So, xg and xga. Questions:
- if every team had converted/conceded goals as expected for the nature of chances that arose, xg and xga tell us the expected match results, correct? So do we read the below scores as what should on average have happened (rounded)?
Liverpool 2 - 1 Norwich
West Ham 1 - 3 Man City
Bournemouth 1 - 2 Sheff Utd
Burnley 1 - 1 Southampton
Palace 1 - 1 Everton
Watford 1 - 1 Brighton
Spurs 3 - 1 Aston Villa
Newcastle 0 - 1 Arsenal
Leicester 1 - 1 Wolves
Man Utd 2 - 1 Chelsea
- are chances personalised for teams and players, or are they based on the general percentage likelihood of a player in a particular position converting a chance, i.e. given the MC performance we would normally expect THEM, or a team in general, to score 3? Did MC outdo their own norm, or just the Prem League norm? And is that 3 goals in any match with those chances, or against West Ham in particular with those chances?
- should I be reading that MU, MC, Brighton and Burnley flattered? Got a bit lucky? Will their xg now increase based on their goals scored at the weekend?
And yes, I know this is only 1 set of fixtures and a small data set, just trying to ensure I fully get what is going on here.
xg
- raoul
- Dumbledore
- Posts: 5050
- Joined: 13 Oct 2005, 18:32
- FS Record: Tied for OR 1st worldwide in FPL in GW5 in 21/22 (only to then finish 1m), and ranked 1st in TFF late in season 21/22 (eventual finish 95th). Won an old game called Football Fantastic. Best Fanteam finish 117th and cashed £150. Best FPL finish 2.4K.
- FPL:
- Billy Bongo
- FISO Knight
- Posts: 12000
- Joined: 21 Jul 2013, 22:18
- FPL:
Re: xg
You dont need stats to show those 4 teams were flattered a little bit by the score lines, MOTD was enough tbh
In all 4 games awful defending was the issue rather than exceptional poaching with one exception , Kane ( not a great footballer but a world class finisher )
It's why the stats can confuse, Kane is a great example. Everyone moans he does nothing but he then takes the chances like he does, simply world class at being Harry Kane
XG for Harry Kane should be banned
Sent from my SM-G975F using Tapatalk
In all 4 games awful defending was the issue rather than exceptional poaching with one exception , Kane ( not a great footballer but a world class finisher )
It's why the stats can confuse, Kane is a great example. Everyone moans he does nothing but he then takes the chances like he does, simply world class at being Harry Kane
XG for Harry Kane should be banned
Sent from my SM-G975F using Tapatalk
- TheRumourMill
- Grumpy Old Man
- Posts: 1432
- Joined: 03 Dec 2017, 23:15
- FS Record: FPL: Similar to Everton - haven't cracked the top 10k glass ceiling yet, neither have I had and sub 500k shockers either!
International tournaments: Better, 11k in WC 2018 and top 1000 way back at Euro 2012! - FPL:
Re: xg
To answer in order:raoul wrote: ↑13 Aug 2019, 10:52 So, xg and xga. Questions:
- if every team had converted/conceded goals as expected for the nature of chances that arose, xg and xga tell us the expected match results, correct? So do we read the below scores as what should on average have happened (rounded)?
Liverpool 2 - 1 Norwich
West Ham 1 - 3 Man City
Bournemouth 1 - 2 Sheff Utd
Burnley 1 - 1 Southampton
Palace 1 - 1 Everton
Watford 1 - 1 Brighton
Spurs 3 - 1 Aston Villa
Newcastle 0 - 1 Arsenal
Leicester 1 - 1 Wolves
Man Utd 2 - 1 Chelsea
- are chances personalised for teams and players, or are they based on the general percentage likelihood of a player in a particular position converting a chance, i.e. given the MC performance we would normally expect THEM, or a team in general, to score 3? Did MC outdo their own norm, or just the Prem League norm? And is that 3 goals in any match with those chances, or against West Ham in particular with those chances?
- should I be reading that MU, MC, Brighton and Burnley flattered? Got a bit lucky? Will their xg now increase based on their goals scored at the weekend?
And yes, I know this is only 1 set of fixtures and a small data set, just trying to ensure I fully get what is going on here.
Yes that's broardly it. Of course rounding brings its own error margin, say a match where one side records xg of 0.51 and the other 1.49, you could say it "should" have been 1-1 due to rounding, but in reality one side was almost a goal better in terms of xg. You also have to be careful of phases of play. For example on Saturday Bournemouth's xg included 2 high xg chances which were responsible for the bulk of their xg. The problem was one of them was Wilsons point blank shot he hit at the keeper, and the other was mephams goal on the rebound from that Wilson shot. So in reality Bournemouth created even less than initially appears from just looking at the xg numbers. This is why its important to drill down a bit further into the stats, or watch MOTD or the games themselves to see how the chances arose.
Chances are not personalised. The numbers are for an average player in an average league playing for an average team etc. its from thousands of data points across many years and many leagues and players. So if you think a player is an above average finisher, or below average, apply your own judgement to the raw numbers.
MU, MC, Bri, Bur were flattered, yes. the way their xg is calculated wont change though. What will slightly change is the bookies odds on them scoring in the future, or predicted points models like in the FFScout RMT members area, or fplreview.com, patricians model, etc.
- blahblah
- FISO Viscount
- Posts: 108823
- Joined: 13 Oct 2005, 18:46
- Location: .. he thinks that he knows something which he doesn't, whereas I am quite concious of my ignorance.
- FPL:
Re: xg
Yep it is all about the average so think Kun et al vs Benteke et al and also Alisson vs Adrian. I touch on the latter in my RMT ramble as to why WC'ing to Salah has a certain appeal.
- raoul
- Dumbledore
- Posts: 5050
- Joined: 13 Oct 2005, 18:32
- FS Record: Tied for OR 1st worldwide in FPL in GW5 in 21/22 (only to then finish 1m), and ranked 1st in TFF late in season 21/22 (eventual finish 95th). Won an old game called Football Fantastic. Best Fanteam finish 117th and cashed £150. Best FPL finish 2.4K.
- FPL:
Re: xg
Thanks for that TRM, very useful.TheRumourMill wrote: ↑13 Aug 2019, 11:17To answer in order:raoul wrote: ↑13 Aug 2019, 10:52 So, xg and xga. Questions:
- if every team had converted/conceded goals as expected for the nature of chances that arose, xg and xga tell us the expected match results, correct? So do we read the below scores as what should on average have happened (rounded)?
Liverpool 2 - 1 Norwich
West Ham 1 - 3 Man City
Bournemouth 1 - 2 Sheff Utd
Burnley 1 - 1 Southampton
Palace 1 - 1 Everton
Watford 1 - 1 Brighton
Spurs 3 - 1 Aston Villa
Newcastle 0 - 1 Arsenal
Leicester 1 - 1 Wolves
Man Utd 2 - 1 Chelsea
- are chances personalised for teams and players, or are they based on the general percentage likelihood of a player in a particular position converting a chance, i.e. given the MC performance we would normally expect THEM, or a team in general, to score 3? Did MC outdo their own norm, or just the Prem League norm? And is that 3 goals in any match with those chances, or against West Ham in particular with those chances?
- should I be reading that MU, MC, Brighton and Burnley flattered? Got a bit lucky? Will their xg now increase based on their goals scored at the weekend?
And yes, I know this is only 1 set of fixtures and a small data set, just trying to ensure I fully get what is going on here.
Yes that's broardly it. Of course rounding brings its own error margin, say a match where one side records xg of 0.51 and the other 1.49, you could say it "should" have been 1-1 due to rounding, but in reality one side was almost a goal better in terms of xg. You also have to be careful of phases of play. For example on Saturday Bournemouth's xg included 2 high xg chances which were responsible for the bulk of their xg. The problem was one of them was Wilsons point blank shot he hit at the keeper, and the other was mephams goal on the rebound from that Wilson shot. So in reality Bournemouth created even less than initially appears from just looking at the xg numbers. This is why its important to drill down a bit further into the stats, or watch MOTD or the games themselves to see how the chances arose.
Chances are not personalised. The numbers are for an average player in an average league playing for an average team etc. its from thousands of data points across many years and many leagues and players. So if you think a player is an above average finisher, or below average, apply your own judgement to the raw numbers.
MU, MC, Bri, Bur were flattered, yes. the way their xg is calculated wont change though. What will slightly change is the bookies odds on them scoring in the future, or predicted points models like in the FFScout RMT members area, or fplreview.com, patricians model, etc.
View Latest: 1 Day View Your posts