Mid table obscurity.

[Davefevs]

Just now, VT05763 said:

Correct me if I am wrong but are you not saying that if you include the Fulham goals on Kings record then we have conceded more goals this season with him in the team but if you don't count those 6 then we haven't ? 

That is what I thought you meant by "skewed" eg the Fulham game makes it look worse.

My point was that you cannot pick and chose the stats to support a narrative. In this case we concede less goals with King in the side, if you take out the Fulham game.

Apologies if I have misunderstood

No, I just said it was skewed, didn’t say take it out.  I wasn’t agreeing or disagreeing with Swan either.  It provides some added context to an “average”, which by virtue of being an average doesn’t always tell the full story, especially for someone who’s not played many minutes.  I could’ve just replied with a figure, e.g. 2.09 per 90.  I could provide the whole squad….here it is.

Looking for extremes in data, rather than just averages, was one of my big learnings in analysing data from my time spent at Warwick Business School, especially when taking small sample sizes.  I surveyed customer satisfaction survey scores from London Zoo (was actually a large sample).  We were presented with averages, made our conclusions, then they gave us the raw data and we saw the odd crap score (and comments) that highlighted an issue with a single employee.  If you’re bored, the other big learning is referenced in my “placebo” thread from a few seasons back.

When I’m buying something from Amazon, I look at the review score, but I always read a few of the reviews where they give 0 stars.  I don’t remove them and recalculate the average, I look at why they gave 0.  Was it a product feature that is key to my usage?  Is it a poor returns experience? Etc.

[chinapig]

1 minute ago, VT05763 said:

Ah O.K. so King hasn't played in many games so we shouldn't use him in any stats. 

That would render quite a few stats useless but good to know.

Thanks.

 

Not quite. In the particular case of whether or not King playing makes a material difference to the number of goals we concede I am suggesting there isn't enough data to draw a meaningful conclusion. I suppose there are other confounding factors such as who did he play against, who else was in the team and so on so you may be unable to attribute the outcome to a single player anyway.

Measures that are personal to him like passing accuracy for instance would still be meaningful though.

[Davefevs]

7 minutes ago, VT05763 said:

Ah O.K. so King hasn't played in many games so we shouldn't use him in any stats. 

That would render quite a few stats useless but good to know.

Thanks.

 

I’ll let Chinapig respond re his own “rules” for what is a good sample size, but Louis Britton’s 4.29 goals per 90 at Championship level might help answer?

[chinapig]

3 minutes ago, Davefevs said:

No, I just said it was skewed, didn’t say take it out.  I wasn’t agreeing or disagreeing with Swan either.  It provides some added context to an “average”, which by virtue of being an average doesn’t always tell the full story, especially for someone who’s not played many minutes.  I could’ve just replied with a figure, e.g. 2.09 per 90.  I could provide the whole squad….here it is.

Looking for extremes in data, rather than just averages, was one of my big learnings in analysing data from my time spent at Warwick Business School, especially when taking small sample sizes.  I surveyed customer satisfaction survey scores from London Zoo (was actually a large sample).  We were presented with averages, made our conclusions, then they gave us the raw data and we saw the odd crap score (and comments) that highlighted an issue with a single employee.  If you’re bored, the other big learning is referenced in my “placebo” thread from a few seasons back.

When I’m buying something from Amazon, I look at the review score, but I always read a few of the reviews where they give 0 stars.  I don’t remove them and recalculate the average, I look at why they gave 0.  Was it a product feature that is key to my usage?  Is it a poor returns experience? Etc.

Plus of course if you calculate the mean it will be distorted by extreme values. My preference is often for the median to smooth out that effect.

Though I often found that users couldn't get their head around the simple difference. A depressing comment on the numeracy of British managers.

[ExiledAjax]

3 minutes ago, chinapig said:

Plus of course if you calculate the mean it will be distorted by extreme values. My preference is often for the median to smooth out that effect.

Though I often found that users couldn't get their head around the simple difference. A depressing comment on the numeracy of British managers.

Median can also better when dealing with something like goals scored per game. Where the mean will produce a fractional value, but the data you're discussing comes only in integers.

Eg our mean goals scored per game is 1.32, but the median (and mode) is 1. Likewise goals against is 1.79, but the median (and mode) is 2.

So in both cases, although rounding to the nearest integer helps, the "average" when including the fraction, makes it all look better than it really is in real world speak.

On some of the other points, small number bias is a huge issue in football stats that you need to bear in mind all the time.

[Davefevs]

5 minutes ago, chinapig said:

Plus of course if you calculate the mean it will be distorted by extreme values. My preference is often for the median to smooth out that effect.

Though I often found that users couldn't get their head around the simple difference. A depressing comment on the numeracy of British managers.

You might like this type of view?

 

[Davefevs]

This is probably a better one.  2 shots in a 13 minute sub appearance, gave him a 13.85 shots per 90 in one game.  He did score from 1 of those 2 shots though!

[Waconda]

36 minutes ago, Davefevs said:

This is probably a better one.  2 shots in a 13 minute sub appearance, gave him a 13.85 shots per 90 in one game.  He did score from 1 of those 2 shots though!

Nice, so when does a players stats become "live". How many games, minutes etc.

Who decides ? Is it subjective or is there a rule ?

12 starts not enough then ?

 With him starting conceded 20 =1.66 per game

Without conceded 24 = 1.14 per game

That is a larger enough sample size IMO.

[downendcity]

3 hours ago, Robbored said:

Skew - to give an oblique direction to; shape, form, or cut obliquely.

You are Suzi Dent from Countdown and I claim my £5!

 

 

[downendcity]

2 hours ago, chinapig said:

Plus of course if you calculate the mean it will be distorted by extreme values. My preference is often for the median to smooth out that effect.

Though I often found that users couldn't get their head around the simple difference. A depressing comment on the numeracy of British managers.

They get “it’s a game of 2 halves “ correct - how much numeracy do you expect of them? 

[Davefevs]

2 hours ago, VT05763 said:

Nice, so when does a players stats become "live". How many games, minutes etc.

Who decides ? Is it subjective or is there a rule ?

12 starts not enough then ?

 With him starting conceded 20 =1.66 per game

Without conceded 24 = 1.14 per game

That is a larger enough sample size IMO.

Why are you asking me?

They can be whatever you want them to be, as long as you call out your method / caveats.  Others can use different methods, or call out / challenge.  Remember I never said anything stats becoming live or not, removing stats, etc….I purely made a comment that King’s are skewed by the Fulham game.  I never made any qualification as to whether that was good or bad.

For info my pic above (re City) is goals scored and conceded whilst they are on the pitch.  That’s a decision by FBRef, the suppliers of the data in this case.  I merely put them in a table and sorted them in order to provide something that is easy to view 

I don’t drop any data in any of my databases.  I extract every player who’s played 1 minute this season.

If you want to use your own method, based on games started, or whatever other method, fine, fill your boots.  Maybe @The Swan and Cemeteryis interested in your results and any conclusions you reach?

Any more questions?  Happy to keep answering!

 

[ExiledAjax]

2 hours ago, downendcity said:

They get “it’s a game of 2 halves “ correct - how much numeracy do you expect of them? 

Yet they demand 110% effort at all times.

[Simon bristol]

23 hours ago, BasSavage88 said:

I still don't think he's right for us whilst there are some improvements there's also alot of worrying steps back 

Yeah im inclined to agree. I dont really see how suddenly in august our defence will be fixed and we might get to see some consecutive wins. We just arent set up to stop crosses coming in and stop goals going in. For that tough a character, discipline in defence should be the pre requisite of everything he stands for.

[Davefevs]

6 minutes ago, Simon bristol said:

Yeah im inclined to agree. I dont really see how suddenly in august our defence will be fixed and we might get to see some consecutive wins. We just arent set up to stop crosses coming in and stop goals going in. For that tough a character, discipline in defence should be the pre requisite of everything he stands for.

Unfortunately he’s not the one who takes to the field.  You can see why he really liked Baker, and why he was keen to show his praise for Atkinson’s ability to put his foot through the ball in early season interviews (pre-injury). I imagine he saw a little bit of himself in them.

Not all managers replicate their own playing style in the teams they manage.

Would you go back to a duller style of football for a better defence, e.g. would you take 3 dull nil-nils over a 4-2 win and two 3-1 defeats that were a bit more end-to-end.  Ignore the goal difference for the sake of the question.

[Robbored]

52 minutes ago, Davefevs said:

Unfortunately he’s not the one who takes to the field.  You can see why he really liked Baker, and why he was keen to show his praise for Atkinson’s ability to put his foot through the ball in early season interviews (pre-injury). I imagine he saw a little bit of himself in them.

Not all managers replicate their own playing style in the teams they manage.

Would you go back to a duller style of football for a better defence, e.g. would you take 3 dull nil-nils over a 4-2 win and two 3-1 defeats that were a bit more end-to-end.  Ignore the goal difference for the sake of the question.

I’ve always gone to AG to be entertained but as we all know that doesn’t always happen regardless of who the manager is.

Some managers like Jordan, Ward and DW liked to play attacking front foot football whereas SoD and both Johnson’s served the dullest football I’ve ever seen at AG.

I believe that when Nige has a fully squad to select from that we’ll see some entertaining stuff once again.

[Waconda]

9 hours ago, Davefevs said:

Why are you asking me?

They can be whatever you want them to be, as long as you call out your method / caveats.  Others can use different methods, or call out / challenge.  Remember I never said anything stats becoming live or not, removing stats, etc….I purely made a comment that King’s are skewed by the Fulham game.  I never made any qualification as to whether that was good or bad.

For info my pic above (re City) is goals scored and conceded whilst they are on the pitch.  That’s a decision by FBRef, the suppliers of the data in this case.  I merely put them in a table and sorted them in order to provide something that is easy to view 

I don’t drop any data in any of my databases.  I extract every player who’s played 1 minute this season.

If you want to use your own method, based on games started, or whatever other method, fine, fill your boots.  Maybe @The Swan and Cemeteryis interested in your results and any conclusions you reach?

Any more questions?  Happy to keep answering!

 

Thanks for confirming, got there in the end.

it is why "stats" and spreadsheets can never be relied on in the real world.

Boris saying crime rate had dropped because he conveniently ruled out fraud as a "real" crime is a perfect example

Bristol City have conceded more goals without Andy King in the side if you remove the "skewed" Fulham loss.

Always believe your eyes on the training ground and pitch.

This is actually what NP has reinstated at City, got rid of much of the stat nonsense. He didn't sign Simpson, James and King based on their "numbers" for instance.

Edited February 28, 2022 by VT05763

[supercidered]

On 26/02/2022 at 21:21, VT05763 said:

He did, that is all that matters, surely ?

 

[City oz]

9 hours ago, Robbored said:

I’ve always gone to AG to be entertained but as we all know that doesn’t always happen regardless of who the manager is.

Some managers like Jordan, Ward and DW liked to play attacking front foot football whereas SoD and both Johnson’s served the dullest football I’ve ever seen at AG.

I believe that when Nige has a fully squad to select from that we’ll see some entertaining stuff once again.

What is a full squad to pick from under this current climate. If you know the answer please tell

[Robbored]

1 minute ago, City oz said:

What is a full squad to pick from under this current climate. If you know the answer please tell

A strange question Oz……………..

Under the current circumstances Nige continues to have experienced players missing so remains unable to select his strongest team.

[City oz]

2 minutes ago, Robbored said:

A strange question Oz……………..

Under the current circumstances Nige continues to have experienced players missing so remains unable to select his strongest team.

Not sure though if Nige knows him self what the real team should be if all players fit. He has tried many combinations but the inconsistency seems to be more of a word lately being splashed around rather than performing. There seems that some players do not to have the commitment that they should week in and week out. Not sure if that is because of injuries or something else going on behind the curtain we are not aware of.

The team over all did try hard at Forest but at the end of the day we were second fiddle to a Forest team that are playing OK but not top six potential.

[Major Isewater]

16 hours ago, ExiledAjax said:

Median can also better when dealing with something like goals scored per game. Where the mean will produce a fractional value, but the data you're discussing comes only in integers.

Eg our mean goals scored per game is 1.32, but the median (and mode) is 1. Likewise goals against is 1.79, but the median (and mode) is 2.

So in both cases, although rounding to the nearest integer helps, the "average" when including the fraction, makes it all look better than it really is in real world speak.

On some of the other points, small number bias is a huge issue in football stats that you need to bear in mind all the time.

Quite.

[Davefevs]

4 hours ago, VT05763 said:

Thanks for confirming, got there in the end.

it is why "stats" and spreadsheets can never be relied on in the real world.

Boris saying crime rate had dropped because he conveniently ruled out fraud as a "real" crime is a perfect example

Bristol City have conceded more goals without Andy King in the side if you remove the "skewed" Fulham loss.

Always believe your eyes on the training ground and pitch.

This is actually what NP has reinstated at City, got rid of much of the stat nonsense. He didn't sign Simpson, James and King based on their "numbers" for instance.

Got where in the end?  You seem to be debating with yourself and raising points of debate that nobody is making, to continue a debate with yourself?

Not sure why you’re trolling my posts?  Feel free to take into DMs if you like, would save clogging up this thread.

 

 

 

[Sheltons Army]

3 hours ago, Davefevs said:

Got where in the end?  You seem to be debating with yourself and raising points of debate that nobody is making, to continue a debate with yourself?

Not sure why you’re trolling my posts?  Feel free to take into DMs if you like, would save clogging up this thread.

 

 

 

Too busy sellotaping together his season ticket .....

 

A prinny fit so to speak 

[PHILINFRANCE]

4 hours ago, Davefevs said:

Got where in the end?  You seem to be debating with yourself and raising points of debate that nobody is making, to continue a debate with yourself?

I have now read this paragraph several times, and am still laughing.

Oscar Wilde would have been proud of such a post.

[Davefevs]

1 minute ago, PHILINFRANCE said:

I have now read this paragraph several times, and am still laughing.

Oscar Wilde would have been proud of such a post.

I’ll have to take your word for that….”Kids in America” was good though!