Part 17, I think, is upon us. Previously we looked at expected points and actual points when using possession as the crucial pivot. Today we shall be using possession and the important 1% of possession in just the same way. Only this time we shall delve a bit deeper than just expected points by possession, and instead look at expected shots for by possession.
Join me after the jump for explanations, colorful graphs and why Swansea were the worst team in the league when looking at actual shots for vs expected shots for.
So here we are again. A quick explanation is needed for how we work out what each teams expected shots for total was.
Average EPL possession over the seasons course 50%
Total shots for in the EPL 2011/12 10891
Average per team 544.5
Average per 1% of possession 10.891 (544.5/50)
Arsenal as an example- 637 shots, 10.84 shots per 1% of Arsenals possession (58.76%). We see that Arsenal's actual shots for were less than their expected shots for.
Expected shots for is a teams possession% x 10.891 (the League's average shots per 1% of possession)
Shots For With Possession%
The above graph simply lists shots for totals by each teams respective possession figure (brackets next to team name). Possession decreases going from left to right.
Shots For Per Teams' 1% Of Possession
We see no clear runaway team by the shots for per 1% of possession metric. This is a great little graph as it doesn't take raw shots totals, but instead tells us what each team did per 1% of possession it had.
West Brom do well for a mid table side with 11.85 shots per 1%. Man City unsurprisingly lead the pack. Swansea are the worst team, and uncomfortably so, registering only 8.14 shots for per 1% of their possession%. They had plenty ball and didn't do a lot with it. We shall see in the next post whether Swansea used that high possession number (57%) to contain the opposition as it certainly doesn't look like they used possession in an attacking sense.
Shots For Per Teams' 1% Of Possession Differential
Here we see which teams were a + or a - team when thinking of shots for per 1% of possession.
Swansea, Stoke and Wolves are all below the expected totals per 1% of possession. Yet, so too are Newcastle who had a + actual points to expected points number (previous post). Man City and Chelsea lead the way here. Liverpool are 4th, which fits the pattern that I have seen previously (good stat counts, couldn't convert shots on target into goals)
Man Utd are the biggest shock. They had 57.78% of possession but only registered 646 shots for, and only 0.29 shots for per 1% of possession better than was expected.
Actual Shots For v Expected Shots For (38 Games)
This is the important graph. This shows the full season data set. The blue bars are a teams actual shots for and the red bar are the teams expected shots for (individual possession% x 10.89--average shots per 1% of possession).
This shows us exactly how well (or, poorly) each team did with their possession. Did your team make the most of it in terms of shots for, did the exceed the expected shots for total? Not many teams did exceed it.
Actual Shots For vs Expected shots for Differential (38 games)
Some interesting results here.
Arsenal are 2.90 shots shy of their individual expected number. Is this what we have always known about the recent Arsenal, pass and not shoot? It's a good plan, for shots from distance (which fans cry out for) is a terrible waste of possession.
Swansea are down 160 shots on the season. 160 is a huge figure and this is the type of result that shows why this is a good little method to use. Swansea, as far as we were told, were an excellent passing and possession side and that we should of been sufficiently impressed by that. I always knew there was a trick to Swansea's possession, that trick was that their possession was relatively harmless. This wasn't a Barca style possession (as the Guardian and Telegraph guys love to call it) this was a toothless, containment style possession with little cutting edge.
West Brom and QPR with more shots than were expected. Villa and Stoke the two standout teams who posted a - expectancy number.
Actual shots for v Expected shots for differential (per game)
Last graph of the game and the one that shows us the most relevant data. This is the +/- per game.
We see some patchy results. Mid-table are more or less around their individually expected shots for numbers. The bottom six sides by possession show a mixed bag of results. Some +, some -. A shot per game + or - expectancy is a fair amount in my opinion.
It's only when we move to the top of the possession table that we see some remarkable results.
Man City, Spurs and Chelsea are all around +2 shots for per game than they were expected to be. Man Utd were only + 0.44 shots for per game than was expected. Swansea are a truly staggering -4.20 shots for per game below expected. That is a staggering number of shot per game that they didn't register according to what their (high) possession total told us they should have.
Shots for as a stat is not one I am normally comfortable using. However, I thought it necessary to give all the commonly available stats the same treatment in terms of actual and expected +/-.
- Expected shots conceded
- Expected shots on target for and against
- Expected goals for and against