A feature of week five in the English Premier League was the occurrence of relatively rare, yet significant events, which invariably impact on the chances of a side winning an individual match.
Penalty kicks were awarded in the Leicester vs. Manchester United game, as well as in the Everton, Palace match, providing three of those four teams with a high quality chance that is converted nearly 80% of the time.
The Leicester match also witnessed a late red card for Manchester United, but it was a day earlier at Swansea where a dismissal greatly eased Southampton’s task against their high flying hosts.
The clash between the two teams just below leaders Chelsea in the table had been highlighted as a contest between a team in Southampton, which had produced underlying chance creation statistics which wasn’t out of step with their lofty position. And a Swansea team which was possibly flattered by their second spot in the table at kick off.
There was therefore the possibility that Swansea’s prospects may have been over stated in relation to their opponents. Of course, there is no certainty that this slight edge would guarantee a Saints victory at Swansea. But, if league position is used as the sole performance criteria, as is often the case, such potentially mispriced games are likely to be profitable in the long term for bettors.
Elsewhere, Villa’s new found defensive solidity, which was also questioned last week, came to an abrupt end at home to Arsenal, while Newcastle and WBA perfomed as mid table teams rather than relegation candidates, as each took points from a home draw with Hull and an away win at Tottenham, respectively.
|Home Team||Home Win %||Draw %||Away Win %||Away Team||Actual Score||Most Likely Score|
Red card costs Swansea
At the Liberty, Swansea, who based on total shot ratio may have been fortunate to receive their impressive return from their early season efforts, were on the receiving end of a more transparent dose of bad luck on Saturday.
Wilfried Bony’s red card was a possible outlier. Strikers, on average commit fouls in less dangerous areas of the pitch and are also more likely to be on the receiving end of poor challenges. Therefore, they tend to be allowed to commit more fouls than either midfielders or defenders before they recieve a yellow card.
Bony’s second challenge was undoubtedly poor, but both offences were committed around the centre circle and were his only fouls of the game. So two yellow cards, to make a red from two challenges might be considered unlucky, especially when committed by a primary striker in non-dangerous areas of the field.
The second yellow card was shown in the 39th minute and despite the rare and often unforgettable occasions where ten men triumph against eleven, conceding a man to the opposition is, on average, a severe obstacle to overcome.
Ten men concede more goals than they would have had they not received a red card and they also score fewer. In the table below I’ve taken data from the 2010/11 Premiership season to show this in action. The share of goals enjoyed by the side that would see red in the game fell from 40% before the card to just below 30% after the dismissal.
|Number of Players||Goals scored by red carded team||Goals conceded by red carded team||Percentage of match goals scored by red carded team|
|11 v 11||28||42||40%|
|11 v 10||15||36||29%|
The average time of the red cards in 2010/11 was the 64th minute, which is typical for the Premier League, so a side, on average experienced the man disadvantage for about 30 minutes. Swansea had to play for nearly twice that length of time against Southampton, highlighting their disadvantage.
It is impossible to predict a red card with any degree of certainty before a game, although the 40% share of goals enjoyed by the red carded side in the table above, does indicate that red cards are, on average, shown more frequently to the inferior side. Often with good reason because they are forced into more tackles in more dangerous situations.
However, we can use a system similarly to the way in which pre game goal expectations for each team declines with time, along with the way in which the proportion of goals are scored and allowed alters upon the issue of a red card, to run a Poisson based simulation for the remainder of a dismissal affected game.
Prior to the start of the game the consensus of Swansea’s chances of winning was around 39%, just before Bony’s dismissal it was around 35%, immediately after, it fell to roughly 10%, which has obvious implications for in play betting.
The shot count also illustrates how the game shifted on the red card. Prior to the 39th minute, Swansea had out shot Southampton 5 to 1, afterwards the visitors enjoyed a superiority of 10 to 0, during which they claimed the game’s only goal.
We had legitimate doubts about Swansea’s high league position based on the rate at which they were creating and allowing chances prior to Southampton’s visit, but the further decline in this data will now be partly down to the red card they received in that game – which should be accounted for.
Similarly, Leicester were expected to see an improvement in results once they had negotiated matches against Arsenal, Everton and Chelsea and they duly won at Stoke, but they are unlikely to be awarded regular spot kicks, never mind the two they received in a subsequent victory against van Gaal’s Manchester United team on Sunday.
Therefore, the persistent fascination in assessing the past and attempting to predict future performance in soccer, not only lies in avoiding natural random variation as repeatable talent, while allowing for the differing ability levels of opponents faced, especially in small sample sizes. Bettors also need to account for the disproportionally large impact of rare and often punitive sanctions – red cards and penalties - that can affect any individual game.
The most likely current EPL position based on chances faced and created after five games
|Team||Ranking based on chances faced & created (lower=better)||Ranking of opponents to date (higher=easier)||Current league position|
In the table above, I’ve firstly ranked the teams by their weighted league position in 10,000 chance-based simulations of all 50 games played in the Premier League to date. This potentially highlights a side that might have been lucky or unlucky in converting their chances and saving those of their opponents.
Villa’s current position of 3rd is still a flattering reward for their shot based attempts and with the points from their narrow wins already banked for the current season, their league position in the immediate future is likely to remain an overstatement of their core abilities.
The most fascinating of the EPL teams remains Newcastle. They are almost equally likely to be found in any position from 3rd to 16th based on shot simulations, giving them a weighted average that is comfortably in the top ten.
The second column ranks the quality of opposition faced to date. For example, Arsenal’s current position of 4th is consistent with the quality and quantity of shots attempted in their games, as they rank 2nd in simulations, narrowly behind Chelsea. But this has been achieved against the easiest, average set of opponents.
In addition, as the data on red cards and penalty kicks shows, bettors should be mindful of which teams have experienced more luck than others.
Interestingly eight of the 20 EPL teams are yet to play in a match, which has contained a spot kick. Leicester are a net two penalties to the good, whilst Manchester United and Spurs have each faced two and received none in return. Palace, Hull and Sunderland are also a net minus one spot kick, while Burnley, Liverpool, QPR, WBA and West Ham have each received one and conceded none in return.
Red cards have been issued to Hull, both Manchester sides, Swansea, Palace, Spurs, West Ham and Newcastle and the teams to benefit, albeit sometimes only for a few minutes, were Leicester, Chelsea, Stoke, Arsenal, Southampton, West Ham and Villa.
The sides with the numerical advantage took over 70% of the game’s goal attempts during play against the ten men, as well as scoring a similar proportion of the total goals.
A single ratings figure is often useful in distinguishing between sporting sides for betting purposes, but the context in which these core stats have been achieved in relation to rare, but significant events, as well as strength of schedule, should also be considered.