Match Report: City silence transfer clamour with timely reminder of grit over goals

[BTRFTG]

43 minutes ago, Davefevs said:

At the end of the day you seem upset that my view of the game was different to yours in that I don’t think Reading dominated.  That’s fine. All about opinions.  

Save I didn't say Reading dominated.

I also asked a relatively simple question in can somebody explain what are these so-called stats mean,  how and who defines them? At which you appear to have gone off on a rant yet fail to answer the basic questions asked. What in heavens name is an 'expected goal', what isn't and what are the differentiators? From the graph which value relates to which event? In the case of Swift's shot surely it's either an expected goal or not, not some part goal?

[ExiledAjax]

1 hour ago, BTRFTG said:

Save I didn't say Reading dominated.

I also asked a relatively simple question in can somebody explain what are these so-called stats mean,  how and who defines them? At which you appear to have gone off on a rant yet fail to answer the basic questions asked. What in heavens name is an 'expected goal', what isn't and what are the differentiators? From the graph which value relates to which event? In the case of Swift's shot surely it's either an expected goal or not, not some part goal?

Mate, there is so much out there to explain to you what xG is. You have access to the internet and are literate so if you really care you can find out yourself. However to assist, and using the same site that @Davefevs used:

"...each type of shot has a goal value assigned to it based on how often shots of that type are scored. For example, if a certain type of shot went in 1 time in every 10, it would have an expected goal value of 0.1). If you add up all these values for each match, you’ll get totals for how many goals a team “should” have scored and conceded, assuming everyone is equally good at converting and saving shots over the long run. That last bit is obviously not true, but shooting and saving are currently difficult skills to quantify and vary quite a bit, and this approach has been shown – and is widely considered – to be an upgrade on simply counting all shots equally."

Genuinely, the guy that compiles these stats is generally more accurate at predicting the finishing position of teams than the bookies are. Look at his results over the past few seasons if you need proof.

Another site clarifies that generally you can expect an xG value to consider:

• Shot location. From where on the pitch was the shot attempted?
• Shot type. Was it taken with the foot or the head? Was the shot a direct free kick?
• Assist type. Was it assisted by a cross, a through-ball or a regular pass? Was the shot taken off a rebound?
• Speed of attack. How much ground did the attacking team cover, and how quickly, before attempting the shot?
• Long balls. Did the attacking move include any very long passes?
• Dribble. Did the shot immediately follow a successful dribble by the attacking player? Did this dribble beat the keeper or an outfield defender?
• Set play. Was the shot attempted off a set play or from open play?

More sophisticated (and expensive) xG data will consider more of these factors, but even the most basic will take the first three and the last one into account. The most sophisticated nowadays also do factor in the average finishing ability of the individual shot-taker as well.

It's true as well that different companies and individuals will generate different xG values for the same match. This doesn't instantly invalidate the entire concept of xG - it just means you need to be careful. I'm currently tracking all of our xG results for its season - I take the values that InfoGol and Experimental361 give and average them out (I'd love to have more sources but can't find more free and reliable sites).

https://www.amazon.co.uk/Football-Hackers-Science-Data-Revolution/dp/1788702050 is an easy to read book explaining the rise of data use in football. It includes a very good explanation of the origin, uses, and limitations of xG. It also mentions Bristol City within the first 10 pages so it gets the seal of approval for that alone.

 

[Davefevs]

3 hours ago, BTRFTG said:

Save I didn't say Reading dominated.

I also asked a relatively simple question in can somebody explain what are these so-called stats mean,  how and who defines them? At which you appear to have gone off on a rant yet fail to answer the basic questions asked. What in heavens name is an 'expected goal', what isn't and what are the differentiators? From the graph which value relates to which event? In the case of Swift's shot surely it's either an expected goal or not, not some part goal?

I assumed with your posting history you’d seen enough mentions on OTIB re xG to have a vague understanding of its use. Thanks to @ExiledAjax for taking the time to explain it.  Apologies if that wasn’t the case.

I only put the graph up to show the flow of chances throughout the game.  I don’t find Experimental361’s xG metrics to be very good, but I do like his graphical timelines, as each vertical line shows a shot.  The Diedhiou missed header and Pato’s goal had about the same likelihood of scoring, approx 1 in 10 (0.1 vertical line on the chart).  A lot of people will say Fam’s was a sitter, but the stats in this case disprove this, from ‘000s of similar type headers from that rough position.  Opta data is much better...but costs a bomb.

1 hour ago, ExiledAjax said:

Mate, there is so much out there to explain to you what xG is. You have access to the internet and are literate so if you really care you can find out yourself. However to assist, and using the same site that @Davefevs used:

"...each type of shot has a goal value assigned to it based on how often shots of that type are scored. For example, if a certain type of shot went in 1 time in every 10, it would have an expected goal value of 0.1). If you add up all these values for each match, you’ll get totals for how many goals a team “should” have scored and conceded, assuming everyone is equally good at converting and saving shots over the long run. That last bit is obviously not true, but shooting and saving are currently difficult skills to quantify and vary quite a bit, and this approach has been shown – and is widely considered – to be an upgrade on simply counting all shots equally."

Genuinely, the guy that compiles these stats is generally more accurate at predicting the finishing position of teams than the bookies are. Look at his results over the past few seasons if you need proof.

Another site clarifies that generally you can expect an xG value to consider:

• Shot location. From where on the pitch was the shot attempted?
• Shot type. Was it taken with the foot or the head? Was the shot a direct free kick?
• Assist type. Was it assisted by a cross, a through-ball or a regular pass? Was the shot taken off a rebound?
• Speed of attack. How much ground did the attacking team cover, and how quickly, before attempting the shot?
• Long balls. Did the attacking move include any very long passes?
• Dribble. Did the shot immediately follow a successful dribble by the attacking player? Did this dribble beat the keeper or an outfield defender?
• Set play. Was the shot attempted off a set play or from open play?

More sophisticated (and expensive) xG data will consider more of these factors, but even the most basic will take the first three and the last one into account. The most sophisticated nowadays also do factor in the average finishing ability of the individual shot-taker as well.

It's true as well that different companies and individuals will generate different xG values for the same match. This doesn't instantly invalidate the entire concept of xG - it just means you need to be careful. I'm currently tracking all of our xG results for its season - I take the values that InfoGol and Experimental361 give and average them out (I'd love to have more sources but can't find more free and reliable sites).

https://www.amazon.co.uk/Football-Hackers-Science-Data-Revolution/dp/1788702050 is an easy to read book explaining the rise of data use in football. It includes a very good explanation of the origin, uses, and limitations of xG. It also mentions Bristol City within the first 10 pages so it gets the seal of approval for that alone.

 

Ta.  

[BTRFTG]

2 hours ago, ExiledAjax said:

Mate, there is so much out there to explain to you what xG is.@Davefevs

 

Thanks for that explanation and that's pretty much the same 'no sh*t Sherlock' explanation given by the UK's leading football mathematician at the year's  Royal Institution Xmas Lectures:

The more you shoot the more likely you are to score;

The more central you are to goal the more likely you are to score;

The closer to goal the more likely to score;

The greater seperation of striker to opponent the more likely to score et al.

So whilst mathematicians are able to create interesting algorithms from quadrillion data points showing players where statistically they should move and what they should do to maximize their talents, in and of themselves they deliver and portend  less than that worked out by many a player unable to work out the square root of 4.

I hazard last night many of the dangerous crosses that went a' begging probably didn't register on any of these indices though were statistically more dangerous (to a talented side, not Reading) than Swift's long-ranger that doubtless did. It's that out of context assessment that renders stats meaningless when applied to specific events. Identical players, identical situation, one scores, one doesn't - why? The maths won't help you. Long term stats predict trends but does not describe or predict what has or will happen at a specific event and the dangerous thing is the reporting of these stats as a reflection of what happened is similarly meaningless, rather less informative than somebody saying:"we should have had 5." For example, I'd love to see maths explaining the last 15 minutes of our away match at Coventry as without context it wouldn't make sense.

I also wouldn't be so disparaging at the weapons bookies muster. I've heard and seen first hand it's they who employ many of the brightest mathematicians and statisticians and the range of detailed analysis they output is astonishing and often predicts those results the rest of us might think out of the blue. But that takes immense levels of big data, not all derived from on-field action, combined with good old fashioned inside knowledge and - chance. That's how they win and is a million miles from the meaningless 'expected goals' stats now becoming so popular and reported as gospel. For example, in the bookies case I've seen comprehensive data of tallies of goals that should have been conceded but weren't and interestingly they don't necessary tally with the expected goals numbers of the opposition. They're different and in that difference lies the tale and context of most matches.

 

[Davefevs]

4 minutes ago, BTRFTG said:

Thanks for that explanation and that's pretty much the same 'no sh*t Sherlock' explanation given by the UK's leading football mathematician at the year's  Royal Institution Xmas Lectures:

The more you shoot the more likely you are to score;

The more central you are to goal the more likely you are to score;

The closer to goal the more likely to score;

The greater seperation of striker to opponent the more likely to score et al.

So whilst mathematicians are able to create interesting algorithms from quadrillion data points showing players where statistically they should move and what they should do to maximize their talents, in and of themselves they deliver and portend  less than that worked out by many a player unable to work out the square root of 4.

I hazard last night many of the dangerous crosses they went a' begging probably didn't register on any of these indices though were statistically more dangerous (to a talented side, not Reading) than Swift's long-ranger that doubtless did. It's that out of context assessment that renders stats meaningless when applied to specific events. Identical players, identical situation, one scores, one doesn't - why? The maths won't help you. Long term stats predict trends but does not describe or predict what has or will happen at a specific event and the dangerous thing is the reporting of these stats as a reflection of what happened is similarly meaningless, rather less informative than somebody saying:"we should have had 5." For example, I'd love to see maths explaining the last 15 minutes of our away match at Coventry as without context it wouldn't make sense.

I also wouldn't be so disparaging at the weapons bookies muster. I've heard and seen first hand it's they who employ many of the brightest mathematicians and statisticians and the range of detailed analysis they output is astonishing and often predicts those results the rest of us might think out of the blue. But that takes immense levels of big data, not all derived from on-field action, combined with good old fashioned inside knowledge and - chance. That's how they win and is a million miles from the meaningless 'expected goals' stats now becoming so popular and reported as gospel. For example, in the bookies case I've seen comprehensive data of tallies of goals that should have been conceded but weren't and interestingly they don't necessary tally with the expected goals numbers if the opposition. They're different and in that difference lies the tale and context of most matches.

 

xG needs to carry a health warning.

You are right about data (which is why I’m surprised you weren’t up to speed on xG), in that in the world of football it lacks context.

I just posted something on the Ratings thread re Obita’s crossing.  It’s still pretty one dimensional, as what our eyes might tell us is one of those unsuccessful crosses, might’ve been because:

- shit striker

- Baker getting a glancing header taking it off the striker’s head

etc

etc.

The problem is people are taking things like xG to justify a single result.  According to the pic I initially put up, Reading win 1.4-1.1 which we know is rubbish.  We are seeing League tables based on xG and City are bottom according to Experimental361.

I try to use the data / pics to back up what I’ve seen with my eyes or help explain something that is not easy to describe in words. 

[BTRFTG]

14 minutes ago, Davefevs said:

xG and City are bottom according to Experimental361.

Aggregation of which is similarly meaningless. If aggregation had meaning after a few seasons we should have scored a similar number of goals to that predicted, though seems to me most games see fewer goals than 'expected' and few where more goals are scored than expected. What's the significance level and after how many games might the numbers tally? They won't for good reason the huge variables between matches (should they be considered discrete events.) We can all think of player combos that work/ don't work together so assuming players know what to do and where to go, why is that? Derek Hales being the prime example. Paired with Micky Flanagan (punches aside) he was a one man goal machine.  Anybody else, cow's backside with a banjo merchant. 

[Davefevs]

4 minutes ago, BTRFTG said:

Aggregation of which is similarly meaningless. If aggregation had meaning after a few seasons we should have scored a similar number of goals to that predicted, though seems to me most games see fewer goals than 'expected' and few where more goals are scored than expected. What's the significance level and after how many games might the numbers tally? They won't for good reason the huge variables between matches (should they be considered discrete events.) We can all think of player combos that work/ don't work together so assuming players know what to do and where to go, why is that? Derek Hales being the prime example. Paired with Micky Flanagan (punches aside) he was a one man goal machine.  Anybody else, cow's backside with a banjo merchant. 

 

[Hampshire Red]

Makes I laugh. If we don't win the same olds have a massive list of Lj or City complaints and their 'fantasy manager' nonsense 

When we win well, away to a good side the same people bring out irrelevant stat crap. 

In May, on one stat from the Madstad matters: 0-1

So for any of those present to witness another away win (its hard being a homer only) please compliment Oles report with other praise on City. This forum doesn't have to be a moaners site

[bobbyhutchscurlymullet]

8 hours ago, Alex_BCFC said:

Not a fan of Watkins but even I would consider giving him a game instead of Weimann who looks way off what he can be.

I believe Watkins is injured at the moment. Out for a while I think I recall Johnson mentioning a few weeks ago