No label defined (Q4697001)

When \(n\) individuals enter a tournament, it is an easy matter to determine the `best' by playing a simple `knock-out' competition. A total of \((n-1)\) games is thus required. However, many contests are team games so that determination of the unique winner (as opposed to the best team) is rather more difficult and there may be criticism of `unfair groupings' and so forth. This `tournament problem' and the associated matter of weighting designs has received a certain amount of attention. In particular \textit{R. C. Bose} and \textit{J. M. Cameron} [J. Res. Natl. Bur. Stand., Sect. B 71, 149-160 (1967; Zbl 0157.032)] produced calibration designs when the total number of players did not exceed 50. Their `balanced weighting designs' (BWD) were constructed according to the criteria: (i) a total of \(n\) players wish to compete; (ii) a game consists of a contest between two teams; (iii) such teams are of equal size, each consisting of \(m\) players (where, necessarily, \(2\leq m\leq n/2)\); (iv) order of players within a team is immaterial; (v) a player must play the same number of times with each team-mate; (vi) a player must play the same number of times against every opponent; (vii) subject to the above criteria, the number of matches is a minimum. The present work differs from the above in that to take account of any `correlation' of playing ability between one contestant and another, all feasible games are played. In addition the approach adopted is somewhat less sophisticated than in [J. Res. Natl. Bur. Stand., Sect. B 71, 149- 160 (1967; Zbl 0157.032)].