Key performance indicators in Australian sub-elite rugby union

Area of Science:

• Sports Science
• Rugby Union Analytics
• Performance Indicators

Background:

• Sub-elite rugby union performance analysis requires understanding key factors influencing match outcomes.
• Previous research often focuses on professional leagues, leaving a gap in understanding sub-elite dynamics.

Purpose of the Study:

• To identify critical performance indicators (PIs) for success in sub-elite rugby union.
• To compare the predictive power of absolute versus relative data sets for determining match outcomes.

Main Methods:

• Utilized data from 17 PIs across 76 matches in the 2018 Queensland Premier Rugby Union season.
• Developed a random forest classification model to analyze win/loss outcomes based on absolute and relative data sets.

Main Results:

• The random forest model achieved 72.6% accuracy using absolute data and 78.1% accuracy using relative data.
• McNemar's test (p=0.84) indicated no significant difference in performance between absolute and relative data models.
• Positive associations were found between match outcome and relative kicks in play, meters carried, turnovers conceded, and initial clean breaks.

Conclusions:

• Match outcomes in Queensland Premier Rugby are predictable using both relative and absolute data sets.
• The predictive difference between absolute and relative data was less pronounced than in professional rugby.
• A successful game plan should emphasize an out-of-hand kicking game, outperforming the opposition in meters gained, and minimizing turnovers.