xG table

r00fie1

Well-known member
There is Hoppe!
There is Hoppe!
Hallelelujah!
Chrisi Wilders Red an white Army!
We`re going up!
We`re going up!
The Mighty Boro`s going up!

I knew we had to be top of something!:love:
 

Laughing

Well-known member
There is Hoppe!
There is Hoppe!
Hallelelujah!
Chrisi Wilders Red an white Army!
We`re going up!
We`re going up!
The Mighty Boro`s going up!

I knew we had to be top of something!:love:
We'll certainly finish higher than we are right now. I have a charity bet so I bloody well hope so
 

Redwurzel

Well-known member
Thats what short term stats do, last 4 and 8 games, typically. They are supposed to, quickly, capture an upturn, or indeed, a downturn in form quicker than season long stats.

Whilst it is historic, it can still be used to predict outcomes, every stat is historic and can be used to some extent or another. Oddly enough relative league position is pretty good at predicting outcomes and it is one of the few current stats that we have.

Generally, if you want to predict the outcome of games you are either a backer or a bookie and to get any sort of edge you have to focus on something other bettors don't. This won't improve your strikerate but it will improve your odds. For example lady jokeys used to make a profit in racing, just backing them blindly as most backers just assumed women would be weaker in a tight finish so tended to lower the horses relative chance in their thinking so are less likely to back it, leading to an inflated pice.
I agree xG is just one form of (historic) stats. The stats on this thread are going back 10 games. I know a betting site that looks at last 4 home or away games and last 2 games to show form. I tend to look at last 4 home or away games, 2 games is a bit too short for me, It then sets that against the opponents stats e.g say Coventry v Boro its got 30.0 for Coventry and 35.50 rating for Boro that gives a a negative rating of 5.50 for the match. A negative rating means the away side is more likely to win. in this case its a small strength rating - an average rating is around 10.0 with a range from 0 to 40.0. If you looked at our away results there is no way Boro should be favs (lost 3, drew 1, won none) anyone would make Coventry clear favs say around 43% Coventry 29% draw 28% Boro - In betting odds 5/4 Coventry 5/2 draw 13/5 Boro. These will not be the bookie odds as the bookies go more on performance stats (Boro touch favs) not result stats. This game is difficult for the sats because Coventry have not played at home?

Ref betting edge - my edge was "George Boateng" if he returned from injury I backed us and vice versa. Bookies often failed to pick this edge up.
 

Laughing

Well-known member
I agree xG is just one form of (historic) stats. The stats on this thread are going back 10 games. I know a betting site that looks at last 4 home or away games and last 2 games to show form. I tend to look at last 4 home or away games, 2 games is a bit too short for me, It then sets that against the opponents stats e.g say Coventry v Boro its got 30.0 for Coventry and 35.50 rating for Boro that gives a a negative rating of 5.50 for the match. A negative rating means the away side is more likely to win. in this case its a small strength rating - an average rating is around 10.0 with a range from 0 to 40.0. If you looked at our away results there is no way Boro should be favs (lost 3, drew 1, won none) anyone would make Coventry clear favs say around 43% Coventry 29% draw 28% Boro - In betting odds 5/4 Coventry 5/2 draw 13/5 Boro. These will not be the bookie odds as the bookies go more on performance stats (Boro touch favs) not result stats. This game is difficult for the sats because Coventry have not played at home?

Ref betting edge - my edge was "George Boateng" if he returned from injury I backed us and vice versa. Bookies often failed to pick this edge up.
I use 4 and 8 games in my deepnet algorithm to pick up trends in a, reasonably short time. I don't use any player information. I would love to but that would give me too many data points.

I even looked at identifying "key" players for each side and using just 2 or 3 sets of stats per player. It would still give me 6 new data points per game per stat.

My algorithm gives a Phi coefficient for each metric I look at, over and under 2.5 and home, away, draw. It's between 0-1. I find the whole process fascinating.
 

Rofesleg

Well-known member
I use 4 and 8 games in my deepnet algorithm to pick up trends in a, reasonably short time. I don't use any player information. I would love to but that would give me too many data points.

I even looked at identifying "key" players for each side and using just 2 or 3 sets of stats per player. It would still give me 6 new data points per game per stat.

My algorithm gives a Phi coefficient for each metric I look at, over and under 2.5 and home, away, draw. It's between 0-1. I find the whole process fascinating.
I'll be honest Laughing, I prefer looking at xG stats than a 2.5 Phi coefficient between 0-1.

Anyway, I've written an algorithm and I must be doing something wrong. It's not fascinating, it's sh1t.
For the last round of Champs League fixtures it predicted the following result:
Real Madrid - 1
Surreal Madrid - Fish
 

BoroPhil

Well-known member
XG is a very broadtool. but it is used to some extent by bookmakers to help predict the future. Hence our odds are relevely short to finish top 6 even though we are in the bottom 3. Bookies are more times right than wrong, as they say have you ever seen a poor bookmaker?

Last season we struggled up the end of November but still finished 7th.

When Leicester won the Premier League is there anybody on here that believed they were clearly the best team in the league that season? (sometimes things fall your way).

True and someone mentioned Brighton earlier - they were hugely underpeforming vs xG a couple of years ago and the bookies regularly had their match odds better than you would expect for a team in their position. So they are clearly using xG (and other data) to set their prices.
If you think xG is rubbish then lump on Boro's opposition - you will get huge value at the moment.
 

Redwurzel

Well-known member
Laughing - look @ stats when a centre half captain is missing (if you haven't already) Another interesting analysis is regular goalie missing.

Would up mind giving me some feedback either on the main forum or private mail - cheers (y)
 

Redwurzel

Well-known member
True and someone mentioned Brighton earlier - they were hugely underpeforming vs xG a couple of years ago and the bookies regularly had their match odds better than you would expect for a team in their position. So they are clearly using xG (and other data) to set their prices.
If you think xG is rubbish then lump on Boro's opposition - you will get huge value at the moment.
Brighton have turned - sometimes it can take a while to regress to the mean say 12 months - the bookies will keep focused on the mean.

Its like betting on tails @ 6/5 (and been excited on getting those odds) and you then there are 5 heads in a row in a coin toss. Its easier to think the chances for heads is much better than evens and tails is bad value, but its not. The tails will start appearing on a normal coin.

A good bookie will keep offering 5/6 on heads and 5/6 on tails.

In this example I believe Boro are similar to tails. Time will tell if our perfromance stats turn into more positive results.

I notive we are now favs with the bookies at Coventry despite having the worst away record in the league.
 

Laughing

Well-known member
Laughing - look @ stats when a centre half captain is missing (if you haven't already) Another interesting analysis is regular goalie missing.

Would up mind giving me some feedback either on the main forum or private mail - cheers (y)
Sure I'll send you a pm
 
Top