SATISTICS!! (Grooaaaannnn)

Ok, ok, hear me out. After some of our discussions in the "Sarri - In/Out" thread regarding statistics, i decided to take a closer look to see what I could find. 

Statistics in my view can help to illustrate a picture. Yes, they can be manipulated (which I might be doing in this post!), but I will attempt to be as objective as possible. I will just layout the stats and let you all draw your own conclusions. If you don't trust stats at all, no worries either. Just consider this the ramblings of a mad man with way too much time on his hands. 

For a few stats below, I'm using a stat called xG (Expected Goals) which I believe @Argo brought up. It's a fairly new type of stat which I find useful to describe the quality of a scoring chance. Basically it's a % probability of a chance going into the goal, based on thousands of previous chances from the same spot in which the chance occurred. Things that are taken into account include position of chance, type of pass that led to chance, how many and how close defenders are to the chance, was there defensive pressure on the chance, which foot the chance is taken with etc. etc.  It does not take into account the player taking the chance, which has obvious issues, but also isolates the chance itself which can be useful. 

We've all heard the phrase "He really should've put that in the back of the net". xG attempts to put a number rating on that chance. 

A penalty using this method is rated around 0.77 or so, which means penalties have historically had a 77% chance of going in. Hazard's goal against Wolves is rated at 0.02, which means it had a 2% chance of going in. For more information of pros and cons to xG, please see this post on Reddit https://www.reddit.com/r/soccer/comments/9a7dz3/xg_what_is_expected_goals/

xG in total over a period of time for a team: The higher the number, the more chances you create (not necessarily high quality, but the highest totals are generally a combination of high volume and high quality). The lower the number, the less chances you create. 

So after researching and plugging in of numbers, here are some rankings I found:

non-xG stats

Goals minus Penalties: 6th in the PL (45).   Man City, Liverpool, Arsenal, Spurs, Man Utd ahead of us. Watford, Everton, WestHam directly below us. 

Total Shots minus Penalties Awarded: 2nd in the PL (471).  Man City 1st. Liverpool, Leicester and Man U right below us. 

Out of the box shots: 1st in the PL (181).  Leceister, Fulham, Southampton directly below us. Man City and Liverpool 5th and 6th respectively. 

Penalty Area Shots, minus Penalties awarded, minus 6 yard box shots: 2nd in the PL (267).  Man City 1st. Liverpool, Everton, Tottenham behind us. 

6 yard box shots: 14th in the PL (23).  Wolves, Palace, Southampton, Brighton right above us. Everton, Bournemout and Fulham right below. 

Goals per Shot (minus penalties): 12th in the PL (0.096)  West Ham, Bournemouth, Brighton right above us. Wolves, Leicester, Newcastle right below. 

Stats taken from www.whoscored.com

 

xG stats (non-penalty)

xG: 6th in the PL (44.71)

xG per Shot Basically average chance quality per shot: 13th in the PL (0.095)  West Ham, Everton, Cardiff right above us. Brighton, Fulham, Leicester, Southampton right below us. 

xG from Aug 11th to Dec 31st: 3rd in PL (33.52)

xG from Jan 1st to Now: 13th in PL (11.18)

I can't find stats for shots taken over half a season. If I did, I would've looked at xG/Shot for first half and second half of season too. If anyone can find them, let me know. 

Stats taken from www.understat.com

 

Make of the above what you will. Remember this is only the attacking phase. I'll post my thoughts below

Caveat: I feel that there are certain flaws with the xG system, because Aguero's sitter that he missed at our backpost was rated 0.45 somehow, and Babel's goal against Liverpool was a 0.7 when it should've been like a 0.95. However, it just means that this model will underrate all non-penalty chances as a whole. Also, as long as a player doesn't take a shot, xG is not counted. So someone could be 1cm away from goal, but decides to kneel down and nod the ball in. Instead, he gets tackled before he can make the shot. That doesn't count into xG. 

Edited March 23, 20197 yr by dustbin719

Sarri-Out Glasses on. I've made my statement about that, so I'm sticking to it. 

The above tells me that we're taking a high volume of shots (2nd in PL), but the overall quality of our chances are low.  We can work the ball right up to the opponent penalty area, but our penetration to get those easy six yard tap in goals is really bad (14th in PL). This confirms my own eye test in recent games, In the beginning of the season, Sarri baited teams to press us high, which then stretches them out introducing spaces between the lines for us to move the ball quickly through. Now teams don't press us that high, or don't press us at all, preferring to focus on retreating quickly once they lose the ball. Once we miss that small, small window to counter attack, we have no way of beating 2 low blocks. 

Teams started to wise up right around the turn of the new year. We drew Southampton 0-0, then lost to Spurs 1-0 right after Jan 1st. Beat Newcastle 2-1, then lost to Arsenal 2-0. This shows in our xG being terrible since Jan 1st. Sarri hasn't been able to get us out of this predicament. Even after potentially finding a solution by playing a bit more conservatively, he reverts back to Plan A for Wolves and Everton. 

Against Wolves, we aboslutely peppered them with shots. By xG alone, we "won" 1.05 vs 0.61. However, after looking closely, none of our shots had an xG above 0.15. Wolves' big chance was rated at 0.58.

 

I understand that the PL website has us missing the 4th most "Big Chances" in the league. This would point to our poor finishing. However, I'd be interested to see how many Big Chances we've generated since Jan 1st. I don't have that info so hard to make a definitive comment on that. 

I have my stats in a table where you can sort by different categories. If anyone is interested, I can try to share it in a google docs here. 

 

 

 

Edited March 23, 20197 yr by dustbin719

12 minutes ago, dustbin719 said:

If anyone is interested, I can try to share it in a google docs here. 

Only if you've got some pictures of trains as well preferably steam ones.?

30 minutes ago, bluehaze said:

Only if you've got some pictures of trains as well preferably steam ones.?

Sh*t how did you know?

I’ve also got a collection of  power rangers figurines in my mother’s basement 

Mate, I wont read that long post because I dont need stats to tell me our attack is sh*te. 

I agree with your assessment. IMO I think this problem has been ongoing beyond this season. The board hasn't been able to fix our attack ever since Costa left and Fabregas' form began to decline two years ago.