Shooting models, expected goals & Pinnacle Sports’ alternative Premier League table

Shooting models, expected goals & Pinnacle Sports’ alternative Premier League table

By Mark Taylor Aug 15, 2014

Tweet

Soccer bettors looking to make a profit on the 2014/15 season should understand the significance of ‘luck’ on a game-by-game basis. This article highlights the limitations of using a teams league position as an indicator of their ability, and explains how shot data is more accurate at determining whether or not a team are performing above or below expectations.

Premier League: ‘the superior 7′ and the ‘threatened 13′

Over the last decade of Premier League soccer it has become evident that the league has split into two distinct groups. The big three has expanded to four or five sides that have a realistic chance of winning the title, while regular top half finishers, such as Everton and Spurs are vying for a Champions League spot.

The remainder of the Premier League sides look towards domestic cups for any realistic chance of silverware, but their main target in August, is to avoid relegation.

Simon Gleave has coined the phrase ‘the superior 7′ and the ‘threatened 13′, which excellently describes the expectations of the 20 Premier League teams at the start of a new EPL campaign.

Movement within the two groups is most evident for the ‘threatened 13′ where talent levels are similar. Soccer is a low scoring sport where a run of good fortune can easily turn possible draws into wins and propel a side from the fringes of relegation to inside the top ten, leading to the illusion of improvement, especially if judged simply by league position.

Is league position a good indicator of a team’s strength?

A top 10 finish is regarded as a successful season for the Premier League’s also-rans, but the record of such a side in the following season strengthens the argument that luck may play a big part in their apparently impressive finishing position.

Over the previous 10 seasons, 37 unfashionable sides have finished in the top half of the table, either by occupying 8th place or below, or by occasionally displacing one of the superior 7 from their position of power. Their average finishing position has been eighth, their goal difference has averaged zero and points total has been 54.

Understandably, the following Premier League season has been met with optimism, and an expectation that an impressive year can be used as a building block to break permanently into the elite group.

However, only six of the 37 sides over the last decade have managed to improve on their league position in the following season, with two other sides holding station. Their average finishing position for the following campaign fell to 12th, goal difference was minus nine and points were 8 fewer at an average of 46. Two sides were subsequently relegated and three others only narrowly escaped in 17th spot.

Therefore, league position may at best, give a broad indication of team quality, especially away from the top half dozen slots. With a recent influx of available soccer data, an opportunity to identify which teams may be riding an unsustainable wave of good fortune or conversely create opportunities, but fail to get a fair reward.

Shots are increasingly challenging goals as the definitive measure of team quality. Initially shots taken and conceded were used, but in recognition that all shots or headers aren’t created equally, shot models, which incorporate shooting zones or actual pitch coordinates are used.

These shot models attempt to estimate the expected number of goals a side might score and allow over a season from the quality and quantity of opportunities they face or produce.

Although not perfect, by looking at shots, we can begin to re-evaluate actual match results to see if the result fairly reflects the efforts of each side on the day, or perhaps help to identify a side which partly benefitted from luck.

Case study: England at the 2014 World Cup

England’s early departure from the 2014 World Cup was naturally regarded as disappointing. Despite being drawn in the most difficult group, a single point from a meaningless final game was a poor reward.

However, a more measured view highlights how narrow the gap between success and failure is when talent is of a similar level and decided over a short run of games.

England out-shot each of their three group opponents. Their expected average number of goals from these shots was slightly superior to those of Italy in the opening game and higher than both Uruguay and Costa Rica in the subsequent two matches.

The leisurely way in which Costa Rica – having already qualified – approached the final game may skew any conclusions drawn from the data. But by re-running each game from Group D using the probabilities that each goal attempt would actually produce a goal, allows us to see how fair the final group table was.

In short, we are giving each side many more opportunities to turn their shooting ability into goals and defensive prowess into saves.

Under this shot based assessment, England’s stock rises considerably, with qualification for the knockout stages becoming more likely than not, over 10,000 group simulations.

This is partly by virtue of the possibly unrepresentative Costa Rica game, but also because of the quality and quantity of chances created against both Italy and Uruguay. Shots therefore, add context to England’s poor points return from two narrow defeats and a draw.

2014/15 Alternative Premier League table

During the forthcoming Premier League season, we will produce an alternative table, based on expected goal models, derived from every actual shot attempted and conceded in each game, incorporating simulations of each league match played as the season progresses.

The aim is to highlight those sides which may be unsustainably under or over performing in the traditional league compared to a broader measure of their attacking and defensive capabilities.

As well as looking at individual games of note from each weekend, the alternative table will also seek to show where teams may have been with a kinder bounce of the ball, or the degree of trouble a poor run of luck could cause.

This is vital for bettors. Recognising a teams relative over their perceived strength, allows them to notice when teams are over or underperforming. Identifying this before the market creates an opportunity to gain a soccer betting edge.

Mark Taylor is a freelance soccer and NFL writer who, along with producing expert content for Pinnacle Sports, also runs his own soccer analytics blog, the Power of Goals.

If you have feedback, comments or questions regarding this article, please email the author or send us a tweet on Twitter.