Elo Rating Calculation for Games (from Chess)

Recently I’ve implemented a rating system for an Android game and found the Elo rating system from chess to be a very good fit and easy to implement. The idea of the ELO system is to increase or reduce points proportional to the strength difference of the players. A player with more points is expected to win, thus if he only reaches a draw will loose points and the weaker player will gain some.

For the examples we use a K factor of 32, which is used by FIDE for rating beginners (a lower K factor is used for masters and will reduce the number of points won/lost).

This is the algorithm implemented in Java for calculating the ratings:

// Returns an array of two doubles: [0] is the new rating for player A, [1] for player B
//
// Parameters:
//   status: 0 = draw, 1 = player A won, 2 = player B won
//   RA, RB: the players points before the game
private double[] calcElo(int status, double RA, double RB) {
	double QA = Math.pow(10, RA/400);
	double QB = Math.pow(10, RB/400);

	// set the desired k factor
	double k = 32;

	double EA = QA / (QA + QB);
	double EB = QB / (QA + QB);	    	

	double SA = 0.5; // draw
	double SB = 0.5; // draw
	if (status == 1) {
		SA = 1;
		SB = 0;
	} else if (status == 2) {
		SA = 0;
		SB = 1;
	}

	double[] res = new double[2];
	res[0] = RA + (k * (SA - EA));
	res[1] = RB + (k * (SB - EB));
	return res;
}