#ThreeAtTheBack - The OptaPro podcast explores how analysts at international federations approach major tournaments… twitter.com/i/web/status/1… 11 Jul

NEWS: OptaPro is attending @simai2018 in Rome today, delivering a presentation on the application of data models &… twitter.com/i/web/status/1… 4 Jul

BLOG: Ahead of tonight’s T20 international, OptaPro have looked into the potential tactics which could influence th… twitter.com/i/web/status/1… 3 Jul

This @SportTechie article discusses how access to technology and data in the dugout may influence decision-making a… twitter.com/i/web/status/1… 3 Jul

A data-driven analysis of Mexico with a focus on their performances in the #WorldCup group stages. Counter. optasportspro.com/about/optapro-… 2 Jul

BLOG: Ahead of their game this afternoon, the Pro team have been analysing Mexico’s attacking approach at the World… twitter.com/i/web/status/1… 2 Jul

BLOG: Using Opta cricket data @FWV_27 investigates IPL bowler tendencies when delivering a free-hit. Jaffa. optasportspro.com/about/optapro-… 27 Jun

Head of OptaPro @BenMackriell shares insight on the analyst’s role during a major international tournament in this… twitter.com/i/web/status/1… 26 Jun

The OptaPro Blog

RSS Feed

Welcome to the OptaPro blog, featuring news and analysis from OptaPro's cutting-edge research team.

GUEST BLOG: Creating chances in the English Premier League

by Mark Taylor from The Power of Goals blog.

Baseball is the sport most often cited as being at the forefront of the use of advanced statistical analysis to both explain and predict events within their game. While much of the upsurge of football related statistical analysis can be traced back to the success enjoyed by baseball, it is also well documented that the two sports are very different creatures. Baseball abounds with set piece scenarios and one to one confrontations where historical expectancies regarding possible outcomes are well documented and backed up by copious amounts of data.

Football by contrast has been dubbed as an invasion sport where team interactions are a much more important facet of the sport. Counting statistics such as goals, crosses and tackles are excellent building blocks on the road to developing a better understanding of the sport, but the importance of on field player interactions are only gradually being explored.

Reua _528380

One possible result of excellent player interaction

One sport that has had a burgeoning statistical underground and in tactical terms lies midway between baseball's individuality and football's team ethos is American Football. The similarities with its near cousin,  football or soccer goes beyond the numerically even eleven verses eleven team confrontations and one particularly fertile ground for cross pollination of ideas lies in the NFL quarterback's relationship with his wide receivers. While opposing players from the 'offense and defense' face off in a largely unseen and until recently unrecorded contest, the creative heart of the offensive side of the ball attempt to gain potentially game changing amounts of territory by skilfully passing, catching and subsequently running with the ball. Essentially when a quarterback passes the ball to a wide receiver, we are seeing many of the same requirements that are on display when a midfielder creates a scoring opportunity for a striker.

A quarterback requires many of the attributes of a chance creator in football. He must have the vision to pick out his intended receiver, the composure to buy himself time and space to deliver the pass and the accuracy to deliver the pass both in stride and without fear that it will be intercepted. Having made a completion, the second component of a successful catch, the yards after the catch rely principally at the feet of the receiver. Speed, strength and awareness of your surroundings each contribute towards accumulation of receiver yards and once again these attributes are similar to those required of a talented goal scorer.

Reu _1451620

QB: vision

In short, passing in the NFL requires an input from both quarterback and receiver and the consensus at the moment appears to favour the bulk of the credit for the completed pass remaining with the thrower and yards subsequently gained belong more to the receiver.

The relationship between scorer and provider in football has largely been one sided. Assist numbers and chances created are recorded, but much of the game changing impact of a goal is usually credited to the scorer. Only recently have goal assists become a regular feature of the postgame data and chances created, but subsequently missed often pass unnoticed.

OptaPro have kindly provided me with much more detailed information that includes who made the goal attempt, where on the pitch that attempt was made and which of his colleagues supplied the pass. This data allows us to begin to move on from mere counting towards quantifying the value of each chance and we can better judge creative players on their ability to create chances rather than leaving them hostages to the finishing abilities of their team mates.

Knowledge of the quality as well as the quantity of chances created by a player has many advantages. From a purely descriptive nature we can judge how well a player performed over the season. He may have enabled his team mates to carve out excellent chances, but through random variations that exist in any sample many were missed and our player's contribution may be obscured. We can also see which combinations of players contributed most to a team's cause. Again this may be partly due to randomness or a hint of the intuitive interaction that is sometimes claimed to exist.

To see how we can use OptaPro's excellent figures, I've used Aston Villa's 2010/11 season and Liverpool's disappointing 2011/12. The pitch coordinates of each assisted shot at goal are used to produce an expectation for the chance to result in a goal, be on target or be blocked. Headers and shots are accounted for, headers become more difficult to save inside the six yard box, but shots become more potent thereafter. Primarily the figures should be used as an added descriptive feature of each team's season.

Aston Villa in 2010/11

Random variation will exist in all of the following results, but hopefully it will be less present in larger samples, so I've used only the most prolific chance creators. The reasons for existence of random variation in the actual outcomes compared to predicted outcomes are numerous. Luck, in the traditional sense will always exist, a defender slipping or a keeper mishandling, but a player may also be "lucky" in the way a pass arrives to him, allowing a perfectly executed goal attempt without the need to adjust or compromise his body shape.

It's inevitably tempting to judge a player merely by what actually happens on the pitch, but to do so would underestimate the degree to which runs of good and bad luck can occur and obscure the true midpoint of their abilities. If we also take note of the cumulative predicted expectancies we can see who may be riding a tide of good or bad fortune and begin to develop a more accurate longer term estimation of their abilities. 

The Reality and the Expectancy of Aston Villa's Main Shot Creations in 2010/11 


Aston Villa created 302 assisted chances in 2010/11 and over half of those were made by either Ashley Young or Stewart Downing, both of whom were to move on in the close season to Manchester United and Liverpool respectively. By moving back a stage in the goal creation process we begin to appreciate the magnitude of the creative loss to Villa caused by the departure of these two players.

Downing and Young created just over 80 chances each for their teammates in 2010/11, the latter was responsible for 10 assisted goals compared to just 7 for Downing. However, Downing gains ground and possibly overtakes his teammate when we look at the cumulative expectancies for chances created by the two midfielders over the season. An average Premiership attack would expect to score on average 11.5 goals from the chances that Downing helped to manufacture compared to just below ten for Young.

Fewer of the Downing chances than expected were on target and if as we suspect the actual taking of the chance lies more with the shooter than the provider, then this could indicate that Downing lagged behind Young in reality because he was unlucky in the player he was setting up with an opportunity. The provider of a decisive football pass may not have much control over who the recipient is or what they subsequently do with the chance.

Aston Villa's Main Creative Players During The 2010/11 Season


The number of chances a player creates is obviously important, but the quality of chance is also a factor. The average goal expectancy per chance shows that Downing was producing higher quality opportunities than the other three top creative Villa players. Albrighton wasn't far behind in quality, but did lag the future Liverpool player in terms of quantity. The combined quality of Downing, Young and Albrighton and their value to Villa begins to be laid bare. In hindsight the reasons behind the Midland team's struggles in 2011/12 were present in the figures from the previous season.

Liverpool in 2011/12

By far the most interesting case study from last season came at Anfield, where  Liverpool demonstrated the ability to create and miss chance after chance and Stewart Downing continued his career as a big money signing.

Disappointingly for Downing he finished 2011/12 without a Premiership goal or assist as Liverpool underperformed especially in front of goal. Downing created fewer chances than he had done in 2010/11 at Villa, just 56 compared to 85. This can be partly explained by the larger creative talent pool at his new club, although countering that Liverpool did create over 50% more chances than Villa had done in the previous year.


Downing - disappointing

However, in terms of the goal expectancy of those chances created, Downing was among Liverpool's best. The actual number of on target attempts recorded from the chances he created was broadly on terms with predictions, as were shots blocked, but the expectation for 6.5 goals created over the league season never materialised.

In contrast, Charlie Adam was second behind Downing with a cumulative goal expectancy for the chances he created of just under 6 and the recipients of his passes obliged by coming in with exactly that number. Every other of Liverpool's main creative players saw the chances they created under perform goal expectation as did inevitably the team as a whole, but no one created as much and reaped so little as Downing.

The Reality and the Expectancy of Liverpool's Main Chance Creators in 2011/12


Downing's average expectancy per chance ranked third among his teammates in his first season at Anfield, and numerically they were slightly inferior in quality compared to those he posted at Villa Park. Gerrard and Bellamy excelled at combining with teammates to produce on average high quality chances, although limited opportunities saw both fail to match the quantity of Downing chances.

Downing certainly wasn't quite the same creative force following his move, but neither was he the major disappointment that zero goals and zero assists would suggest. If we rate him primarily as a creator of chances rather than a goal scorer, then he produced enough high quality and quantity of chances to have registered multiple successful assists. In reality, Liverpool were let down by their finishing and no one more so than Downing the provider.  

Liverpool's Main Creative Players During The 2011/12 Season


One final extension of this type of analysis is to see how potent certain combinations of players were for their clubs. Randomness will predominate rather than repeatable talent in such small sample sizes, but as a descriptive tool over a season it can add value.

At Villa in 2010/11 Albrighton provided Downing with chances that resulted in two goals for the midfielder and Downing reciprocated. Their combined tally of four goals over performed against the predicted goal expectation of 2.6 goals, indicating levels of performance that were probably unsustainable in the longer term had Downing remained on the team. By contrast Darren Bent is undoubtedly a better than average finisher and both Young and Downing in the role of provider benefited from finding him with their final passes. 


Albrighton and Bent

Aston Villa Attacking Partnerships With At Least Seven Chances Created, 2010/11


In his new role of provider of chances at Anfield, Downing could be disappointed not to have broken his assist duck with a little help from Liverpool's main strikers. Both Adam and Gerrard saw two goals apiece arrive in the credit column courtesy of Suarez, while Downing's remained bare despite providing the Uruguayan with chances that were worth more in terms of goal expectancy. Downing feed Carroll with even more ammunition, but he too failed to convert any of the chances.

The combination figures also hint at mitigation for Downing's failure to register a goal, Adam and Suarez provided him with over twenty opportunities combined, but the goal expectancies of those chances were low with average individual expected success rates in the region of less than 1 in 25. Overall many of Dowing's goal attempts were from distance and despite numerous attempts his scoring expectancy over the season was only slightly more than three goals. A disappointment, but he was hardly spurning gilt edged opportunities.

Liverpool Attacking Partnerships With At Least Seven Chances Created, 2011/12


Real goals and real misses will always dominate football reports and debates, but with OptaPro's granular data, in depth analysis may be achieved by applying goal expectations and chance conversion projections.

Randomness is widely unappreciated as a factor in sporting contests, but its presence should not be ignored. What actually happens on the pitch, even over a considerable number of matches is merely a sample of a team's true ability and a slight flexing of a boot at the wrong time may mean that actual results may under or overestimate a player or team's true ability. Use of expectancy may give a more considered view as to what may occur in the future and prevent hasty conclusions being drawn from real, but potentially unrepresentative occurrences. 

Lastly, by recognising where the division lies between attempting and creating chances, we can avoid damning a creative player who provides a run of chances to poor or unlucky finishers or praising one who's fortunate enough to find excellent or currently lucky finishers on the end of his defence splitting passes.  

All images ©Action Images

Posted by Mark Taylor at 13:44


Post a comment