Stats - how much notice do you take?

[Woolly Blue]

So the other day when discussing the potential of us signing Gylfi Sigurdsson on Twitter, I shared a link to a heavily statistics based post which demonstrated that in quite a few areas there was little between him and Ross...

Anyway, it provoked a mini debate about the importance of statistics. Personally, I'm a bit of a 'stats centrist'. I mean, I don't put too much emphasis on them but I do think (when presented with context) they can be useful.

I was thinking of doing a podcast on the topic and therefore wanted your thoughts?

Basically, I want to know how much notice you take of them? Do you proactively seek them out? And how much do they influence your thoughts on certain footballers?

If it's one of our players, who I see week in, week out, I take no notice of stats and go with what I see on the pitch.
For players who I don't see as often, I take more notice of them.

 

[Abraders]

70% of stats are made up doe ladssss

 

[Berenger]

It's the opposite of that. Not sure you read the article, but it's using stats to delve beyond the headline figure - e.g. where and how the assists are happening.

"Top Trumps" would be "Barkley got X assists, Sigurdsson got Y assists, so one of them is better at assists."

Agree the article is a notch up from the latter, but not much. The key issue is context and controlling factors. All the article does is explain why 'chances' in 'chances created' are not all the same. That's just understanding your data.

Useful statistical analysis would go further in trying to control for several factors, *given* understanding of what the data is measuring, such as: Swansea's level, players around him, his role in the team, opposition and so on.

 

[Philw]

The modern statistics fetish wears me down to be honest! Ever since players started wearing those radio powered support bra's therre has been a requirement for a statistics orgy to populate the dross that is skySports news!

I dont care that player A ran 2km more than players b, because many players run around like headless chickens and notch the miles up! Others run intelligently, reducing the miles required but achieving the same or better output? Similarly, all stats about passing are garbage! If i play with a confidence player like Barkley and his confidenc eis low, as soon as I give him the ball, he will give it staright back to me! That equals a successful pass for Barkley but puts me into trouble and possibly loses the ball to the opposition! Similarly, when his confidence is up, he might make far fewer passes, but they will be 30 yard worldies? i know which I would prefer, but reviewing statistics would suggets that the games when he has been passing the ball staright back to who he got it from equates to a better performance, when it simply does not! There are only two statistics that have any real currency, goals scored and (for keepers) saves made!

Blame the moneyball approach. To me it's more interesting with regards to how much the club places value in statistics and this method, particularly with the pickup of the chap from Forest Green and to an extent Steve Walsh and the success of Kante. Players have to be measured somehow but as many have already said, statistics can be manipulated and framed in any which way anyone wants and on the flip-side there's also the immeasurable intangible assets. I try not to place too much emphasis on statistics, but I would be interested to see how much the club does.

http://statsbomb.com/2017/06/passing-percentages-are-mostly-useless-quantifying-passing-ability/

A very interesting read for those that care about such things.

Some stats need to be taken with a pinch of salt. For example the successful passes can easily be skewed when they are nothing but safe passes especially along the back line to and fro. A player receives the ball and passes back to same player = successful pass against player receives the ball, turns into attacking mode and tries a through ball that a defender just gets a toe to it and concedes a corner = failed pass.....we score from the corner!!! Too many grey areas in certain stats. Should be catergorised somehow into forward passes, backward passes, sideway passes etc for a clearer picture. I think the goals stat is the clearest we can get.

The problem with 'stats' is that most stats the average football fan has access to are...crap. Assists, goals, even things like distance run and chances created - like @Berenger said, that's top trumps, not useful statistics. Proper statistical models that take in multitude data points and large data sets can tell you absolutely enormous amounts. Like, all the variables that make a certain type of pass from a certain area into a certain area difficult, and how good is a certain player at completing those passes compared to every Premier League player who's ever tried to?

Things like mentality, grit etc - if nothing shows up as a measurable benefit, then are you just imagining stuff? If a player 'never gives up', then some sort of measurable outcome should occur for it to matter - maybe they make more possession-adjusted defensive actions when in a losing position than the average player or something. Scouting is still very important but proper stats can give you either a wider reaching overview or a more in-depth look at a player to complement traditional scouting. Or they can do something like flag up a player with really weird/interesting stats who you should definitely go have a proper look at.

Stats have become a useful tool in assessing a player, but they are simply that another tool. They are a thing, not good not bad just a thing. It is the people who rely on purely statistics to make an assessment of a player that are bad.

Stats need interpretation and not by Sky or whoever. Proper analytics is being carried out by every major team in every major sport and we rarely get exposed to these metrics.

Analytics discerns which stats are meaningful, scientifically over many years' worth of data, and inform and influence a lot of decisions regarding players/tactics/transfers etc.

Temp.

 

[Goat]

Temp.

????????

 

[Philw]

????????

Was on my phone and needed to switch to pc to write something useful. Tbh most of it's been covered now...

Unfortunately, the football (media) world doesn't use statistics like the scientific community would. They are used to provoke debate or moulded to fit an argument and are therefore pretty useless. i.e...

When people use stats like 'chances created', what constitutes a chance?
Why is 'distance covered' an important stat without the context of where the player is running or what they are actually doing?
etc.

In short, stats in football only mean what you want them to mean.

I think that the difficult thing with football stats is identifying which stats actually give you the best chance of winning matches.
I don't believe that any of the stats that sky present to us show us that picture.

The big problem is that nobody even started thinking about keeping statistics beyond goals scored until recently, and even now it's really difficult to find out really basic things, so the statistics we have don't really tell us as much as we'd like.

Statistics are fine. They're the basic input of any scientific approach to analysis of sport and the output of persistent journaling. The problem is that although statistics in sport is old, football statistics is still relatively undeveloped (or, at least developing), and even fewer know how to discuss statistics that know that they are. And there's an inherent difficulty with football statistics (at least compared to North American sports and sports like cricket, all of which are largely sequential), since it's much harder to know how to divide and which sequences are useful to compare.

The key issue is context and controlling factors. All the article does is explain why 'chances' in 'chances created' are not all the same. That's just understanding your data.

Useful statistical analysis would go further in trying to control for several factors, *given* understanding of what the data is measuring, such as: Swansea's level, players around him, his role in the team, opposition and so on.

Ultimately it comes down to 2 things imo:

1) Context and analysis - on their own, stats are just useless numbers. To make any proper use of them they need to have a context and to be properly analysed. As mentioned above the stats that most of us are exposed to in the football world and much less sophisticated and useful than those in American sports tend to be. See: Moneyball. Someone mentioned how usefull/useless Pass %/Chances Created stats are: In the NBA they have models for how 'good' a pass or chance created is - a pass that creates an open shot worth three points is worth more than one that creates a 'marked' shot worth two points etc.

2) Even the best stats mean nothing on their own. As others have mentioned, they are just one tool/part of the process. The 'eye-test' is just as important, and the two should be used in conjunction to confirm/deny the analytical findings of each other. See various examples people have posted about Ross/Siggy.

 

[SerenityNigh]

Was on my phone and needed to switch to pc to write something useful. Tbh most of it's been covered now...

Ultimately it comes down to 2 things imo:

1) Context and analysis - on their own, stats are just useless numbers. To make any proper use of them they need to have a context and to be properly analysed. As mentioned above the stats that most of us are exposed to in the football world and much less sophisticated and useful than those in American sports tend to be. See: Moneyball. Someone mentioned how usefull/useless Pass %/Chances Created stats are: In the NBA they have models for how 'good' a pass or chance created is - a pass that creates an open shot worth three points is worth more than one that creates a 'marked' shot worth two points etc.

2) Even the best stats mean nothing on their own. As others have mentioned, they are just one tool/part of the process. The 'eye-test' is just as important, and the two should be used in conjunction to confirm/deny the analytical findings of each other. See various examples people have posted about Ross/Siggy.

All data needs interpretation

 

[Philw]

All data needs interpretation

Is that like Every Good Boy Deserves .... ? (or Never Eat Shredded Wheat)

 

[SerenityNigh]

Is that like Every Good Boy Deserves .... ? (or Never Eat Shredded Wheat)

Dunno, actually I think of it like the biblical texts (1 Corinthians 14.27 specifically) on speaking in tongues--it's useless/harmful without interpretation.

 

[Woolly Blue]

@Adam-GOTTV

If you ever ask this question in a thread again, please ensure you add a poll to the thread with options something like :-

• Very important, I'm a numbers freak
• Yea, a bit like
• Away with you and your Football Manager ways
• Cheese on shots on target

You can then decide whether or not to take any notice of the stats provided by the poll.
Thanks for your future co-operation on this.

 

[Efcjake]

The only ones that matter are team wins and team goals. Individual stats are near useless in this game and fail to show sufficient context for a 90 minute match.

 

[Mikey_Fitzgerald]

I'm all about the eye test over stats tbh

 

[Groucho]

I'm all about the eye test over stats tbh

Look, I've just seen someone on twitter say to Gana "love you mate, better stats than Kante".

Now he means well, I'm not ridiculing him, but it seems to have become a thing to quote stats ad verbatim.

Like you, I'd rather see for myself.

 

[Mikey_Fitzgerald]

Look, I've just seen someone on twitter say to Gana "love you mate, better stats than Kante".

Now he means well, I'm not ridiculing him, but it seems to have become a thing to quote stats ad verbatim.

Like you, I'd rather see for myself.

I love Gana, but Kante is on his way to being the next Makalele

Kante should win the next Ballon D'or IMO

He won't, because they rarely give it to defensive players, but all being fair he'd win it hands down

He's essentially decided the destination of the Premier League title two seasons running

 

[NilesRossly]

I love Gana, but Kante is on his way to being the next Makalele

Kante should win the next Ballon D'or IMO

He won't, because they rarely give it to defensive players, but all being fair he'd win it hands down

He's essentially decided the destination of the Premier League title two seasons running

Disagree. He was a key part but Hazard/Mahrez were more important.