Statistical rubbish

[jt64]

Someone once said that football's a simple game, you put it in at one end and keep it out the other. So as I've got a bit of time on my hands today, I thought I'd run some numbers on this principle: if you're on the pitch when we score, I give you one point. If you're on the pitch when we concede, I take a point away. For anyone that's ever watched ice hockey, it's exactly the same as their plus/minus figure. So, without further ado:

Raw figures:

Spoiler: wall of text - for context, our goal difference is +2

8 - Mark Howard
5 - Matt Edgar, Conor Sammon
4 - Jose Baxter, Kieran Wallace
3 - Jamal Campbell-Ryce
2 - Kieran Freeman, Chris Basham, Che Adams, Stefan Scougall, James Wallace, Craig Alcock, Matt Done, Harrison McGahey
1 - Jay McEveley, Terry Kennedy
0 - Louis Reed, Billy Sharp, Mark McNulty, Michael Higdon, Paul Coutts
-1 - Neill Collins, Martyn Woolford
-3 - Ryan Flynn, Dean Hammond, Bob Harris
-4 - Callum McFadzean, Jamie Murphy
-6 - George Long

Per 90 minutes (minimum of 270 minutes played):

Spoiler: another wall of text

0.60 - Stefan Scougall
0.57 - Harrison McGahey
0.53 - Mark Howard
0.50 - Matt Edgar, Conor Sammon
0.47 - Kieran Wallace
0.43 - Matt Done
0.36 - Jamal Campbell-Ryce
0.32 - Jose Baxter
0.15 - Che Adams
0.12 - Jay McEveley
0.11 - Kieron Freeman, Chris Basham
0.00 - Louis Reed, Billy Sharp, Paul Coutts
-0.06 - Neill Collins
-0.22 - Martyn Woolford
-0.53 - Dean Hammond
-0.75 - Bob Harris
-0.84 - Ryan Flynn
-1.50 - George Long

We're nearly 20 games in, so while some things may be affected by sample size (Long's only played 4 games and coincidentally the entire team shat the bed in two of them, so I wouldn't read too much into the keeper figures and call for him to be dropped), I'd say:

- maybe some of Adkins' loan signings aren't quite as awful as we think. Sammon and Edgar have both played over half the available minutes, and per game seem to be contributing a lot more than many of the other options at their position. Maybe look to get Edgar starting with someone other than Collins, i.e. the only real regular contributor with a negative figure? That said, the less said about his other acquisitions, the better

- Wallace's figures surprised me, think it was pretty harsh for him to get dropped for Harris, and while Bob's could be sample size related and may improve, I'd at least look to give Kieran another run

- Maybe a central midfield of Basham holding and Baxter/Scougall doing everything else might not be awful?

- We've used a fucking ton of players

edit - this is league only by the way

 

[mattbianco1]

-4 - Callum McFadzean, Jamie Murphy

Good job we sold Murphy then.. He was shit

Seriously though, I like this. Well done.

We went 4 on the bounce early in the season. Sammon was our new hero. Edgar was playing great.. WTF has happened?

 

[I should coco]

For anyone that's ever watched ice hockey

You were doing so well until.......

Zzzzzzzzzzzzzzzzzzzzzzzzzz

 

[Alfreton14Blade]

Going by the top scores (& who's avaialable), that'd give us a starting XI of:

Howard
Freeman Edgar McGahey KWallace
JCR Basham Baxter Scougall
Adams Sammon

 

[PokerBlade]

Sometimes statistics get in the way of analysing humans whose performance can fluctuate for reasons that are either not known, not predictable, or not representable in stats.

On the other hand, whenever I see or hear discussions about other sports, it makes me think how far behind football is in the analysis.

Radio Sheffield's question to Adkins after the game was something like "Will you have to reassess things before Saturday?". Um yeah, he will. Meanwhile he never faces the question as to why he brought two centre backs on, why we have so few shots on goal (I haven't looked at stats for that, but I think it's been true of us relative to opposition), etc.

When you see the breakdowns they give for ice hockey or basketball it's on another level to football.

 

[Geordie Blade]

Long's only played 4 games and coincidentally the entire team shat the bed in two of them

I really like the post, but the nerd in me is wary of statements like the quote above. Isn't the whole point of your exercise to look for correlations between independent variables (e.g. Long playing) and independent variables (goals scored/conceded)? If there's a strong relationship, you can't just write it off as "coincidence"; that's what confidence intervals and statistical significance are for. It's quite possible that it is only coincidence that we've capitulated in 50% of Long's appearances, but it is also possible that there is a cause-effect relationship that needs to be looked at.

 

[Pollingtonblade]

Sometimes statistics get in the way of analysing humans whose performance can fluctuate for reasons that are either not known, not predictable, or not representable in stats.

On the other hand, whenever I see or hear discussions about other sports, it makes me think how far behind football is in the analysis.

Radio Sheffield's question to Adkins after the game was something like "Will you have to reassess things before Saturday?". Um yeah, he will. Meanwhile he never faces the question as to why he brought two centre backs on, why we have so few shots on goal (I haven't looked at stats for that, but I think it's been true of us relative to opposition), etc.

When you see the breakdowns they give for ice hockey or basketball it's on another level to football.

Those fucking idiots at RS and the Sheffield Star wouldn't know journalism if it kicked them in the face.

 

[Geordie Blade]

Sometimes statistics get in the way of analysing humans whose performance can fluctuate for reasons that are either not known, not predictable, or not representable in stats.

I disagree with this. Yes, the performance can depend on a whole range of factors (some of which are known and some that are not known), but statistics deal with averages where a lot of that 'noise' is cancelled out. If you think that there are factors that can influence things, then you put it into the analysis. If there is something that you hadn't thought of, then you're not doing statistics properly. If there's something that you couldn't have thought of, then you still undertake tests to see how much 'unexplained' variation there is in your results. Humans are very variable, but we still use statistics to make decisions about which medicine works best when you're ill (despite knowing that the drugs don't work on everyone, and that each individual person varies in terms of age, weight, gender, diet, family history, etc.).

The main problem that you'd face with an analysis like the one above is 'confounding'. This happens when you see a link between two things (e.g. Ryan Flynn is linked with a poor outcome) and make an assumption that the cause-effect relationship is in one particular direction (i.e. "we concede more goals when Flynn plays, therefore he is crap"). It could be possible that Flynn gets picked for the more difficult games because he offers a more defensive role than, say JCR. In that the case, the cause-relationship would be in the opposite direction (i.e. "Flynn tends to get picked for the games in which we're more likely to concede goals"). It's difficult to so stats to work this sort of thing out without running a randomised controlled trial.

[Edit - wow, that was boring, sorry!]

 

[jt64]

Isn't the whole point of your exercise to look for correlations between independent variables (e.g. Long playing) and independent variables (goals scored/conceded)?

The whole point was that I'm bored this afternoon

That said, you have a point, although I think keepers will always be a special case - you never know when they're playing well unless the team as a whole is playing badly, as they're not going to have much to do. It's not like we're Bayern whereby Neuer could have let in every single shot on target and they'd still be top of the league (actually true by the way). I'm mostly looking for something that may not necessarily be picked up in your orthodox figures.

 

[PokerBlade]

Geordie, if we analyse something like penalty conversions by a player, we might never gain enough data on them to be sure our conclusions are meaningful. How many standard deviations away from being a good penalty taker is Sharp this season? We'll never see hundreds of penalties by him to get a meaningful sample size. Where as something like pass completion for a player in a certain position could easily be analysed.

Statistics are very useful in sport and I'm not at all rejecting them. I'm just saying that the limitations can sometimes be underestimated.

 

[OllyBlade]

Sometimes statistics get in the way of analysing humans whose performance can fluctuate for reasons that are either not known, not predictable, or not representable in stats.

On the other hand, whenever I see or hear discussions about other sports, it makes me think how far behind football is in the analysis.

Radio Sheffield's question to Adkins after the game was something like "Will you have to reassess things before Saturday?". Um yeah, he will. Meanwhile he never faces the question as to why he brought two centre backs on, why we have so few shots on goal (I haven't looked at stats for that, but I think it's been true of us relative to opposition), etc.

When you see the breakdowns they give for ice hockey or basketball it's on another level to football.

http://experimental361.com/2015/10/26/season-so-far-league-1-24-oct-2015/

I recommend this website. Admittedly this is a month old now, but shows that actually we were taking quite a lot of shots relative to the division. It suggests our problem was more the other end, or in getting a midfield balance, as we were also facing a lot of shots too. It looks like they are putting out new stats though for the last month at the moment.

Geordie, if we analyse something like penalty conversions by a player, we might never gain enough data on them to be sure our conclusions are meaningful. How many standard deviations away from being a good penalty taker is Sharp this season? We'll never see hundreds of penalties by him to get a meaningful sample size. Where as something like pass completion for a player in a certain position could easily be analysed.

Statistics are very useful in sport and I'm not at all rejecting them. I'm just saying that the limitations can sometimes be underestimated.

There are would seem to be a lot more secondary statistics you'd want to know in football as well, for things where you might have enough samples. E.g. rate of successful passes as a %. You might also want to know what the average length of pass the player made was, their average position, how many passes directly led to a shot, etc, else the stat could be misleading.

 

[stringjunior]

I like a good stat, and these are good stats, in a bad sort if way....

 

[William Henry Foulkes]

Gestalt

 

[PeterNdlovu081]

Why is Murphy on there?

 

[jt64]

Why is Murphy on there?

Because he played a game?

 

[PeterNdlovu081]

Because he played a game?

When?

 

[jt64]

When?

I'm glad you've got Gillingham out of your memory, I'm still trying

 

[PeterNdlovu081]

I'm glad you've got Gillingham out of your memory, I'm still trying

I'm glad we got rid of Murphy. We haven't lost 4-0 again since.

 

[Hunkerby]

The Per 90 minutes stat really highlights how bad the 'Adkins old boy' signings have performed this season.
The trio of Sharp, Woolford & Hammond all towards the bottom of the list.

 

[grafikhaus]

I recommend this website. Admittedly this is a month old now, but shows that actually we were taking quite a lot of shots relative to the division. It suggests our problem was more the other end, or in getting a midfield balance, as we were also facing a lot of shots too. It looks like they are putting out new stats though for the last month at the moment.

But there are 'shots' and 'shots'. J C-R had a 'shot' into the kop in the Southend game, and that's where it ended up. In the kop. The MoTD stats. are notoriously bogus. '% Possession' means something if you're Barca but is tedious shit if you're Swansea (finally been found out) or United. A 'shot on target' could be one that bounces a few times before the goalie 'throws his cap on it', whereas one which screams just past the post has more success ratio. If you get my drift.

And that's the true result of employing useless, 'safe' managers. While we are happy to bore teams into submission with endless, pointless sideways and backwards tripe it's easy for those with vested interests- the board, the manager, Radio Sheffield - to blame the fans for being 'impatient'.

If there is a more 'patient' bunch of supporters out there than United's, I'd like to know who they are. That's why the likes of McNulty can get away with taking the piss. Clough chose public humiliation as his way to address McN's frailities. Adkins' was to get rid of a highly-gifted player. Portsmouth seem to have found the magic formula - management. I'm sure the clue is in the job description.

 

[Deaneandagana]

Someone once said that football's a simple game, you put it in at one end and keep it out the other. So as I've got a bit of time on my hands today, I thought I'd run some numbers on this principle: if you're on the pitch when we score, I give you one point. If you're on the pitch when we concede, I take a point away. For anyone that's ever watched ice hockey, it's exactly the same as their plus/minus figure. So, without further ado:

Raw figures:

Spoiler: wall of text - for context, our goal difference is +2

8 - Mark Howard
5 - Matt Edgar, Conor Sammon
4 - Jose Baxter, Kieran Wallace
3 - Jamal Campbell-Ryce
2 - Kieran Freeman, Chris Basham, Che Adams, Stefan Scougall, James Wallace, Craig Alcock, Matt Done, Harrison McGahey
1 - Jay McEveley, Terry Kennedy
0 - Louis Reed, Billy Sharp, Mark McNulty, Michael Higdon, Paul Coutts
-1 - Neill Collins, Martyn Woolford
-3 - Ryan Flynn, Dean Hammond, Bob Harris
-4 - Callum McFadzean, Jamie Murphy
-6 - George Long

Per 90 minutes (minimum of 270 minutes played):

Spoiler: another wall of text

0.60 - Stefan Scougall
0.57 - Harrison McGahey
0.53 - Mark Howard
0.50 - Matt Edgar, Conor Sammon
0.47 - Kieran Wallace
0.43 - Matt Done
0.36 - Jamal Campbell-Ryce
0.32 - Jose Baxter
0.15 - Che Adams
0.12 - Jay McEveley
0.11 - Kieron Freeman, Chris Basham
0.00 - Louis Reed, Billy Sharp, Paul Coutts
-0.06 - Neill Collins
-0.22 - Martyn Woolford
-0.53 - Dean Hammond
-0.75 - Bob Harris
-0.84 - Ryan Flynn
-1.50 - George Long

We're nearly 20 games in, so while some things may be affected by sample size (Long's only played 4 games and coincidentally the entire team shat the bed in two of them, so I wouldn't read too much into the keeper figures and call for him to be dropped), I'd say:

- maybe some of Adkins' loan signings aren't quite as awful as we think. Sammon and Edgar have both played over half the available minutes, and per game seem to be contributing a lot more than many of the other options at their position. Maybe look to get Edgar starting with someone other than Collins, i.e. the only real regular contributor with a negative figure? That said, the less said about his other acquisitions, the better

- Wallace's figures surprised me, think it was pretty harsh for him to get dropped for Harris, and while Bob's could be sample size related and may improve, I'd at least look to give Kieran another run

- Maybe a central midfield of Basham holding and Baxter/Scougall doing everything else might not be awful?

- We've used a fucking ton of players

edit - this is league only by the way

I have to say I like stats, numbers, ratios, trends and all things related to inverted U shape diagrams - it all adds to the wonderful complexities and nuances of football.

I also like honesty and hard work & coming from Sheffield I understand & can reconcile with people who have this disposition as part of their makeup - the point being, that in his recently published book "the crazy gang" Dave Bassett alluded to one of his biggest strengths in his career which was his ability to recognise the major cultural characteristics of the fans of the clubs he was managing and to produce a team that both displayed those characteristics and produced performances to match.

Compare & contrast his teams to the current bunch.

 

[Bartleby]

We went 4 on the bounce early in the season. Sammon was our new hero. Edgar was playing great.. WTF has happened?

we lost some momentum with that international break, plus Adams came back with a virus and Edgar seemed to struggle after the break as well.