Notational Analysis in Sport

Notational analysis is the whole process of systematically recording match events into a structured coding system and analyzing the resulting counts. A performance indicator is a specific output of that process: a function of the counts — typically a ratio, percentage, or composite — that summarizes an aspect of performance. Hughes and Bartlett stress that a valid performance indicator should relate to successful outcomes and be normalized to an appropriate exposure, so that it discriminates good from poor performance rather than merely reflecting volume of play.

There is no universal number; it depends on the sport and the variability of the indicator. The recommended approach is to compute a running mean of each indicator as matches accumulate and observe when adding another match no longer changes the average appreciably — the point of stabilization. Volatile indicators stabilize slowly and need more matches; stable ones settle quickly. Reporting a profile from too few matches risks treating ordinary match-to-match noise as a real tendency.

Because the entire analysis rests on the assumption that events have been classified consistently. If two competent observers would assign the same events to different categories, the counts — and every indicator built from them — are arbitrary. Establishing reliability through inter-observer and intra-observer agreement (Cohen's kappa for categories, percentage error for counts) verifies that the coding system has clear, applicable definitions. Categories with unacceptable error should be redefined or merged before the data are used for decisions.