Winner Determination: Instant Run-off

The series instant run-off winner determination rule was first used during the inaugural Computer Poker Competition in 2006. Though it has gone by many different names over the years, the idea remains the same: competitors aim to maximize the total number of series won across the opponents.

Under this winner determination rule, the outcome of each series of matches is either a win, loss, or tie. More specifically, if the total bankroll for agent A is higher than the total bankroll for agent B (during the matches they play against one another), agent A wins the series. If they are equal, it is a tie. Otherwise, agent B wins. The winner gets +1 point, the loser gets -1 point, and there are no points in the event of a tie.

In contrast to instant run-off over the total bankroll, which the competition has also used, the series instant run-off uses an instant run-off technique over these points. The total points for an agent is the sum of the agent's (-1/0/+1) score in all series it played with the remaining bots. Formally, the steps of the ranking algorithm are:

  1. Initialize the set of remaining bots S to be all bots.
  2. Consider all matches between bots in S, and compute the total points of each bot
  3. Define T to be the bots with the lowest total points. These will be ranked below the other remaining bots in S, and tie with each other. Remove T from S
  4. Repeat 2 and 3 until S is empty

Historically the series instant run-off has been used to favour the approximate equilibrium solution approach to poker AI as opposed to techniques that would learn online and attempt to maximize their bankroll.