Burnley: Is it more than just good fortune?
Football is becoming a tale of expected goals. Playing the game according to statistical analysis is a rough road with significant bumps along the way.
Usually a composed, considered presenter, Jeff Stelling was vehement in his vilification of the nerdy, 'lacking the soul of football' stat. To some extent, he has a point. Football is not a game played on excel spreadsheets. It is played between white lines on green grass by human beings. Uninterrupted flow over a large expanse is difficult to compute and model.
Yet, from another perspective, Stelling is totally wrong. How many times has he uttered the infamous phrase, "he should have scored there?" Expected goals puts a number on that statement, as in he should have scored 57% of the time.
That is all the figure is: a model that provides a snapshot depicting how often a particular shot should find the back of goal. It's based on the historical trends of a variety of factors, from the position of the shot, the angle, type, how and whence the final pass was delivered, plus several other variables that influence a shot's potency.
There are flaws. Few models incorporate the shooter's ability, or the skill and range of the defenders and goalkeeper involved. There are also chances not factored into a team's expected goals. If a shot was not taken because of a last-ditch tackle, even when the striker 'should have scored', it would not be involved in the calculations.
Generally speaking, though, expected goals provides a good overview of a game's pattern, individual performance, and of an overall season for a particular team.
Burnley football club's unexpected success shines a light on expected goals application. When perusing the table, Sean Dyche's typically embattled Lancs are difficult to ignore. Sixth in the league--they were briefly up to fourth--with just two losses in their last eight games, the Clarets are having a remarkable season. Conventional wisdom suggests they shouldn't be.
Opta's expected goals model has Burnley placed in 18th with just 13 expected points. That's an extreme example from a wide range of possibilities, but it illustrates the general point. Dyche's team is outperforming its statistics.
Consider the number of goals Burnley has scored and the number of points it has accumulated. Analyzing the Premier League table, the general trend is one goal equals one point.
The club with the greatest variance between goals scored and points recorded, excluding Burnley, is Chelsea. with -6. The Blues have scored six fewer goals than they have taken points. Burnley's variance is -16. The Clarets have somehow conjured 32 points from just sixteen goals.
Confoundingly, Arsenal and Spurs, two points above and one behind Burnley respectively, have goal differences of +11 and +13. Burnley's goal difference is just +4. Burnley's goals are far more valuable, in terms of points, than any other team in the Premier League. This is a trend that, in theory, should not persist throughout the whole season. In 2018, the 'smart' money will be on Burnley falling down the table.
The same story conclusion awaits when you dig into the underlying statistics. Per game, Burnley rank 14th in shots, 18th in possession, 20th in pass completion rate, 16th in key passes completed, 1st in shots conceded, 3rd in saves and 1st in shots blocked. In terms of the baseline statistics for both the offensive and defensive performance, Burnley are consistently towards the lower reaches of the Premier League. Yet, the Turf Moor club is sixth.
Assuming Burnley is extremely lucky is difficult to believe. Seventeen games into the season is usually a sufficient sample size to iron out anomalies. Yet, the only other conclusion is to make Jeff Stelling's day by saying the statistics are lying. Burnley is shocking perception, the scientific, logical and even footballing perception. And it is wonderful to watch. Long may it continue.