Jump to content
Sign in to follow this  
CambridgeCanary

Late and/or Negative Substitutions

Recommended Posts

There has a lot of discussion of this issue and I saw this piece in the Fink Tank, a statistics based analysis which Daniel Finkelstein writes in the Times every Saturday.   I copy here just because as so often, statistical analysis shows that what we think we kn ow is not as clear cut as it seems.

 

 

Daniel FinkelsteinFink Tank

Published at 12:01AM, November 17 2012

The ability to make a good substitution is regarded as one of the tests of a good manager. Which makes me suspicious. It prompts me to use my Fink Tank rule of thumb — nothing makes any difference.

Against Shakhtar Donetsk last week, Chelsea replaced Fernando Torres with Daniel Sturridge as stoppage time began. How can that change things? Then Chelsea scored a winner with almost the last touch of the ball. Randomness, or an indirect but nevertheless related event?

While conducting another piece of work some time ago, the Fink Tank came down on the side of randomness. Replacing one player with another in the same position does not alter the game in any significant way. This was not because of the timing of the substitution, it was down to its nature.

Now Dr Henry Stott, Dr Mark Latham and Gabriella Lebrect have been probing deeper. What about substitutions that replace one player with another in a different position?

Well, the first point is that the greatest number of substitutions are of the tactically pointless type. We looked at 4,384 substitutions between the start of the 2010-11 season and last weekend; 2,333 involved replacing a player with someone else in the same position. Only 284 involved an all-out change — a defender for a forward.

There were 756 mildly defensive changes (defender on for midfielder or midfielder for attacker). And there were 1,011 mildly aggressive substitutions — attacker on for midfielder or midfielder for defender.

Most of this takes place late in the match. In other words, managers are pretty conservative. But is the less conservative approach worth it? Does it make a difference? Yes, is the clear Fink Tank answer.

The most obvious finding — and one that the data allows one to make with confidence — is that defensive substitutions reduce the number of goals you concede. And the more defensive the substitution, the greater the reduction.

In other words, fans who worry about their team trying to defend for the last few minutes may be worrying about the wrong thing. They should be more worried about the manager keeping a winning team on the field and not changing because the present players are doing well.

The evidence in favour of attacking substitutions is less clear. It is not unequivocally the case that aggressive changes produce more goals. This needs further research. But in the meantime, the time to shout at the manager to make changes is when you are winning, not when you are behind.

 

 

 

 

Share this post


Link to post
Share on other sites
Statistics typically mean nothing. The simple reality is that Hughton hasn''t made many attacking substitutions this season because he hasn''t had confidence in the players on the bench to make a difference.

Morison, Jackson, Howson, Bennett. The four attacking players that he has had on the bench. How many times have Morison, Jackson, or Howson, made the slightest bit of difference to the course of a game this season? None?

We''ve invariably looked far less dangerous when withdrawing Holt for Morison or Hoolahan for Howson.

Bennett is a bit different, because at least twice this season he actually has come on and made a difference, and fans argue that we should see more of him (I agree).

He simply now has more options on the bench. Kamara who can play either wing or up front, probably even the Hoolahan second-striker role, and Becchio instead of Morison, and I expect that we will now see Hughton making subs earlier if we look completely impotent.

Share this post


Link to post
Share on other sites

Oh of course!  The old statistics mean nothing chant.  What rubbish.  Statiistics do not answer every question but their use can help understand issues and to predict likelihoods rather than rely on randon guesswork (like Mr Lawrenson).

 

My challenge is simple.  If you want to argue that statistics mean nothing then argue that Lambert was a better manager than Roeder without using statistics, for those who will say that anything can be proved by statistics then use statistics to show Roeder was a better manager than Lambert.

 

If you have a big enough sample then you can understand trends and you do gain understanding of facts over myth.

 

What the chart shows is that Lambert with his early substitutions is more unusual than Hughton with his late ones.  Perhaps we were spoilt for three years,

Share this post


Link to post
Share on other sites
This is the conclusion of the study:

1) Defensive substitutions result in less goals being conceded.

2) Like for like substitutions hardly ever effect a game.

3) They haven''t done enough research to determine whether attacking substitutions result in more goals.

Wow, what a brilliant piece of completely obvious research conducted by a company who fund their research selling Football Betting systems.

I now feel confident enough to continue to state the outright obvious and completely sold on the idea of spending money on a football betting system which will make me a millionaire.

Yay for non-publicly funded and completely unregulated or accessible statistical studies, what would the world do without them?

Share this post


Link to post
Share on other sites
The only trouble with those statistics are that they aren''t filtered on tactical substitutions, the majority in the 80th minute are to freshen up legs i would imagine.

And while Statistics can gauge trends etc. they''re of no relevance if your defence is getting ripped to shreds in the 15th minute, or likewise the opposition are clearly there for the taking, and they have weaknesses that could be exploited.

Share this post


Link to post
Share on other sites
No, it is research undertaken by a private business named Decision Technology who have contracts with fabulously responsible businesses like Tesco and e-On, who are paid by Castrol and the British Premier League, whilst also being beneficiaries of money from the European Union.

The also do election forecasting and financial forecasting software. They are effectively a professional firm of spin doctors. I admire the naivety that you show, but you do realise that with a large enough set of numbers you can make statistics say whatever they want them to? With that exact same set of numbers a skilled set of statiscians would be able to manipulate data to present the completely opposite conclusion.

That is precisely why people like e-On, Tesco, and political parties pay them money to "consult" them.

Alistair Campbell''s wet dream.

Share this post


Link to post
Share on other sites

"with a large enough set of numbers you can make statistics say whatever they want them to?"
Actually, no you can''t.  There is no size of data set for where the sun rises that will allow you to say it is in the west.
And you''d be better off with a really small set of numbers if you want them to say what you want them to.

Share this post


Link to post
Share on other sites
No but there are statistics which show that dependent on the season the sun may rise north or south of east, the same statistics will show you that the sun only rises due east two days of the year and sets due west two days of the year.

Of course, if you data only includes two samples from the precise days of the spring and autumn equinoxes, the conclusion could be that the sun always rises in the east and sets in the west.

And if you are a muslim you will believe that the sun will rise in the west before the day of judgement. No statistics or scientific data can 100% disprove their belief.

Statistics show that the sun has never risen in the west since records began.

Statistics also show that 23% of the people on this planet also believe that the sun will rise in the west one day.

Share this post


Link to post
Share on other sites
[quote user="singupcarrowroad"]Statistics typically mean nothing. The simple reality is that Hughton hasn''t made many attacking substitutions this season because he hasn''t had confidence in the players on the bench to make a difference.

Morison, Jackson, Howson, Bennett. The four attacking players that he has had on the bench. How many times have Morison, Jackson, or Howson, made the slightest bit of difference to the course of a game this season? None?

We''ve invariably looked far less dangerous when withdrawing Holt for Morison or Hoolahan for Howson.

Bennett is a bit different, because at least twice this season he actually has come on and made a difference, and fans argue that we should see more of him (I agree).

He simply now has more options on the bench. Kamara who can play either wing or up front, probably even the Hoolahan second-striker role, and Becchio instead of Morison, and I expect that we will now see Hughton making subs earlier if we look completely impotent.[/quote]
In one sentence you downplayed this whole thread and took it back to what you thought... not bad going, by anyones standards.

Share this post


Link to post
Share on other sites
Judging by the lack of participation I am the only person keeping this thread alive with any sort of activity, it always amuses me how people point out the diversion of a thread without considering the likelihood that it would otherwise have sunk without a trace.

Share this post


Link to post
Share on other sites
[quote user="singupcarrowroad"]Judging by the lack of participation I am the only person keeping this thread alive with any sort of activity, it always amuses me how people point out the diversion of a thread without considering the likelihood that it would otherwise have sunk without a trace.[/quote]
Oh no, don''t get me wrong - I''m not belittling the attempts at keeping an interesting thread about. The more the merrier. I was just interested in the manner in which you replied, it didn''t seem to pay the original post any credit at all. That being said, I''m more inclined to believe you anyway. I don''t think looking at statistics and saying "There is your answer" is a good method... substitutions are effective on a game by game basis and any numer of factors could contribute to whether they help you win the game or not.

Share this post


Link to post
Share on other sites
I see where you are coming from completely stig, and perhaps I should have given the OP more credit.

I guess what I was trying to say is that irrespective of what the statistics were saying about Hughton in relation to every other manager in the EPL, his record this season would probably have been very different if he had more options at his disposal.

Share this post


Link to post
Share on other sites
The statistics do however show that the highest frequency for substitutions is 75-80 minutes, which is the very thing Hughton is criticised for doing, i.e. "making changes too late."

Share this post


Link to post
Share on other sites

I''d be interested to see what Hughton''s average substitution time is, I bet it is a lot later than 75 minutes. Usually the second sub gets 5 minutes + stoppage time.

Share this post


Link to post
Share on other sites
[quote user="singupcarrowroad"]I see where you are coming from completely stig, and perhaps I should have given the OP more credit.

I guess what I was trying to say is that irrespective of what the statistics were saying about Hughton in relation to every other manager in the EPL, his record this season would probably have been very different if he had more options at his disposal.[/quote]I think that''s a reasonable supposition.  It would be worth looking at what he did at Newcastle where he had a relatively rich squad in terms of depth and quality.While I can''t be bothered to do that, as I don''t care who''s right and who isn''t, it''s interesting that with a strong Newcastle side he played a 4-4-2 formation in 39 games of the 46...http://en.wikipedia.org/wiki/2009%E2%80%9310_Newcastle_United_F.C._season#StatisticsAnd at Birmingham 4-4-2 was also preferred, though I have found no data to back this up, just a quote from the man himself:“My preferred formation is 4-4-2 or a form of 4-4-2, 4-4-1-1, with one

playing a little bit more off the front, because I think that gives you a

little bit more in the opposition half. It allows you to keep the ball

more in their half.”http://www.edp24.co.uk/sport/norwich-city-fc/chris_hughton_ready_to_reveal_first_clues_on_norwich_city_formation_1_1453277

Share this post


Link to post
Share on other sites

Good thread Cambridge. And good input here from Mister. I think much of this is people being disappointed so playing the "you don''t wanna do it like that" card at Hughton. There is really no evidence that substitutions could have changed a draw into a win. I could just as easily have changed a draw into a defeat. Hughton also only made late substitutions in the wins against Man Utd and Arse. Should a defensive manager have brought on a defender? If we''d conceded a late goal in either game would posters have been here chastising Hughton for not making defensive substitutions?

 

Worthy brought on a defender in the play off final and we conceded a goal which ultimately lost it for us. He has been criticised because of the result. If he''d not made that substitution and we''d still conceded he''d have been criticised for not doing. Hindsight makes criticism that much easier and a manager should only really be judged on his results and league position when measured against realistic expectations.

 

 

Share this post


Link to post
Share on other sites
We started the season with Plan A, we are now playing Plan B.

Now that Hughton has options that might make a difference we may see some more positive changes.

What is the point in making changes to make your team weaker?

Share this post


Link to post
Share on other sites
[quote user="nutty nigel"]

Worthy brought on a defender in the play off final and we conceded a goal which ultimately lost it for us. He has been criticised because of the result. If he''d not made that substitution and we''d still conceded he''d have been criticised for not doing. Hindsight makes criticism that much easier and a manager should only really be judged on his results and league position when measured against realistic expectations.

[/quote]

 

Nutty in that particular case I think Worthy was seriously unlucky. He made a decision to use our final substitution with the game at 1-0. If I remember correctly didn''t Three Lungs then suffer an injury almost immediately afterwards that left him bascially walking around the pitch as he couldn''t run meaning we were bascially down to 10 men?

 

Davo

Share this post


Link to post
Share on other sites

As often with football stats, I''d like to know more about how they''ve treated the numbers.  They say putting on a defender for a midfielder is a defensive change but how often does that really happen ?  Most Prem teams play 4 at the back and you wouldn''t add a 5th defender.

 

Typically a defensive change to me would be taking off one of two strikers for, or replacing an attacking midfielder with a defensive one, or both.  From the summary of the research I''m not sure if this is what they''re meaning though.

 

Similarly they say most substitutions are tactically useless as it''s a change of player in the same position.  I''d say it depends, you can put a different player into the same position and it can be a tactical change depending on the individuals.

 

Looking back, under Lambert when we were behind at 60 minutes (and we were often behind then) he''d always bring on 2 subs and often change to a more attacking formation.  It was almost as if out gameplan worked best when we fell behind (e.g. Villa away) !  Under Hughton we are behind less often, and if we''re drawing or winning he''s clearly reluctant to change the team. Also Lambert learnt the lesson of not using 3 subs up when you''re winning, against Sunderland at home when we were 2-1 up with 10 minutes to go, Vaughan got injured and we had a sticky end to the game with 10 men and were very relieved to hold out. TBH you''d have thought that was pretty obvious but I noticed he never did it again in his time with us.

 

I can''t deny that it frustrates me to have the second sub coming on in the 87th minute if we''re losing or drawing. I wish he''d give players at least 10 minutes to make an impact.

Share this post


Link to post
Share on other sites
[quote user="singupcarrowroad"]I''d be interested to see what Hughton''s average substitution time is, I bet it is a lot later than 75 minutes. Usually the second sub gets 5 minutes + stoppage time.[/quote]I''d like there to be 100% less clothing on your avatar...... [A]

Share this post


Link to post
Share on other sites

[quote user="Xavid Fox"]You don''t need stats to realise Hughton''s substitutions are poor at best. No plan B[/quote]

Out of the first seven games he played 4-4-2 a total of 4 times.

Having got nothing but draws he changed to a 4-4-1-1 which led to a ten game unbeaten run.

In Footy there is no such thing as plan A or plan B. You put out a side in a formation best suited to them and to match the opposition. If it doesn''t work you change it if you have the personnel.

Share this post


Link to post
Share on other sites

[quote user="chicken"][quote user="Xavid Fox"]You don''t need stats to realise Hughton''s substitutions are poor at best. No plan B[/quote]

Out of the first seven games he played 4-4-2 a total of 4 times.

Having got nothing but draws he changed to a 4-4-1-1 which led to a ten game unbeaten run.

In Footy there is no such thing as plan A or plan B. You put out a side in a formation best suited to them and to match the opposition. If it doesn''t work you change it if you have the personnel.[/quote]

 

OK, but for shorthand, is it alright if we call it plan B when the initial formation/lineup hasn''t worked and is then changed ?

Share this post


Link to post
Share on other sites
[quote user="Xavid Fox"]You don''t need stats to realise Hughton''s substitutions are poor at best. No plan B[/quote]So he started a game with one set of players and ended with a different one. Isn''t that by definition a change of plan?And have you read singup''s frequent contributions (you should have, over 30% of the posts on this thread are by him [;)] )? The point he is making is that Hughton doesn''t have confidence that the alternative players available would be better than Plan A.

Share this post


Link to post
Share on other sites

Now I''m trying to think of occasions last season where Lambert made substitutions to turn draws into wins. I can remember a couple where Holt and/or Hoolahan were introduced and we manged to gain one point by getting a draw from a losing position. But I don''t remember him using subs to gain the two extra points from a draw to a win. Perhaps it''s my age. I remember games from the 60s and 70s like they were yesterday but I''m struggling with last season. Can anyone help me?

 

 

 

 

Share this post


Link to post
Share on other sites
[quote user="Phillip J Fry"]According to my calculations (taking times from whoscored.com) the average time that Hughton makes his first substitution is around 67 minutes. [/quote]
Whats more interesting to note is how varied his substitution times are. He''s made a substitution at half time 4 times so far this season. However 8 times this season he hasn''t made a substitution until after the 80th minute. It should also be noted that most of his ''earlier'' substitutions (i.e. before the 67th minute) come when Norwich are already two (or more) goals down or are due to injury. Tactical substitutions tend to be made later.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×
×
  • Create New...