Betfair football trading – More on Expected Goals – xG

Betfair football trading – More on Expected Goals – xG

so I did a video last week about
expected goals XG and how you could use it when you’re trading on Betfair and I
sort of skirted over the top of the topic and had a few comments about could
I do another video on XG so that’s what I’m gonna do in this video please like
and comment on the video below that will allow me to produce better quality
videos and more of them in the future yeah so the first thing is you know
thank you for the likes and comments and I know I always ask for them as I just
have but it’s important because when you do that it gives me some feedback and
tells me stuff that you’re interested in so the more likes I get on the video and
then the more of those videos I’ll do and you know I’ll obviously put a few
other videos in and around other content but it does give me a good steer in
terms of what videos you’re interested in and I read all of the comments and
when I look at videos that have been popular then I’ll tend to produce more
though so please keep on liking commenting and all of those things
because that is really helpful but yeah after doing the XG video last week there
were quite a few comments I got on that so I thought I’d follow up with a little
bit more depth in terms of what XG is how it can help you with your trading on
Betfair and in your football trading in general and when in when perhaps you
shouldn’t use it and some of the good things about it some of the bad things
about it so one of the comments I received was you know how accurate is it
so what I actually do is I went off and looked at all of the stats that we have
had from the last match that we that we lost looked at all the way back over the
last season and beyond and when you look at the number of goals that were likely
to be scored then the number of goals on average per match more or less match
what XG has been forecasting now it’s important to understand the HD models do
vary between different companies and different analysts and so on they may
have different weightings slightly based upon a number of things but more or less
there was almost an exact match so if we look at the difference between in goals
between the home and away teams it was 0.39
over the period that we were analyzing and if you look at XG it was 0.37 so you
can see it’s very very close and of course you know goals do does not equal
actually XG is the percentage chance of scoring from a certain position whereas
a goal is that that shot actually went in but to highlight a couple of little
things for you you know you can look at the matches that took place last night
and you can see a difference in terms of the way that you would use ex-chief
within a match itself so I’m just checking where I go the back again so if
you look at the matches from last night we had a couple of interesting
situations that developed that would skew the way that you interpret XG so
Wayne Rooney scored from his own half last night now if you replay and have a
look at that goal which I’m unable to do on this video you’ll have to find it and
replay it what you’ll notice is that the goal keeper goes to clear the ball makes
it but if a hash of it the ball ends up with Rooney in his own half and
admittedly it was a great strike but he did shoot into an empty net so if you
look at a lot of XG models on that particular match they were ranking the
chance of scoring from that distance as virtually 0 0.01 was what’s the model
that I’ve got threw up but of course that didn’t take account of the fact
that there was no goalkeeper in the goal and so of course you’re gonna bounce to
be much more likely to score if there is no goal keeper in the goal that’s that
thing that goes without saying really doesn’t it so a lot of XG models were a
bit skewed last night because of that so I think that sort of shows that it’s
useful to have statistics but you shouldn’t necessarily blindly follow
them because when you look at the output from a lot of the models that came out
the following morning very few people had adjusted or accounted for the fact
that there was no goalkeeper in the goal at the moment which the ball would
strike now I couldn’t score from the halfway line and I’ll own my own half
but a professional footballer was probably quite likely to get the ball
quite close or on target at that particular point and the shot itself was
pretty good it was hit quite hard very firmly and it was towards the middle of
the goal and it was lofty over the players so was was the chance
high of scoring a goal not particularly but it certainly wasn’t as low as 0.01
so you know I think that that’s why it’s helpful to watch a match as well because
you can get a feel for it typically the way I use statistics is you even if
you’re just watching a horse race pre off most definitely in play or a tennis
match in play or a football match I tend to find that you do get a much better
feel for it but having a feel for it isn’t good enough on its own in my book
what you need to do is you actually need to watch the sports that you’re trading
but then you need to be able to quantify it as well or at least make an attempt
to quantify it so if you’re watching a tennis match you can see the momentum
swing backwards and forwards but when you’ve got the stats to back up what’s
going on that will help you because very very often what you’ll find is that you
know you’ll get a certain feeling for something but in fact the stats don’t
back that up so Manchester City were playing
Southampton last night Sampson played pretty well I have to say
even though it’s gonna be a biased opinion but if you actually look at the
number of shots that Manchester United had they had 26 shots which was an awful
lot of shots and twelve of them are on target and yet the score ended up to one
and so home son only had two shots on target and they only had six shots now
if you read those stats they said this goes back to where we started last week
which is if you look at the total number of shots that can give you a guide but
it doesn’t give you the detail so this is a great example when Rooney was a
good example of where X G sort of got it slightly wrong but this is an example of
where X G gets it absolutely right because if you look at the number of
shots that they had in the number of shots that were on target there’s a very
high number so perhaps you’d expect Man City to absolutely whitewash that but if
you’re watching the match you could see that that wasn’t really taking place and
that in fact said Hampton were playing pretty well now I didn’t trade it
because I can’t trade my own team so impossible for me to get involved in
that but from I could see from the setup that so Hampton had that they were
stopping a lot of dangerous text format City so the team was set up to do
that man said he had a lot of possession at 74% per possession they were creating
opportunities but Sir Hampton were limiting those opportunities to shots
that were outside of the box or from unfavorable positions so when you
actually look at the XG of the match you could actually see if you plotted it
from the number of shots taken and the chances you could see both of them going
up step by step almost together now unfortunately for sir Hampton Stirling
got a very late goal which was totally undeserved I must say but you could
actually see the progression through the match was not as skewed as the top level
stats would have suggested so the top level stats were suggesting that man
said you were all over Sampson which they sort of work but it was a tactical
decision on behalf of Sir Hampton to play the game that sort of way and they
were looking for quality breaks and so Hampton actually pretty much matched the
XG of Man City last night because when they did get those opportunities there
were very high quality opportunities and in fact they hit the bar missed from
very close in and you know if one of those had gone in then the match would
have had a different outcome but when you look at the XG stats for Man City
you find that most of them were quite low XG shots but if you added all of
them up together then you would actually see that both teams you know one had a
certain type of star which was generally attacking one had a defensive style and
they were cancelling each other out so you could see quite clearly through the
match that that was what was happening you could see it with your eyes but that
may have misled you because of the amount of shots that City had but XG
refines that and told you what the true balance between the two teams were and
that was actually pretty close which is quite pleasing for me but not so much
for round City fans but if you look at other matches mmm that are very
one-sided what you tend to find is that the XG so if you plot the number of
shots and the XG of each individual shot with a graph from 0 to 90 you’ll see
that you know the graphs that have gradually steps up like this and if
you’ve got a team that’s dominant then that XG keeps stepping up keep
stepping up and you can see that the massive gap appears between the two
teams now of course in football you know you could get a fluke eagle but
generally that tells you that one side is very dominant and you should do a
stretches you should be biased towards that side and likewise you know if you
look at a nil nil draw then if you plot to the XG then they may be creeping up
at about the same rate but the overall additive value of that the sum of all
those XG values may actually be pretty low so therefore it’s sort of saying
well these teams are evenly matched but neither of them are likely to score or
it’s unlikely to school and it can be quite a good precursor for the total
number of goals in the match and the way that it’s biased to one team or the
other and so much better measure but if you’re just using outright shots and
when you compare that of course you’re going to get slight anomalies but the
actual difference between the number of shots and the number of goals that comes
through from XG is pretty you know pretty good you’re not seeing a massive
deviation and in fact if you look at the standard deviation of goals in a match
then the XG value the standard deviation of X G value is actually much much lower
so that’s telling you that in fact it’s discounting more information it’s like
it doesn’t guarantee that what is going to happen will happen but generally you
find that XG presents a better view of exactly what’s going on in the
underlying market for the reasons that I’ve highlighted between man City and
also you know other types of matches while the top level may be showing you
one certain thing XG gives you much deeper discount in terms of the level of
activity that’s going on within the match and how that’s likely to play out
so yeah generally pretty accurate standard deviation is lower than the
number of goals so that’s discounting better quality information it’s actually
one of the ways I suggest you forecast everything is that you have an opinion
over here before the match starts the match starts
and and your opinion is over there so gradually the two of them should get
adjusted until you get closer together so it’s generally incremental changes
you make to change your opinion rather than just saying no that’s not going to
work and I’m switching strategy and actually gives you that and that’s
typically why I tend to use their what I’m looking at football matches in play you

6 thoughts on “Betfair football trading – More on Expected Goals – xG

  1. like getting blood from a stone but where does a leyman get the data, even if it is just a basic xG model and not a million variables like opta?

  2. Hi Peter. Firstly can I say how much I am learning from all your multifarious videos. Thanks so much for them. Secondly, a question about the validity of the xG measure. I have an app called Infogol that gives xgdata and maps at half and full time. Two matches from this evening from full-time. Sociedad v Barcelona, xG 0.99 v 1.57, actual goals 2 v 4. Even more interesting Liverpool v Man City xG 0.8 v 0.95, actual goals 4-3. I realise these are only two matches from this evening, but particularly with the latter Liverpool game how valid is xG as a predictive measure? Apologies if I am missing something?

  3. Hi Pete, can you SHOW us as well as telling us, please? Would also like your views on football trading on match outcome hedged with correct score or <2.5? Appreciate your videos and the time you put in. Cheers.

Leave a Reply

Your email address will not be published. Required fields are marked *