A Least Squares Family of Cubic Curves with an Application to Golf Handicapping

Standard least squares procedures are developed for the simultaneous computation of several members of a family of curves from observed data. The nonlinear system which occurs is solved by a method of successive approximations. The application which led to this problem involves the question of fair golf handicapping. By smoothing the results of more than a million golf matches, played by computer using real scores reported by real golfers, it is found that the currently recommended system of the United States Golf Association leaves the stronger player with an extortionate advantage over the weaker in two man competition, whether match or stroke play. At least in principle a handicap system should equalize the chances of winning. A way of achieving this is suggested by the experiment.