ABSTRACT
Soccer (or association football) is themost popular sport in theworld, with an estimated 3.5 billion fans and 250 million players worldwide (Sporty Desk, 2015). Betting on outcomes in soccer matches is also very popular, unsurprisingly, and the value of the soccer betting market in 2012 is estimated to be between £500 billion and £700 billion (Keogh and Rose, 2013). Consequently, statistical modelling of outcomes in soccer matches is popular among researchers, both in academia and industry, not only for the potential for financial returns but also for the challenges that suchmodelling presents. This is not to say that betting drives all research in statistical modelling in soccer andmany interesting problems relating to tactical questions (e.g. Wright and Hirotsu, 2003; Hirotsu and Wright, 2006; Brillinger, 2007; Tenga et al., 2010; Titman et al., 2015); team, player and manager rating (e.g. Knorr-Held, 2000; Bruinshoofd and Weel, 2003; Schryver and Eisinga, 2011; Baker and McHale, 2015); competitive balance and outcome uncertainty (e.g. Koning, 2000; Buraimo and Simmons, 2015); match importance (e.g. Scarf and Shi, 2009; Goossens et al., 2012); tournament outcome prediction (e.g. Koning et al., 2002; Groll et al., 2015) and tournament design and scheduling (e.g. Scarf et al., 2009; Goossens and Spieksma, 2012; Scarf and Yusof, 2011; Lenten et al., 2013) have been studied. Nonetheless, modelling results and scores, and other in-match outcomes, both straightforward (e.g. first player to score) and unusual (e.g. number of player cautions), motivated by the search for betting market inefficiency, have been a major motivational factor in the development of the state of the art. In this chapter, our aim is to describe the state of the art in the statistical modelling of match results and scores in particular.
