A series of games is played in a match, but the termination criteria is slightly more complex, and statistically justified, than, say, "best 4 out of 7".
The match is played until one player shows him or herself to be statistically significantly better than the other. Practically, one will probably have to set the desired margin of error quite large to keep the expected number of games down to a reasonable number. There are simple models (logistic?) to determine statistical significance.
Next, assume consecutive games are NOT independent. Incorporate into the model "streakiness", where a player having a "bad day" is likely to play several bad games in a row. Things might get game theoretically interesting for the win streak player to intentionally throw a game to make it less appear that the other player is having a bad day.
Inspired by chess matches, though I predict many matches will end in "ties", running out of time or money to continue the match.
No comments :
Post a Comment