Every season since I started blogging about Arsenal, which coincides with Arsenal moving to the new stadium*, the Gunners have topped a different table. In the eight years I’ve been doing this Arsenal have won the Fair Play table, the Injury Table, the money table, the “over-achieved for total spend” table, the “Referees have done one on us Table” from Debatable Decisions (defunct), and the Total Shots Ratio table. Sometimes Arsenal have been so good that that have done the double or triple and topped multiple tables in a single season! But all those tables pale in comparison to the new table Arsenal top. I am proud to announce that thanks to the international week Arsenal are still top of the table, baby. The “Expected Goals Table” that is.
I’m being cheeky but the Expected Goals table is actually not a bad table to top. The “xG Table” as it is known is the logical outgrowth of the Total Shots Ratio table and xG is a damn fine tool for predicting how many goals we would expect a team to score and concede over a given number of shots.
If you’ve been around my blog for any amount of time you know that I am a fan of total shots ratio. It’s not a perfect analytic tool but it makes a lot of sense, has a high rate of correlation, and it’s really simple. Basically, the best teams shoot more and simultaneously limit the opposition’s shots and the result is that they tend to win more. This, however, isn’t always true.
The Invincibles had the lowest Total Shots Ratio (0.63) of any Arsenal side from 2000-2013. The only Arsenal side that had a lower TSR was last season’s Arsenal side, which had a TSR of just 0.60. Meanwhile, the Arsenal side of 09/10 had the best Total Shots Ratio I’ve seen in a long time at .69. If you remember, that was the season Arsenal leapfrogged into… third place. Arsenal probably could have finished better than third except that Arsenal allowed the opposition 13% conversion rate and as a result, shipped 41 goals that season. Up until that season, Arsene’s Arsenal had only allowed 40+ goals twice. After that season, 40+ is the new normal.
So, how did TSR not predict that Arsenal would finish 3rd? 2009/2010 was the first season, if you remember, that the opposition started to figure out how to play against Arsenal’s high defensive line and take advantage of Arsenal’s lapses in concentration at the back. Conceding against corners, set plays, headers, in addition to the route one counter attacks were all hallmarks of that team and remain the main method teams employ to beat Arsenal now. So, while we know that Total Shots Ratio is a good measure, it’s also flawed. Some shots are simply better than others!
This is a drum I have been beating for years. Shots in the box are converted at a higher rate than shots outside the box and shots in the “prime areas” are converted at an even higher rate. Moreover, headers, even in prime areas, are converted at a lower rate and, unless you’re Luis Suarez**, headers from outside the prime areas are almost never converted. We also know that crossing the ball is not as effective as playing a through-ball for an assist, that corners are a low percentage shot, and that one-on-one’s with the keeper are terrific chances. This is something you’ve probably heard me say here and on Arseblog News: all key passes are equal, but some key passes are more equal than others.
This is where the newest metric, Expected Goals, comes in to play. What folks like Michael Caley, have done is taken the idea that some shots are better than others and created a much more detailed version of the Total Shots Ratio. By analyzing where the shots are taken, not just how many, and where the teams are allowing the opposition to get shots and using them in a ratio, Caley is able to come up with a much more accurate predictor of League Performance.
And if you even take a cursory look at Caley’s tables, Arsenal are tops: they top Total Shots Ratio, they top Danger Zone Ratio (DZR), DZR minus crosses, Expected Goals, Expected Goals Against, Expected Goals Ratio, Strength of Schedule (actually they are third), and Adjusted Expected Goals Ratio. So, why are Arsenal in 6th place? Well, several reasons.
First, I have no criticism of the work that Caley has done. This is a tremendous boon to the stats community and something I have wanted to do for years but haven’t been able to put together the time. His model isn’t wrong, his model shows that Arsenal are generating great shots in great areas while limiting the opposition to fewer of those same shots. On average, we expect that Arsenal would score more goals and concede less goals than other teams who are taking fewer and allowing more of those same shots. The problem is that Arsenal often buck stats trends.
Like I showed above with the Invincibles and with the 09/10 Arsenal side the peculiarity of Arsenal is that while the boss plays the averages and uses them in his analysis of the game, the particulars, the Achilles heel, of Arsenal continue to be exploited.
It’s not always the same heel that teams nip at. As you can see from the 7amkickoff Index, this season, the thing that teams are picking on is crosses. In the 11 matches Arsenal have played in the Premier League, Arsenal have conceded 6 goals off crosses and a further 5 of those crosses have been headed. Arsenal have conceded 13 goals in 11 games when the xG numbers say we should have conceded about 8. I mark that increase down almost entirely to headed goals conceded off crosses: a low percentage shot that Arsenal seem vulnerable to this season.
That brings me neatly to the weekend’s match against Man U. Arsenal are the most prolific crossing team in the League and Man U are just 1 cross per game average less that Arsenal. Man U get almost all of that crossing from just one player: Angel Di Maria. ADM is 1st in the League in crosses attempted, he’s 3rd in accurate crosses, he’s 2nd in accurate corners, he’s 3rd in the League in generating shots off crosses, and tied with Fabregas for 1st in the League with throughball key passes. If there was any player Arsenal don’t want to face on Saturday, it’s him.
Those who see the cup as half full will see that statistically Arsenal are doing the right things to prevent goals and to score goals. As Wenger would probably point out, the Premier League race is a marathon and this one statistical aberration of headed goals conceded off crosses should revert to mean. Thus, if Arsenal just keep doing what they are doing they should be a shoo-in for the top four.
Those who are cup half empty, spilling out into the streets which run red with Achilles’ blood, will probably point to Arsenal’s vulnerability to the same old faults and say “statistics don’t tell the whole story.”
*Kick 7amkickoff out of football!
**Scoring headers from this distance is almost unheard of. I would bet that there hasn’t been a header scored from the top of the box in 10 years. I know that Wilshere’s goal was beautiful to watch but given the rarity and skill on display here how this goal wasn’t even nominated for goal of the season is beyond me.