[Albion] 'Enough with the data obsession, in the real world only goals for and against count'

[CheeseRolls]

Whether you like it or not, there’s a reason that all major clubs and organisations in elite sport are heavily invested in Data Science.

Which is that elite sport is big business and anything you can do to gain a marginal advantage, is a worthwhile investment.

There is however an element of truth in what Jim White and his ilk are saying. The most important factor here is that these stats are just indicators. It isn't just that we scored nil is more important than we have a high Xg (it absolutely is by the way). There are two specific questions to ask.

1 What actions can a manager take to address a high number of goals created versus goals scored, that is so different from the options to address a lack of goals scored?

2 Statistical indicators are general markers. A player who is poor at heading, one footed, easily brushed off the ball, plays with his head down etc, is going to miss chances based on those weaknesses regardless of whether the stats say he should have scored. You don't need an Xg score to know some players need to spend more time working on their technical weaknesses.

 

[Stato]

Just like in every other thread about stats/data it basically comes down to:

One set of people who have decided that the team/manager/players suck and therefore really wants stats/data to be bollocks, especially if it indicates that the team/manager/players are doing some things well.

One set of people who have decided that the team/manager/players are doing most things right and therefore wants stats/data to be interesting, especially if it indicates that the team/manager/players are doing some things well.

The difference between these sets of people is that pretty much everyone working in professional sports agree with the second set of people that stats and data are useful tools, hence every ****ing club gathering and using it.

Your first point is key. All statistics do is measure things. What you measure, how you measure and which measurements you decide to care about after you have your results is subject to risk of bias. The old adage about 'lies, damned lies and statistics,' should really speak of there being lies, damned lies and lies about statistics.

For me, one of the most interesting concepts discussed in Adam Curtis's latest documentary series was the suggestion that our decisions and choices are often made before our conscious mind is aware of it. It made me consider that the 'gut' feeling that people talk about and that research indicates can play a bigger role in decision making than logic or reason, is actually our conscious mind interpreting instructions it may be getting from the unconscious. If this is the case, its unsurprising that our conscious mind subsequently looks for patterns or evidence that support the decisions we have already made.

The approach of Billy Beane, as recorded in 'Moneyball' was that deliberately ignoring your gut feeling and believing the numbers that stats can provide can give you an advantage over those who refuse to do it. Baseball is of course a sport separated into small incidents of action that have limited metrics that are far easier to measure and seek to gain advantage from than a football match which is more subject to 'Lorenzian' ripple effects. This could mean that an understanding of chaos theory needs to be added to an understanding of statistics in order to provide a fuller picture. In other words, we can improve our chances by maximising all measurable performance statistics, but if that bloody butterfly continues flapping his wings, Lee Mason will lose the plot, the barn door will flap away from any of our shots and Christian Benteke will actually volley a ball where he intended it to go.

Someone accused [MENTION=1200]Harry Wilson's tackle[/MENTION] of being pretentious when he talked about (I'm paraphrasing because I can't find the post) constructing incidents of disorder, but his point was a good one. Incidents of disorder/disruption often influence football results: Palace's goal came from Ben White's skied ball; A lot of what Villa achieve is down to outcomes and reactions to the unpredictability of Grealish. It seems mad to talk about creating or controlling moments of chaos and turning them to your advantage, but I am starting to think that the absence of an assertive presence who can turn these things in our favour is part of our problem. We rely on the plan and when the plan is not working, we need the players on the pitch who can proactively change things. I'm reminded of the New Zealand rugby union coach (not enough to remember who it was, Warren Gatland perhaps?) who moved on from Clive Woodward's hyper analytical coaching methods by trying to teach his players to be their own on-field coaches and have the wherewithal to understand things whilst in the middle of play and change the plan to compensate. Unfortunately, I think that the football players capable of doing this are like hen's teeth and we currently don't seem to have any candidates.

 

[Johnny RoastBeef]

Very little other than showing "we are actually creating chances", which is something most of us are already aware of but its... nice (I suppose?) to be able to back it up with numbers. Its useful in portraying chance creation. For a coach its pretty much useless except in some circumstances - like if you have a player that repeatedly is taking shots in angles and situations where its very unlikely to score, it will be pretty evident through xG and you can tell him "dont do that shit, do this instead".

Personally I dont think xG is anywhere near as exciting as a lot of other data. Data showing how often Brighton win the ball within 5 seconds of Maupay pressuring their defender or how often Tariq Lamptey beats his defenders, things like that make me hard. xG... not so much.

Here is a filthy little Graph for you.

 

[raymondo]

Those poor managers in the Seventies. They’d watch their teams play well in defence and midfield and then bungle every chance of scoring, but they wouldn’t know what was happening and that they needed a goal scorer as there was no data science to tell them.

 

[Baldseagull]

There are a lot of pundits and journalists trying to use Brighton as a stick to beat Xg/Xp with. Our failure to turn dominance into results enables them to wave us at anyone who cares about stats and say 'Look, it's goals that count. Not your maths!' As if everyone who watches football doesn't already know this. What they are generally saying is 'I've done this job for years without having to understand this stuff and I don't see why I should have to learn anything new.'

They fail to realise/acknowledge that the only reason that people are talking about Xg etc. in connection with Brighton's results this season is because they are a massive deviation from the mean. We are not proof for the pointlessness of a modern statistical approach, we are actually the exception that proves the rule. Our anomaly just points to possible areas for further study.

I suggest the sample size is too small at the moment, and we go on to smash everyone 6-0 for the rest of the season.

 

[Stato]

I get your point but it is a rather sweeping generalization. Not rating XG as a worthwhile measure is not an indicator for being a dinosaur or lacking understanding. I have used Maths all of my professional career. I am pretty well qualified in statistics and probability. I do not fault the theory but I do question people who over interpret the results and then search out unbelievers in order to gratify their superiority complex. XG is an interesting tool, that’s all.

I'd agree with you, but I just don't really see evidence that people are persecuting unbelievers. I've only ever seen it presented as an interesting measurement tool and I would be as suspicious of anyone making greater claims for it, as I would of anyone saying that measurement and analysis are not worthwhile.

 

[raymondo]

Your first point is key. All statistics do is measure things. What you measure, how you measure and which measurements you decide to care about after you have your results is subject to risk of bias. The old adage about 'lies, damned lies and statistics,' should really speak of their being lies, damned lies and lies about statistics.

For me, one of the most interesting concepts discussed in Adam Curtis's latest documentary series was the suggestion that our decisions and choices are often made before our conscious mind is aware of it. It made me consider that the 'gut' feeling that people talk about and that research indicates can play a bigger role in decision making than logic or reason, is actually our conscious mind interpreting instructions it may be getting from the unconscious. If this is the case, its unsurprising that our conscious mind subsequently looks for patterns or evidence that support the decisions we have already made.

The approach of Billy Beane, as recorded in 'Moneyball' was that deliberately ignoring your gut feeling and believing the numbers that stats can provide can give you an advantage over those who refuse to do it. Baseball is of course a sport separated into small incidents of action that have limited metrics that are far easier to measure and seek to gain advantage from than a football match which is more subject to 'Lorenzian' ripple effects. This could mean that an understanding of chaos theory needs to be added to an understanding of statistics in order to provide a fuller picture. In other words, we can improve our chances by maximising all measurable performance statistics, but if that bloody butterfly continues flapping his wings, Lee Mason will lose the plot, the barn door will flap away from any of our shots and Christian Benteke will actually volley a ball where he intended it to go.

Someone accused [MENTION=1200]Harry Wilson's tackle[/MENTION] of being pretentious when he talked about (I'm paraphrasing because I can't find the post) constructing incidents of disorder, but his point was a good one. Incidents of disorder/disruption often influence football results: Palace's goal came from Ben White's skied ball; A lot of what Villa achieve is down to outcomes and reactions to the unpredictability of Grealish. It seems mad to talk about creating or controlling moments of chaos and turning them to your advantage, but I am starting to think that the absence of an assertive presence who can turn these things in our favour is part of our problem. We rely on the plan and when the plan is not working, we need the players on the pitch who can proactively change things. I'm reminded of the New Zealand rugby union coach (not enough to remember who it was, Warren Gatland perhaps?) who moved on from Clive Woodward's hyper analytical coaching methods by trying to teach his players to be their own on-field coaches and have the wherewithal to understand things whilst in the middle of play and change the plan to compensate. Unfortunately, I think that the football players capable of doing this are like hen's teeth and we currently don't seem to have any candidates.

I do think that the big change in our style of football has required the players to think a lot about what they are doing...but think too much I reckon. Sometimes I see we have 4 or 5 players in the opposition's box, so many passing opps and possibilities to think (albeit quickly) through, that no-one has the (unthought) feeling left to just take a touch and smash it at the goal!!!

 

[Farehamseagull]

While not a huge xG sucker, it is remarkable and interesting how often it turns out to be quite accurate in the long run.

Two good examples: last year Mason Greenwood scored a lot of goals but had low xG. People said "well shows xG doesnt matter, Greenwood will keep scoring!"... and look at him now.

And Liverpool. Last season they were 22 points above their xP (expected points) and plenty of people said "well shows xP doesnt matter, Liverpool will be in the top 2 next season!"... and look at them now.

While these are just two examples, it shows that while xG is perhaps not useful it often signals what is actually happening and thus in the long run is likely to come reward you or bite you in the ass. Hopefully for Brighton, the "reward" comes in the upcoming games and the next season!

They are very good examples. As are Southampton who as I pointed out during their good run were consistently outperforming their xG including against us. This run they're on now was coming.

 

[Neville's Breakfast]

I'd agree with you, but I just don't really see evidence that people are persecuting unbelievers. I've only ever seen it presented as an interesting measurement tool and I would be as suspicious of anyone making greater claims for it, as I would of anyone saying that measurement and analysis are not worthwhile.

There have been quite a few posts even on this thread that have been rather dismissive of the view that XG is not quite so illuminating as its supporters would have us believe. XG has been a recurrent theme in the ongoing Potter in/out debates. To one side it is evidence of overthinking and to the other (and it is to this that I am referring) it has been used as evidence that all is well with the managerial strategy as we will eventually get ‘what we deserve.’ The XG stats are actually so amazingly good that this argument has now gone further, with the assertion that the problem must be the club letting GP down with recruitment. Classic example of stats being used unquestioningly to ‘prove’ prior opinion.

 

[Guinness Boy]

Yes. You said it in a context where you were arguing against someone asking why clubs have sports science teams if its as useless as some of you seem to believe. You switch freely between "the results is what matters" and "joy is what matters" depending on what suits you - if we mix the two, it would suggest that "the joy of leaving all the stats/data crap out" would also result in better results, correct me (and elaborate) if I'm wrong?

Or what was really the meaning behind the whole insinuation that "sports science sucks because kids ask their parents after games if they scored or not" thingy?

FFS you've taken an interpretation you made yourself to prove a point I didn't make.

I'm sure at a professional level stats and sports science helps, just as tactics, facilities, crowd etc. All these external factors. But if you reduce football ONLY to a spreadsheet operated by data nerds and accountants you suck all the joy out of it. See also VAR's 30mm armpit offsides.

Secondly if you don't score you don't win. The article in the OP is fundamentally right about that.

So it's fairly easy to see that scoring = joy.

 

[Deleted member 2719]

In fact, enough already. Just thinking about Saturday makes me so angry at our appalling SHITHOUSE "forwards". They are just toss. F**k off the lot of you, and when you get there f*ck off some more. Yes, you Connolly. I don't care that you're 21 or whatever. You're confident enough to wear that stupid haircut. Just f*ck off to Rotherham. And you "Welbz". Sorry, another lovely guy I'm sure, but you are absolutely crap. I knew you would be. You, Ali J, are also crap. At least your overhead kick against Chelsea went in - we'll be able to repeat that on your highlight montage so that we can get it up to 30 seconds. Good job as we'll have leave off your pathetic header against 10-man Sheff Utd when my 10 year old would have nodded it in. I'll stop now, there are others. You f**king know who you are.

Yes the quality and the cool heads under pressure is not there at the moment.

Your rant really does no good at all for anyone apart from yourself.

Are you sure you not one of these entitled fans.

You disgust me.

 

[Stato]

It’s been a recurrent theme in the ongoing Potter in/out debates. To one side it is evidence of overthinking and to the other (and it is to this that I am referring) it has been used as evidence that all is well with the strategy as we will eventually get ‘what we deserve.’ The XG stats are actually so amazingly good that this argument has now gone further, with the assertion that the problem must be the club letting GP down with recruitment. Classic example of stats being used unquestionably to ‘prove’ prior opinion.

I'd interpreted those posts as trying to argue, not that all is well, but that because performances metrics point to improvements in many areas, even though results haven't followed, we should hold our nerve and carry on the path, trying to improve the things within our control. 'Keep The Faith' as MA would have put it. You're right that those who like the style will hang on to the metrics to support their decision and those who don't will dismiss them as irrelevant. Personally, although I can't predict which way it will go, I'm in the former camp, just because trying to do something new seems far more compelling than the alternative, which is just to try to pragmatically survive until we don't. The EPL can get a bit dull for the dozen or so makeweight teams, so trying to achieve something very ambitious in playing style may at least make the journey to the eventual fall that faces virtually all teams in the top division a bit more interesting than that of say Stoke or Palace.

 

[The_Viper]

And we need stats to tell us this?

With the amount of simpletons demanding Potter gets the sack and we bring Big Sam in ourselves, I would say that any metric possible is needed to educate these people.

 

[KeegansHairPiece]

�� I do think that the big change in our style of football has required the players to think a lot about what they are doing...but think too much I reckon. Sometimes I see we have 4 or 5 players in the opposition's box, so many passing opps and possibilities to think (albeit quickly) through, that no-one has the (unthought) feeling left to just take a touch and smash it at the goal!!!

So in our first season when we did quite well, we scored 21 goals in our first 26 games, 5 less than this season. Last season under Potter we'd scored 31 at 26 games played, 10 more than our first season in the division.

Under Potter we've scored 65 goals in 64 games that's led to 67 pts. For the previous 2 seasons we scored 69 goals in 76 games which gave us 74 pts. Baring in mind 1 player got 25 of those 69 goals (36%).

Now, what I would put to you is if we had a player like Murray was 2017-19, with the tactics we are now playing, with the increased chances we are creating, with the space he'd make for others, we'd be on 35+ goals for this season because I feel almost certain a striker with a real nose for goal would be on 12+ for us by now, and we'd be sat on 30+ pts because we'd have got a draw and a win out of the last 2, and likely have beaten Fulham or Villa or both.

 

[golddene]

Those poor managers in the Seventies. They’d watch their teams play well in defence and midfield and then bungle every chance of scoring, but they wouldn’t know what was happening and that they needed a goal scorer as there was no data science to tell them.

Thing is though RossyG in the seventies we could remedy any shortcomings by entering the transfer market and buy a replacement, not saying it was better then especially because we could just as easily lose the players that are creating for us when other bigger clubs came calling.

 

[blue-shifted]

Call me Mr Tyranasaurus Rex, but I completely fail to see what stats actually tell you that you can't tell, just by watching the game.

Our strikers, are under performing their XG? Oh really?
Ryan was having a poor season. Wow, really useful this

I can sort of see the use in a scouting sense. Ie if you're scouting 1000 defenders, it's easier to look at percentage headers won on a spreadsheet that it is to watch 1000 games, but when you actually have the players at the club, can anyone give me an example of some insight that a statistic could give you that you COULDN'T get just by watching the game?

 

[Simster]

You disgust me.

Oh no, how will I cope?

But seriously, I'm just delighted that you've graduated from village idiot to being able to form an opinion without reading what to think in the Daily Mail beforehand. So very well done for that. (I'd pat you on the head patronisingly at this point, if I could)

 

[Seagull58]

But what element of possession should we care about?

Liverpool 0 Brighton 1 - Brighton 37% possession
Brighton 1 Spurs 0 - Brighton 43% possession
Leeds 0 Brighton 1 - Brighton 33% possession
Brighton 1 Palace 2 - Brighton 75% possession

3 classic Hughtonesgue performances and one classic potterball performance.

 

[blue-shifted]

Surely you don't. Surely you're aware you can't watch every action, by every player, at all times... and remember it ALL flawlessly for retrospective comparisons, without any bias, without any momentary lapse of concentration, or false memories.. nothing... just a flawless analyses of performances... all from memory... that spans the season.

No. I'm not sure any clever person thinks "I won't bother writing any of this down, I'll just watch it and go with my gut instinct". Or "there's all these useful figures based on what occurred, but who needs them when I can see what happens and go with what feels right!"

Or .... you can just watch the games back?

 

[CheeseRolls]

There have been quite a few posts even on this thread that have been rather dismissive of the view that XG is not quite so illuminating as its supporters would have us believe. XG has been a recurrent theme in the ongoing Potter in/out debates. To one side it is evidence of overthinking and to the other (and it is to this that I am referring) it has been used as evidence that all is well with the managerial strategy as we will eventually get ‘what we deserve.’ The XG stats are actually so amazingly good that this argument has now gone further, with the assertion that the problem must be the club letting GP down with recruitment. Classic example of stats being used unquestioningly to ‘prove’ prior opinion.

I agree - I was reading a paper recently on Artificial Intelligence applied to chest scans. In one study highlighted the algorithim was able to isolate a particular heart condition and predict with a high degree of accuracy the patients who would die within 12 months of a scan. However when you consider that the patients had to die in order to validate the results and that the AI was not able to propose any practical intervention to prevent this. What are you actually left with?

Using an Xg stat to try and prove a point, just means you have taken the time to read the stats. It shows you what HAS happened and gives you as a supporter ZERO ability to change the future course of events. If you think that somehow means your point of view is right, you are sadly deluded.