Scribd
Upload
57 views2 pages

Systematic Approach To Solving Games

This document outlines a systematic approach to solving game theory problems: 1) Represent the game using normal, Bayesian normal, or extensive form based on whether it is simultaneous or sequential and involves multiple types of players. 2) Determine each player's strategy set based on their information sets. 3) Eliminate strictly dominated strategies when possible. 4) Choose the appropriate equilibrium concept such as Nash equilibrium, trembling-hand perfect Nash equilibrium, Bayes-Nash equilibrium, or subgame perfect Nash equilibrium. 5) Solve for the equilibrium by finding best responses, checking for mixed strategies, and applying backward induction or iterative elimination of dominated strategies as needed.

Uploaded by

Diane Yang
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
57 views2 pages

Systematic Approach To Solving Games

This document outlines a systematic approach to solving game theory problems: 1) Represent the game using normal, Bayesian normal, or extensive form based on whether it is simultaneous or sequential and involves multiple types of players. 2) Determine each player's strategy set based on their information sets. 3) Eliminate strictly dominated strategies when possible. 4) Choose the appropriate equilibrium concept such as Nash equilibrium, trembling-hand perfect Nash equilibrium, Bayes-Nash equilibrium, or subgame perfect Nash equilibrium. 5) Solve for the equilibrium by finding best responses, checking for mixed strategies, and applying backward induction or iterative elimination of dominated strategies as needed.

Uploaded by

Diane Yang
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Systematic approach to solving games

1) Write down the game (when feasible). Fill in payoffs for each cell of the matrix or terminal node.
a) Normal form if the game is simultaneous move without multiple types.
b) Bayesian Normal Form if the game is simultaneous move with multiple types.
c) Extensive Form if the game is sequential.
2) What is each players strategy set?
a) First, determine what each players information sets are.
b) A players strategy must specify what action to take in each of the players information sets.
3) When possible, eliminate strictly dominated strategies. 1
a) Do not eliminate weakly dominated strategies.
b) Do not eliminate strictly dominated actions (yet).
4) Determine what equilibrium concept to use:

Simultaneous move

Multiple
types

Single type

No weakly
dominated
strategies

NE

Weakly
dominated
strategies

THP NE

BNE

Sequential move

Perfect
info

Backwards
Induction

Not perfect
info

Single
type

SPNE

Multiple
Types2

PBE

5) Solve. Remember that an equilibrium specifies the players strategies (their actions at all their
information sets, not just the information sets that are reached in equilibrium). If asked for the
equilibrium outcome, you can just specify what actions get played in equilibrium, and the resulting
payoffs.
a) Nash equilibrium:
i) Find each players best-response function, which gives that players optimal strategy, given
the strategies played by the other players. (In the normal form, this entails simply
underlining best responses.)
ii) Solve best-response functions simultaneously to get the set of pure-strategy Nash equilibria.
(In the normal form, this is just finding cells where all payoffs are underlined.)
iii) If there is more than one pure-strategy Nash equilibrium, look for equilibria in mixed
strategies: if a player is mixing, she must be indifferent between all pure strategies that she
plays with positive probability.
1
2

This could theoretically cause problems if youre solving PBE without refinements. I wouldnt worry about it, though.
You may want to use PBE in cases where there are not multiple types. PBE more generally allows us to deal with beliefs.

b) Trembling-hand perfect Nash equilibrium:


i) Solve for the set of Nash equilibria, as above.
ii) Eliminate weakly dominated strategies from consideration either as pure strategies or as
parts of mixed strategies (but do not iteratively eliminate weakly dominated strategies). 3
c) Bayes-Nash equilibrium:
i) Solve exactly as you would solve for Nash equilibria, but expected payoffs will be a
function of the probabilities over types.
d) Backwards induction Nash equilibrium:
i) Find and mark best responses at the terminal decision nodes (the decision nodes closest to
the terminal nodes). When there are ties, consider all tied options.
ii) Trim the tree: Given the best responses at the terminal decision nodes, find best responses
at the decision nodes just before the terminal decision nodes.
iii) Iterate until you have found best responses at all nodes.
e) Subgame perfect Nash equilibrium:
i) Identify all subgames. A subgame starts at a singleton node, contains all successor nodes,
and does not split information sets.
ii) Find the set of Nash equilibria in the terminal subgames (the subgames closest to the
terminal nodes). When there are multiple Nash equilibria, consider all of them.
iii) Using the results from (ii), identify the payoffs from reaching each terminal subgame.
Using these payoffs, find the set of Nash Equilibria in the subgames one step before the
terminal subgames.
iv) Iterate using this type of backward induction by subgame until you have solved for the
whole game.
f) Perfect Bayesian Equilibrium (consists of both strategies and beliefs):
i) Eliminate strictly dominated actions in any information set.
ii) Propose equilibrium strategy for one player (usually the sender, who has multiple types).
iii) Compute other players beliefs based on the proposed play.
iv) Calculate the best responses of other players, conditional on their beliefs.
v) Check if any type of the first player has an incentive to deviate. (When necessary, find the
set of off-path beliefs consistent with this play.)
vi) Repeat until you have checked each possible pure strategy for the first player (not involving
strictly dominated actions).
vii) Apply Cho-Kreps Intuitive Criterion, if asked to do so: if a message is equilibrium
dominated for a type, then assign probability zero to the message coming from that type.

This isnt 100% correct, but should suffice for our purposes.

You might also like