One area that I have been increasingly interested in over the course of the season in terms of statistics and analytics is that of scoring chances. We know that City have consistently been at the top end of clubs in terms of their conversion rate and that during the period where City slipped from top spot they were still creating chances, but underwent a noticeable drop off in terms of converting chances to goals.
There a variety of sources of data providing numbers around chances and shots, and whilst useful these can at times be misleading. For example, shots is very wide definition to cover an ‘attempt on goal', essentially classifying shots inside and outside the box, on and off target and blocked as one and the same.
The OPTA definition is:
Shot on target
Any goal attempt that:
a) Goes into the net
b) Would have gone into the net but for being stopped by a goalkeeper's save
c) Would have gone into the net but for being stopped by a defender who is the last man.
Shot off target
Any goal attempt where the ball is going wide of the target, misses the goal or hits the woodwork.
Any goal attempt heading roughly on target toward goal which is blocked by a defender, where there are other defenders or a goalkeeper behind the blocker.
We also now have data available for chances created too. Again, OPTA's definition is:
Assists plus Key passes.
OPTA also define an assists and key passes as follows:
The final pass or pass-cum-shot leading to the recipient of the ball scoring a goal.
The final pass or pass-cum-shot leading to the recipient of the ball having an attempt at goal without scoring.
What myself and shuddertothink wanted to quantify and measure was what is truly a scoring chance. We know from the definitions above that not all shots are equal: a shot from the six yard line with just the goalkeeper to beat can be classed in equal measure to a thirty-yard shot that sails high over the bar.
We are both close followers of the NHL and in turn some of the analysis being done to try and analyse a game that - very much like football - is fluid by its nature as opposed to one that contains static, set plays (such as cricket or American football). There has been some interesting work in this area (notably Corsi and Fenwick numbers and a writer called David Staples has too been looking at this area to try and classify how a scoring chance can be truly measured.
Staples leans heavily on work done by Roger Neilson, a former hockey coach who pioneered certain aspects of analysis back in the 1970's and 1980's around this subject and influenced other coaches in the game. Staples, on the blog The Cult of Hockey has a nice background post about Neilson numbers and also provides a clear description as to what he considers constitutes a scoring chance.
Over the past few seasons Staples has measured ‘Neilson numbers' for the Edmonton Oilers in the NHL, with the 2011/12 season findings here. He and some colleagues are also conducting a wider exercise in this years NHL playoffs [link] which I have too being following with interest. This has not been without question though, with it being felt in some quarters (there is a good debate on this article in the comments section) that the measure is not entirely faultless.
However, we both feel that the methodology applied by Staples is a workable one when looking at scoring chances in football and have looked at the past half a dozen or so City games, studying the scoring opportunities that existed in them to try and define a scoring chance in football (based on the principle of the ‘Neilson numbers'). What is of interest is that the number of chances we have seen in games are somewhat lower than the official numbers that are provided.
My final ‘working' definition of the criteria I have included as being considered a scoring chance is as follows:
- Any attempt on goal from inside the box, whether the outcome is a goal, save, clearance, block or the shot is wide or over the bar.
- Any attempt on goal from outside the box that forces the goalkeeper (or a defensive player positioned on the goal line) into a save.
The above was decided upon as with not all shots being equal, a shot from outside of the box that does not force a save (let alone one that is on target) should not be considered a scoring chance. We have though included all shots from inside the box as - similarly to Staples' scoring ‘zone' in hockey - inside the eighteen yard box I considered to be a legitimate scoring chance when a shot is made regardless of outcome due to the proximity to the opposition goal and the increased threat it carries.
Again, after looking at the past few games in trying to hone the definitions we have settled on allocating a + mark to all players that are involved in a scoring chance, where:
- They are positioned in what is generally considered to be the final third zone (or oppositions defensive area) on the pitch, or;
- Where a player is outside of the final third zone, they play a ball into the final third zone that results in a scoring chance.
I haven't veered too much from what I originally began with but have made a couple of tweaks along the way, notably to introduce a measure to allow a greater opportunity for defensive players to be allocated a + mark as, unlike in hockey, the playing area is far greater in size and there are a greater number of players on the area at any one which means in football there is less scope for defensive players to be involved in scoring chances. Therefore:
- Where a defensive player makes a block or clearance that essentially stops a scoring chance (as per my definition) that player will also be allocated a + mark.
Using the final working definition I have looked again at Monday's derby to establish the scoring chances in the game and to allocate ‘Neilson numbers' to players of both sides. Within this measure we have also accounted for both the number of minutes and number of touches that a player has, ostensibly to further establish the effectiveness (or ineffectiveness) of players involved in scoring chances (click the image to enlarge):
Given we have just one game so far based on this particular analysis we can see many players with a zero in certain categories; particular on the defensive side with United failing to get shot on target. The exercise has though highlighted the contribution (unsurprisingly) of both David Silva and Sergio Aguero, but also Pablo Zabaleta from an offensive standpoint.
This though is an exercise that is still very much a work in progress (and maybe by no means a perfect one at this point) but it is one that we are comfortable enough in putting ‘out there' at this stage. We will also repeat the exercise for the remaining two games with the plan to look at this area over the whole of the 2012/13 season.
As usual, any comments would be welcome.