and AVB unlucky. I take my hat off to you. That's dedication and makes for some interesting discussions.

The notion that we can measure luck and put a number to it is fascinating, but flawed certainly for football.

 

[Trotter]

Re: Tim Sherwood - Head Coach

Would it be remiss of me to counter the fact that all team regress with the actual facts. That is for the first 6 games of this season we were basically at mean (ranging between 100.8 and 99.8) and from then on moved away from the mean almost every week, ending at 93.6 (based on Sh/Sv% (or 86.7 based on SOT%/SOTC%), with our PDO going away from the mean after 8 of the final 10 games under AVB.

 

[Mulletperm]

Re: Tim Sherwood - Head Coach

**** me. Seems like you need a Masters in statistics to follow football these days.

 

[glorygloryeze]

Re: Tim Sherwood - Head Coach

**** me. Seems like you need a Masters in statistics to follow football these days.

You certainly do in this tread; I work with data and some stats, but even I'm lost

 

[Trotter]

Re: Tim Sherwood - Head Coach

It is basically just someone trying to use their view of some flawed statistics and assumptions to show the fact that even though we were playing poor under AVB, and our shooting and saving %'s were rubbish under AVB, that is was all just down to pure luck, and he would have turned it all around, and his view that AVB was our saviour therefore continues and Sherwood is a lucky chancer, rather than have to admit that AVB just had the team and individuals under-performing, getting worse each week, and currently Sherwood has them over-achieving and creating better chances, and improved results. The argument is just not credible.


And by the way I do have a masters degree in statistics, so I am certainly in my comfort zone

 

[Jurgen the German]

Re: Tim Sherwood - Head Coach

But we clearly weren't. I can name the games on one hand where that was the case, the rest of the time it was dour football where we were bailed(not Baled, that was the previous season) out thanks to fortunate penalties, own goals, centre backs hitting 30 yard half volleys into the bottom corner, stoppage time winners. What we were doing under AVB was completely unsustainable, a team with a 15:21 goals record should've been well in the bottom half, it was GOOD luck that we weren't, not bad. AVB's gameplan was to squeeze the life out of the game, if he could win it 50.01 to 49.99(I'm not talking about possession here) then he was thrilled, it was awful, the cracks were getting bigger and bigger and the walls well and truly came crashing down at home to Liverpool. I'm glad that chancer is gone.

Agree with every word of this.

 

[Bullet]

Re: Tim Sherwood - Head Coach

And by the way I do have a masters degree in statistics, so I am certainly in my comfort zone

Nice one, that will put the cat amongst the pigeons! Please explain further, e.g. My point above about the way he selected the high/low performers over the first 19 games... rather than another method of selecting them... did that make any sense to you?

 

[Jurgen the German]

Re: Tim Sherwood - Head Coach

Like Spurmeup mentioned earlier, I like debate as much as the next guy and I could see why some people wanted AVB to finish out the season but it's getting comical now to see the depths Scara is having to plunder to make AVB look like a genius and Sherwood look like a poor man's Arry. We weren't dominating games under AVB, having 70% of the ball but only creating one or two meaningul chances a game if we were lucky and resorting to pot shots from 25 yards out is not dominating a football match.

Under Sherwood we are getting more than 2 players into the opposing box and the team generally look far more at ease in this system. As Rossi said, it's not lucky that Sherwood decided to bring Ade back into the fold, decided that Eriksen is the main man rather than playing Holtby who isn't as good and it's not lucky that he gave Bentaleb a chance as he knows the kid can play.

I think it's literally Scara and GB on an island :lol:

 

[africanspurs]

Re: Tim Sherwood - Head Coach

You certainly do in this tread; I work with data and some stats, but even I'm lost

i've got a masters in this and no matter what the data shows with that guys chosen methodology i dont care.....i prefer what i am seeing now and it would take long to get something together to prove we are doing better now not down to the luck of the green but through a tactical shift

 

[braineclipse]

Re: Tim Sherwood - Head Coach

It is basically just someone trying to use their view of some flawed statistics and assumptions to show the fact that even though we were playing poor under AVB, and our shooting and saving %'s were rubbish under AVB, that is was all just down to pure luck, and he would have turned it all around, and his view that AVB was our saviour therefore continues and Sherwood is a lucky chancer, rather than have to admit that AVB just had the team and individuals under-performing, getting worse each week, and currently Sherwood has them over-achieving and creating better chances, and improved results. The argument is just not credible.


And by the way I do have a masters degree in statistics, so I am certainly in my comfort zone

So, the opinions of the other side according to you:

-The PDO numbers were all down to pure luck under AVB.
-AVB would have turned it around
-AVB was our saviour
-Sherwood is a lucky chancer

Appreciate your input and great to get the opinion of someone with a background to understand something, but do you have to go so over the top when characterizing those who disagree with you? Apart from the possible exception of perhaps one poster (that isn't Scara) this doesn't really seem like a fair characterization of anyone on this forum. I would have thought that with a masters in statistics anything other than liberal arts you would realize that straw manning your opponent's argument like this doesn't strengthen your case, it rather makes you seem like you haven't understood the argument.