Friday, November 2, 2012

Team Shots +/-

One of the elements which feeds into the new weekly team and player forecasts is how each team has performed against their opponents to date, compared to how the rest of the league fared (for future posts I'll refer to this as +/-). For example, in gameweek nine, Southampton managed to register 15 total shots against Tottenham; eight inside the box (SiB) and seven outside (SoB). So far this year though, Spurs have been comparatively strong away from White Hart Lane, surrendering just 11 total shots on average; six SiB and five SoB. For this gameweek then, we can calculate that Southampton exceeded other teams’ efforts by 33% for SiB (eight compared to a league average of six) and 40% for SoB (seven compared to a league average of five).

By averaging this data over the season I believe we get a better picture than simply looking at goals per game or even shots per game, as the above data is effectively opponent adjusted. At this point in the year, with just four or five home/away fixtures played there’s a risk of small sample size fluctuations, but nevertheless I believe it’s fairly useful in identifying teams who are consistently out shining the league average and thus should be in line for new or continued success in the future.

To see what this looks like, I thought I’d paste all the data to date below and then pick out a couple of teams to highlight:

Though each number has some use in predicting future performance, I’m generally going to focus on SiB as they obviously correlate best to goals scored and are a generally sound indication of a team creating quality chances rather than just taking pot shots.

  • The team that jumped out the most for me was Everton. They have been simply dominant both home and away at generating SiB, leading the league in both instances. Their conversion rate of SiB to goals is a moderate 12% which suggests there could be an element of some of these chances being headers from long balls (right Gerrard?) which tend to not be converted as well as shooting opportunities, but even so, these are impressive numbers for a side whose fantasy players come in the ~7.0m bracket and they underscore just how valuable that group should continue to be.
  • On the negative side at home, two surprising sides to note are Arsenal and Chelsea who have both only managed to perform to the league average through their four/five home games. Both still rank in the top handful of teams so there’s nothing to worry about per se, though the fact they’ve had limited success against apparently weaker sides may suggest that some regression might befall them when tougher teams visit the Emirates and Stamford Bridge. Indeed, the three best teams they’ve collectively faced this year have all held them below league average SiB by a significant amount (Arsenal: Chelsea -14%; Chelsea: Newcastle -49%, Man Utd -11%).
  • Sunderland have simply been brutal in the shot department, which won’t come as a surprise to anyone who has been looking at Steve Fletcher’s fortunes in any detail. The fact that he was able to convert his first handful of shots into goals really masked this team’s poor attacking play and it’s no surprise that goals have been hard to come by of late. On a bright note, they did manage to register 21 total shots with nine inside the box against Newcastle in GW8, but until that level of production can be met on a regular basis, the 10% of managers who own Fletcher should be concerned (even though his conversion rate of shots on target to goals of 45% isn’t crazy the fact that he’s hit the target with 79% of his shots is. Over the first seven games he’s taken 14 total shots, which based on his historical SoT% would give him just four SoT which would translate to two goals if you’re feeling bullish, rather than the five he registered so far).
  • If you have a defender facing Southampton at home you’ll likely be licking your lips and banking the clean sheet now, but the data would suggest we ease up a little on that excitement. Southampton have impressed in almost every attacking metric we have but their shot totals away from home are particularly notable. Only five teams (MCI, EVE, ARS, LIV, MUN) have fared better on their travels versus league average with the Saints managing to score in every road trip so far (despite playing at ARS, MCI, EVE and MHU). The question is whether or not Southampton have the quality to really expose the easier sides they will face in the coming weeks, but at least so far, the data likes them to not only get by offensively, but excel.
  • Man City’s data looks a bit odd as they have simply crushed teams away from home (59% above average) while their home totals are useful yet don’t standout (16%). Note, however, that they are still second in the league in SiB at home, it’s just they’ve done so against weaker opponents and thus don’t get credit for ‘overachieving’. In City’s particular case this might not be a cause for concern as there’s an argument that they will simply create enough chances against anyone to win the game, whether that means exceeding expectations with 13 SiB at Fulham, or just doing enough to get by Swansea with just seven SiB. This one is worth monitoring as City welcome TOT, EVE and MUN in three of the next four home games, so we will be able to see if they’ve truly taken a step back offensively, or were merely just toying with their opponents and doing enough to simply win the game (as asset which is often attributed to be the mark of ’Championship’ team).

So that’s a little taste of the kind of analysis we can get from this data. I don’t plan to post it every week but it’s contents will be included in all the posts along with, of course, the team and player forecast models.


chemikills said...
This comment has been removed by the author.
chemikills said...

Nice work this is what I explained to you a few weeks ago, except I did it with the defence.
Considering doing it with Def too? I think that's where it's a more useful metric.
Because comparative performances in attack can get heavily skewed by open games when both teams are really going for it?

Chris Glover said...

Yeh I've done defense too, I just wanted to highlight the attack as I was looking at it this afternoon. I'll post a similar thing next week on the defense. I think I possibly stole this from you then, so thanks :)

chemikills said...

haha np the problem with it is that it's hard to judge new form with these comparisons because fixtures are always all so different

stuckert said...

One additional thing to think about is adjusting for score effects. I know this is something that always has to be done in hockey (I'm a Canadian), because "being behind" usually motivates teams to play a more offensive style.

The famous case is the Edmonton Oilers from the 1980s. If you don't adjust for score effects (i.e. only calculate shots for / against ratios when the score is tied), then actually it looks like they were statistically quite poor. But the fact was that often they raced out to a big lead early in the game, and then sat back as their opponent peppered them with shots trying to catch up.

In footy it should be the same, if not even more drastic. I imagine it would make the calculations much more complex, but really I feel the SIB numbers should only be taken from times during the game when the score was tied.

Am I on to something or just crazy? Here's an article about score-adjust Fenwick in the NHL:

stuckert said...

By the way, I should also add that you're doing an amazing job on this site. I read almost every article you post, and quite a few have helped me out in building my team.

This site is exactly what I wished I were writing myself! Keep up the great work.