Term 1 Lecture 2 Flashcards
(11 cards)
Whats a game in extensive form?
A game in extensive-form captures information about the order of moves, which is ignored in the strategic-firm model of a game in the previous section
Whats a game tree?
Diagram that describes the details of the sequence of decisions and actions players take when they play a game
- tree must have a unique initial node and a unique path from the initial node to every successor node
Why is ‘Nature’ included?
We want to allow for the possibility that there is some randomness in the game, so we introduce nature with node name 0 into the game tree
What is a pure strategy?
An instruction book for a player on how to play the game, must be complete and tell the player what to do at every point at which he/she makes a move
- even include apparently redundant instructions for when other players make mistakes
Whats a mixed strategy?
A random choice of a pure strategy for player
- player 1 -> (p,q,r, 1-p-q-r) for example
- key feature is that the randomisation occurs before the game is played and they stick to it
Whats a behavioural strategy?
When a player chooses probabilities for actions at each decision point
- key feature: randomisation happens at each decision node
Games so far have what property about information
Everyone playing the game knows everything about past events in the game
Whats an information set?
An information set is a collection of nodes with the property that, each node in the set has the same player’s name, each node in the set has the same actions available
- can be used to describe situations where a player’s past actions are hidden from others - like simultaneous
- IS can be used to describe situations where a player has private information
- IS can be used to describe situations where a player forgets what they knew previously
Existence of IS changes the definition of pure strategies:
A pure strategy for an incomplete information extensive form game is just an instruction for every info set where it must make a move
- player cant distinguish between decision nodes, so player must choose the same action for all nodes in the same information set
Backwards induction
- what it works for
- what it provides
- what it proves
- works for all finite, perfect info games and can be used to find credible NEs
- once we work through an extensive form game to find NE, have created pure strategy for every player in the game, called SPNE
- proves existence of NE in a finite perfect info game - called Zermelo’s theorem
Not all NE are reasonable
Sometimes the given NE can be based on a non-credible threat
- so SPNE might be a better solution concept as players play optimally at every decision point