Jump to content

Statistically... Things are Heading in the Right Direction (Nerd Alert)


Recommended Posts

47 minutes ago, Tamworthowl said:

I love this sort of analysis and stats, but I can't help thinking that all it says is:

 

7 clean sheets in ten is a good thing, and a big improvement on earlier in the season;

It is, but we've still a long way to go. 2 of those clean sheets were against a team in the division below. A very good, high-scoring team, but still a team in the pub league. 3 were against teams now in the bottom 6. 1 was against 10 men for most of the game, and the other was against a team managed by Tony Pulis. On average, we've scored only just over a goal a game this season, and it's now over 2 months since we scored twice. But we do seem to have turned a corner results-wise.

Link to comment
Share on other sites

2 hours ago, Tamworthowl said:

I love this sort of analysis and stats, but I can't help thinking that all it says is:

 

7 clean sheets in ten is a good thing, and a big improvement on earlier in the season;

Hull away was a bit of a freak result (although I suspect similar analysis from a Hull perspective would show it wasn't for them);

We are now creating more chances, but still not scoring enough.

 

I'm not sure that we need the analysis to tell us that.

 

Element of truth in that, but on the flip side, earlier in the season there was A LOT of false hope regarding our promotion chances. 

 

Using this anslysis, it was clear our form

aroumd that time was unsustainable, and that unless we tightened up defensively, we’d fail to maintain that run of form. 

 

We didnt improve, and our fortune waned, and we began to get the results our performances deserved. 

 

Like all stats, they tell a segment of the story, and they need to be applied appropriately. But I, for one, think the xG model is as good a model that exists (although it isn’t without its limitations). 

Link to comment
Share on other sites

 

Millwall 0.68 - Sheff W 0.36

 

A 0-0 draw a fair result. Neither side created an opportunity with a better than 11% (0.11 xG) chance of scoring. Statistically, our best chance of scoring coming from Forestieri's woeful free kick (0.08). Fletcher's last minute opportunity had a xG of 6%. 

 

It was our 3rd worst attacking display of the season (after Hull (0.15) and Brentford (0.17) away). 

Link to comment
Share on other sites

3 minutes ago, frastheowl said:

 

Millwall 0.68 - Sheff W 0.36

 

A 0-0 draw a fair result. Neither side created an opportunity with a better than 11% (0.11 xG) chance of scoring. Statistically, our best chance of scoring coming from Forestieri's woeful free kick (0.08). Fletcher's last minute opportunity had a xG of 6%. 

 

It was our 3rd worst attacking display of the season (after Hull (0.15) and Brentford (0.17) away). 

 

Two question strike me from this post:

 

Why wasn't Joao's attempt higher scoring than Fletcher's? To me Fletcher's seemed a much more difficult chance than Joao's.

 

Were we really less of a goal threat yesterday than we were at Bramhall Lane? We were camped in our own half, how come we "scored" more in that game than yesterday?

Link to comment
Share on other sites

Been saying a much more dumbed down version of what the op has put into facts and stats for a long time.

 

Basically we create sod all and often the opposition dominate and have clearer chances.

 

Lots of fans disagree, often pointing to x amount of shots per game as if they all chances.

 

Well presented op and cheers for putting the work in.  

  • Thanks 1
Link to comment
Share on other sites

19 hours ago, Tamworthowl said:

 

Two question strike me from this post:

 

Why wasn't Joao's attempt higher scoring than Fletcher's? To me Fletcher's seemed a much more difficult chance than Joao's.

 

Were we really less of a goal threat yesterday than we were at Bramhall Lane? We were camped in our own half, how come we "scored" more in that game than yesterday?

 

My mistake actually. Fletcher's chance had a 0.06 xG value, whereas Joao's had a 0.07xG. Both were similar...Fletcher's was from a closer distance than Joao's but he was off balance, but Joao's effort was under less pressure. 

 

Regarding the game against Sheff U, Morgan Fox's chance in the 8th minute when he put it over the bar from 6 yards out had a xG value of 0.59 (59% probability of scoring). Interestingly enough, in the game against United, if you use a npxG (Non-penalty xG) model, we created slightly better xG than United did. 

 

And it's that type of game which really brings xG into it's own. Looking at basic statistics, possession, shots at goal, shot on target etc, United battered Wednesday. But, penalty aside, United never troubled the Wednesday goal with any real threat, and the xG values show that. Fox's chance, discounting McGoldrick's penalty miss, was easily the best chance of the game. The next best quality chance was Norwood's free kick in the last minute, which had a 0.08 xG value. 

Link to comment
Share on other sites

On 11/02/2019 at 10:04, frastheowl said:

 

I like my stats. And I particularly like the usage of the ever more popular (and common) Expected Goals statistics. And a warning...if you don't like graphs...click Back now. 

 

Essentially, for every effort on goal, a number is given (actually a percentage) to indicate how good a chance that is. Lots of factors are taken into account, whether it's a header, a shot, an open net, a dead ball situation etc etc. 

 

As a crude rule, the teams who create a high Expected Goals number (and concede a low Expected Goal number) tend to be the better performing sides in the division. They can indicate trends, and whether runs of form are sustainable (more on that later), if teams genuinely are unfortunate or if a team is statistically punching above their weight. 

 

And using the InfoGol website, things are starting to look up for our rejuvenated side following the dismissal of Jos. 

 

UNADJUSTEDNONRAW_thumb_1.jpg.c487d5b593d66ea3b2ead7bfcb73a3bc.jpg

 

This graph shows the Expected Goals difference in each of our game weeks. Example, 1st game of the season, Wigan Away. Wigan created 2.42 xG, whilst we created 1.36 xG, therefore our xG Diff was -1.06. 

 

The Brown shading represents Jos' time here, Orange Bullen's, Green Agnew/Clemence's and finally Blue is Bruce's. 

 

As you can see, throughout Jos' period in charge we were constantly conceding better chances than we were creating. Even when we had the run which culminated in us sitting in the playoff positions, we were still performing to a negative xG difference. Indicating the run was unsustainable, and heavily reliant on good fortune (either underperforming opposition strikers, overperforming goalkeeping or overperforming in our own finishing...i.e. Reach's run of wonder strikes). 

 

Bullen began to change things for the better, although we only managed to outperform the opposition in one of his three games in charge. Agnew oversaw our best performance of the season...the 1-1 draw against Birmingham. We created a high xG in that game, and limited Birmingham to very little. In Bruce's 1st two games, granted against two of the league's worst sides, we've put in two dominant displays in terms of xG. 

 

For the 1st time this season, we've managed to maintain a positive 5 game average xG diff for a sustained period of time. Taking the Hull game as an anomaly, since game week 21 (Rotherham) we've seen a definite trending improvement. 

 

Essentially, what this indicates is, if we start to find our shooting boots and maintain the current levels of performance, our results will continue to improve. 

 

image.png.572149eed76a64978ec244cd79265691.png

 

Our offensive xG has actually remained at a similar level throughout the season, between 0.75 and 1 for the most part. Indicating, on average, we're creating enough chances to score about a goal a game. What has changed, aside from a 5 game stretch between GW 15 and 19, is we are currently underperforming to our xG, whereas our over-performance earlier in the season we down to our ability to score goals from positions with a low xG. Prime examples are Reach's goals against Boro, Leeds and WBA or both goals away at Reading (combined xG of 0.09 resulted in a actual goal score of 2.0).

 

image.png.7fb12c11b1981a38fc1e50e058ad4fdf.png

 

But, without doubt, our biggest improvement since Jos' departure is our ability to prevent the opposition creating good quality chances. Almost instantly from the game week of Jos' departuure (GW22), our xG against numbers have nosedived. In only one game since Jos' departure have we conceded a xG of more than 1.0, against Hull, which is beginning to look more and more like an anomaly...an off day. 

 

Overall, there's reasons to remain positive. I'm certain Bruce will want us to start to ensuring we convert the chances we create. If we do so, and maintain our recent trend of conceding few good quality chances, results can only continue to improve. 

Who decides what's a good chance and what's not? 

Link to comment
Share on other sites

It's going to take some time, but I'm pretty confident that Brucey will get us firing in the right direction before too long. 

 

Whether he can do that with this current set of players is up for debate, but the one major positive has been the keeper and back four. 

 

Needless to say, Moustachio's time at the club was nothing short of an embarrassment. If Chansiris grand plan to avoid P and S was promotion this season, then (by appointing Manuel) he dropped so many bo11ocks that there are no more bo11ocks to drop. 

Link to comment
Share on other sites

2 hours ago, LiamAP22D said:

Who decides what's a good chance and what's not? 

 

Each effort on goal is given a xG value, indicating how likely it was to score a goal from that situation. 

 

The value given is dependant on a whole host of factors, the biggest two been distance and angle from the goal. Other factors include whether it's a set piece, a header, defenders positioning, from a through ball etc. etc. eventually come together to give the opportunity a xG value which is essentially a decimalised percentage of the chance registering a goal. 

 

So a chance in the middle of the goal within the six yard yard is likely to give a high (>0.7) xG, whereas a speculative long shot from 30 yards out on the half volley is likely to give a low (<0.05) xG. Throughout the game, these chances get added up to give a xG for the team. 

 

You can then compare the xG to how many goals a team actually score (or concede) to see if the team is overachieving (or underachieving) its expected xG. When Wednesday went through the period earlier in the season when Reach scored a host of wonder goals and Joao chipped in with a couple also, we seriously overachieved our expected xG. Generally speaking, teams overachieving their xG struggle to replicate it over a period of time (they're lucky). The only team in recent years who maintained it over a prolonged period of time were Reading when they finished in the playoffs a couple of seasons back. Generally, the league table correlates with the teams xG diff reasonably well. For example, this seasons top five placed teams in the Championship, would all occupy top five positions if the league table was decided on xG diff. 

Link to comment
Share on other sites

 

16 hours ago, bigrbuk said:

What's the lowdown for Saturday's xG at Rotherham then? Where should I stick my money if that's correct?

 

Genuinely fascinated by this and have a grand idea about how to use the data alongside my Python learning

 

The infogol app predicts, based on the season's xG, a slight favouring towards a Rotherham win (41%), with the draw at 28% and a win for ourselves 31%. So it's a tough game to call really.

 

And if you consider the two teams most recent form, that doesn't really change either. Both sides have been unfortunate not to pick up more points than they have. Rotherham's xG points value over their last six games is 9 points, however they have only achieved 3 points in that time. They were on the end of an undeserved defeat away at Ipswich, creating 1.15xG and conceding only 0.38...yet found themselves on the end of a 1-0 defeat. 

 

They arguably deserved a point in the 4-2 defeat at home to Brentford (xG 1.82 - 2.07), and should've beaten Wigan at the New York Stadium creating 0.61xG more than their opponents. 

 

Over the course of the last six games, Rotherham have more or less matched their xG with the actual goals they've scored. But they've conceded far more goals (10) than their xG (7.15). Without looking at the goals they've conceded, that suggests either a run of bad luck with some good strikes scored against them or poor goalkeeper (or a mixture of both). 

 

Ourselves on the other hand, have suffered with our inability to convert chances. Creating an xG of 7.22 over the six game period, we've only converted 3 of those into goals. Whereas our defensive displays have pretty much matched our conceded xG (4.77), conceding only four goals in those six games. 

 

For me, I'd probably opt for a low scoring draw. However, if Wednesday start to correct the discrepancy between xG created and goals scored, then an away victory is on the cards. 

 

 

  • Thanks 1
Link to comment
Share on other sites

On 11/02/2019 at 10:04, frastheowl said:

 

I like my stats. And I particularly like the usage of the ever more popular (and common) Expected Goals statistics. And a warning...if you don't like graphs...click Back now. 

 

Essentially, for every effort on goal, a number is given (actually a percentage) to indicate how good a chance that is. Lots of factors are taken into account, whether it's a header, a shot, an open net, a dead ball situation etc etc. 

 

As a crude rule, the teams who create a high Expected Goals number (and concede a low Expected Goal number) tend to be the better performing sides in the division. They can indicate trends, and whether runs of form are sustainable (more on that later), if teams genuinely are unfortunate or if a team is statistically punching above their weight. 

 

And using the InfoGol website, things are starting to look up for our rejuvenated side following the dismissal of Jos. 

 

UNADJUSTEDNONRAW_thumb_1.jpg.c487d5b593d66ea3b2ead7bfcb73a3bc.jpg

 

This graph shows the Expected Goals difference in each of our game weeks. Example, 1st game of the season, Wigan Away. Wigan created 2.42 xG, whilst we created 1.36 xG, therefore our xG Diff was -1.06. 

 

The Brown shading represents Jos' time here, Orange Bullen's, Green Agnew/Clemence's and finally Blue is Bruce's. 

 

As you can see, throughout Jos' period in charge we were constantly conceding better chances than we were creating. Even when we had the run which culminated in us sitting in the playoff positions, we were still performing to a negative xG difference. Indicating the run was unsustainable, and heavily reliant on good fortune (either underperforming opposition strikers, overperforming goalkeeping or overperforming in our own finishing...i.e. Reach's run of wonder strikes). 

 

Bullen began to change things for the better, although we only managed to outperform the opposition in one of his three games in charge. Agnew oversaw our best performance of the season...the 1-1 draw against Birmingham. We created a high xG in that game, and limited Birmingham to very little. In Bruce's 1st two games, granted against two of the league's worst sides, we've put in two dominant displays in terms of xG. 

 

For the 1st time this season, we've managed to maintain a positive 5 game average xG diff for a sustained period of time. Taking the Hull game as an anomaly, since game week 21 (Rotherham) we've seen a definite trending improvement. 

 

Essentially, what this indicates is, if we start to find our shooting boots and maintain the current levels of performance, our results will continue to improve. 

 

image.png.572149eed76a64978ec244cd79265691.png

 

Our offensive xG has actually remained at a similar level throughout the season, between 0.75 and 1 for the most part. Indicating, on average, we're creating enough chances to score about a goal a game. What has changed, aside from a 5 game stretch between GW 15 and 19, is we are currently underperforming to our xG, whereas our over-performance earlier in the season we down to our ability to score goals from positions with a low xG. Prime examples are Reach's goals against Boro, Leeds and WBA or both goals away at Reading (combined xG of 0.09 resulted in a actual goal score of 2.0).

 

image.png.7fb12c11b1981a38fc1e50e058ad4fdf.png

 

But, without doubt, our biggest improvement since Jos' departure is our ability to prevent the opposition creating good quality chances. Almost instantly from the game week of Jos' departuure (GW22), our xG against numbers have nosedived. In only one game since Jos' departure have we conceded a xG of more than 1.0, against Hull, which is beginning to look more and more like an anomaly...an off day. 

 

Overall, there's reasons to remain positive. I'm certain Bruce will want us to start to ensuring we convert the chances we create. If we do so, and maintain our recent trend of conceding few good quality chances, results can only continue to improve. 

was it cloughie who once said 'football is played on grass, not paper'?

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...