Stats & Analytics

[craig 5,199]

47 minutes ago, DMAA said:

Morelos leading in xG, but his actual goals trails his xG quite significantly, maybe consistent with a dip in the form for most of the season. 

He's starting to look much like the player we have been used to.  Seems to be playing with a smile on his face again.  Fingers crossed it contnues

[DMAA 3,658]

2 hours ago, craig said:

He's starting to look much like the player we have been used to.  Seems to be playing with a smile on his face again.  Fingers crossed it contnues

As I said in the match thread, I have been his biggest critic for quite a while (which isn’t saying much to be fair), and I thought he was outstanding against Benfica. He was a joy to watch. It would be a huge boost to us if he can carry on like that given the way everyone else is playing. 

[Tannochsidebear 2,429]

8 hours ago, DMAA said:

The vast majority of stats are just basic facts.
 

xG is a metric which is calculated using a strictly defined and consistently applied methodology. 
 

It is a fact, you just have to understand the methodology and process to understand what it is a measure of. The number isn’t random or subjective, it provides an accurate measure of the frequency of a compilation of facts. 

I appreciate the explanation but the “fact” remains that whether something is a good chance or a half chance, and whether a team’s play should result in 5 goals or 6 according to xg is based entirely on opinion, and the beauty of our beautiful game is that it is entirely unpredictable and made up of so many parts. Methodology is not facts, it is opinion. 
 

I would perhaps be more swayed if they took a game, say the Hamilton game, and showed us exactly how they got to the xg number, what they counted, what score they gave it, and what they didn’t count. My guess is there would be considerable argument to be had in how they got to that number. 
 

Of course we can all agree that the Hamilton xg was exactly 0.0! No debate to be had there!

[DMAA 3,658]

32 minutes ago, Tannochsidebear said:

I appreciate the explanation but the “fact” remains that whether something is a good chance or a half chance, and whether a team’s play should result in 5 goals or 6 according to xg is based entirely on opinion, and the beauty of our beautiful game is that it is entirely unpredictable and made up of so many parts. Methodology is not facts, it is opinion. 
 

I would perhaps be more swayed if they took a game, say the Hamilton game, and showed us exactly how they got to the xg number, what they counted, what score they gave it, and what they didn’t count. My guess is there would be considerable argument to be had in how they got to that number. 
 

Of course we can all agree that the Hamilton xg was exactly 0.0! No debate to be had there!

This is what I’m trying to explain, it isn’t done like that. It’s all pre-defined. For example, an attempt on goal from particular coordinates on the pitch will be assigned a pre-defined xG score. The pre-defined score is based on historical data, the percentage of shots from that position that have historically resulted in a goal. Opinion doesn’t come into it, it’s pretty much at the stage where a computer can calculate it, as I understand it. They tweak the models to take into account extenuating circumstances, and the models are increasingly accurate in predicting the score based on the attempts on goal made in a game. 
 

I should say I am not an expert in this and there will be bears out there, including the guys who run these Twitter accounts, who are much better qualified than me to explain how they work. But the key point is that the stats are generated from an objective statistical model which is informed by a mountain of historical data. 

Edited November 11, 2020 by DMAA

[Rousseau 10,823]

9 hours ago, Tannochsidebear said:

I appreciate the explanation but the “fact” remains that whether something is a good chance or a half chance, and whether a team’s play should result in 5 goals or 6 according to xg is based entirely on opinion, and the beauty of our beautiful game is that it is entirely unpredictable and made up of so many parts. Methodology is not facts, it is opinion. 
 

I would perhaps be more swayed if they took a game, say the Hamilton game, and showed us exactly how they got to the xg number, what they counted, what score they gave it, and what they didn’t count. My guess is there would be considerable argument to be had in how they got to that number. 
 

Of course we can all agree that the Hamilton xg was exactly 0.0! No debate to be had there!

You're not getting it.

 

The 'fact' of "whether something is a good chance or a half chance", to put it your way, is calculated through statistical modelling. 

 

Opta takes 300,000 shots (what's that 20,000 games, assuming 15 shots per game?), looking a variety of variables (position of the shot, angle, etc. - quite a few) and comparing how many times that shot was actually scored, to get the probability of that type of chance being scored.

 

So, their model then takes all that data and compares it to a chance that one of our players had on Sunday, to get the probability of that chance being scored. That's what would be 'expected' to score based on 300,000 shots in 20,000 games.

 

To keep it simple, penalties have an xG of 0.76, meaning they are scored 76% of the time. That's a fact. Better players will exceed that, because they're better players: the xG describes that fact.  

 

It takes nothing away from "the beautiful game". It's giving us more facts with which to describe the game. 

[Gonzo79 15,299]

8 minutes ago, Rousseau said:

It takes nothing away from "the beautiful game".

That's subjective.  

 

Breaking things down into numbers and diagrams can detract from the overall enjoyment but that depends on the viewer/supporter/fan/critic's point of view and goes for anything artistic/creative (which football can be when it's at its best).  

 

Each to his own.

[Rousseau 10,823]

The hostility to this is embarrassing. It's just ignorance. (Not necessarily here, just in general.)

 

Those that criticise are the same that'll declaim, "Oh-- he should have scored there! That's a sitter!".

 

Yes, the xG backs you up on that. It just describes what you intuitively think.

 

What's wrong with that?

 

No wonder this country achieves little of any relevance in football: we're still in the dark ages, held back by 'received wisdom', as other countries have modernised and developed and shot past us. 

 

Rant over.  

 

Apologies in advance.

[compo 7,226]

And whereabouts can you get a copy of the stats for a game so you can examine them ,

[Rousseau 10,823]

3 minutes ago, Gonzo79 said:

That's subjective.  

 

Breaking things down into numbers and diagrams can detract from the overall enjoyment but that depends on the viewer/supporter/fan/critic's point of view and goes for anything artistic/creative (which football can be when it's at its best).  

 

Each to his own.

"The Beautiful game" is still there. We all still get caught up in the emotion of it. 

 

But then some of us, afterwards, want to look beyond that to see that stats of the game; to see how it really was. 

 

I was bouncing on Sunday (and sick on the Thursday!). Then on Monday I wanted to look at the stats to see how dominant, or not, we actually were. In the same way people re-watch games, to catch things they missed. 

[Rousseau 10,823]

1 minute ago, compo said:

And whereabouts can you get a copy of the stats for a game so you can examine them ,

For us fans, it's mostly through Twitter etc. -- and the few @DMAA posts here. 

 

The data packages that clubs and pundits use are bloody expensive!