ABSTRACT
Performance indicators (PIs) are widely regarded as a means of objectively measuring sporting performance in terms of success (Hughes and Bartlett, 2002). However, their use assumes that performance can be understood, described and, by inference, improved through their measurement (Lames and McGarry, 2007). Whilst some sports have identified a correlation between frequency-based performance indicators and match outcome, e.g. basketball (Csataljay et al., 2009) and squash (Vučković et al., 2009), no consistent relationship has yet been identified for rugby union (Vaz et al., 2011). This may not be a surprise given the complex nature of rugby union and the variety of techniques and tactics that teams can choose to deploy from one match to the next. For example, van Rooyen et al. (2010) studied ruck frequency in the 2007 World Cup and found that different styles of play between the pool and knock-out stages of the competition led to winning. However, research in rugby union has suggested key PIs such as setpieces (Jackson and Hughes, 2001; Ortega et al., 2009; Sasaki et al., 2007) and territorial gains and losses (Hughes and Franks, 2004; Correia et al., 2011). However, very few attempts have been made to combine this information into a performance model, with the exceptions of Bracewell et al. (2003) and Jones et al. (2008). Potential weaknesses of these models are that they compare performance to a historic average and are subject to issues of volatility and the influence of confounding variables.
