Mixed strategies: the graphical method

Mixed strategies: the graphical method

Spec 5.4 · Edexcel A-Level Further Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • A mixed strategy plays each strategy with a fixed probability, chosen at random each time.
  • Rose's expected pay-off against a column is the sum of probability of row ×\times entry.
  • For a 2×n2\times n game plot a line per column and find the highest point of the lower boundary.
  • For an n×2n\times2 game plot a line per row and find the lowest point of the upper boundary.
  • State every probability, including zeros; the value is the long-run average gain to Rose.

Mixed strategies

With no stable solution, play each strategy with a fixed probability. Rose's expected pay-off weights each entry by its probability.

When a game has no stable solution, a player who always uses the same pure strategy can be exploited. A mixed strategy chooses each strategy with a fixed probability, at random each time. Rose plays row ii with probability pip_i, with ∑pi=1\sum p_i=1.

Against any single column, Rose's expected pay-off is the sum of (probability of row ×\times entry). Both rows must be weighted: pp and (1−p)(1-p).

Her optimal pp maximises the worst of these, where they are equal.

0.10.20.30.40.50.60.70.80.9112345xy(1/3, 3)Column 1: 2 + 3pColumn 2: 4 − 3p
Rose's expected pay-off against each column, as p varies.

Worked example

For (5124)\begin{pmatrix}5&1\\2&4\end{pmatrix} Rose plays row 1 with probability pp. Find her optimal strategy and the value of the game.

Rose plays row 1 with probability pp. Against a column with entries 3 (row 1) and 7 (row 2), her expected pay-off is...

The 2×n2\times n graphical method

Plot Rose's expected pay-off against each column as a line. She picks pp at the highest point of the lower boundary.

If Rose has two rows and Colin has nn columns (n=2,3,4n=2,3,4), let Rose play row 1 with probability pp, 0≤p≤10\le p\le1. Each column gives a straight line for her expected pay-off against that column.

Colin will choose the column worst for her, so her guaranteed pay-off is the lowest line at each pp (the lower boundary).

She picks pp at the highest point of the lower boundary, where an increasing line meets a decreasing one. The height there is the value of the game.

Example: (614253)\begin{pmatrix}6&1&4\\2&5&3\end{pmatrix} gives lines 2+4p2+4p, 5−4p5-4p and 3+p3+p. They meet where 5−4p=3+p5-4p=3+p, so p=25p=\frac25 and the value is 175\frac{17}{5}.

0.10.20.30.40.50.60.70.80.91123456xy(2/5, 17/5)Column 1: 2 + 4pColumn 2: 5 − 4pColumn 3: 3 + p
The lower boundary peaks where columns 2 and 3 meet.

The lines 5−4p5-4p and 3+p3+p meet at which value of pp?

The other player's strategy

The two lines meeting at the peak show which columns Colin uses. Solve that 2×22\times2 game for him and give the others probability 0.

The two lines that meet at the peak tell you which of Colin's columns he actually uses, and only those two columns are played.

Solve the resulting 2×22\times2 game for Colin: let him play one of the two columns with probability qq, make Rose's expected pay-off equal for both rows, and solve for qq. His other columns have probability 0.

State the probability for every strategy, including those never played.

0.10.20.30.40.50.60.70.80.91123456qPay-off(1/5, 17/5)Row 1: 4 − 3qRow 2: 3 + 2q
Rose's pay-off against Colin's probability q of column 2, for each of her rows: the lines meet at q = 1/5, so he plays column 2 with 1/5 and column 3 with 4/5.
  • Column 100
  • Column 215\frac15
  • Column 345\frac45

Worked example

Continue the example using columns 2 and 3. Colin plays column 2 with probability qq.

Colin has 3 columns and only columns 1 and 3 are on the optimum. What probability does column 2 get?

The n×2n\times2 method

Work from Colin's side with probability qq on column 1, and find the lowest point of the upper boundary.

If Rose has nn rows and Colin has two columns, let Colin play column 1 with probability qq.

Each row gives a line for Rose's expected pay-off, and Colin picks qq to minimise the highest line: the lowest point of the upper boundary.

Example: (266143)\begin{pmatrix}2&6\\6&1\\4&3\end{pmatrix} gives 6−4q6-4q, 1+5q1+5q, 3+q3+q. The upper boundary is lowest where rows 1 and 2 meet: 6−4q=1+5q6-4q=1+5q, so q=59q=\frac59 and the value is 349\frac{34}{9}. Row 3's line is only 329\frac{32}{9} there.

Rose then plays only rows 1 and 2: 6−4p=1+5p6-4p=1+5p gives p=59p=\frac59.

0.10.20.30.40.50.60.70.80.91123456xy(5/9, 34/9)Row 1: 6 − 4qRow 2: 1 + 5qRow 3: 3 + q
The upper boundary is lowest where rows 1 and 2 meet.

In an n×2n\times2 game, which point on the graph does Colin choose?

Exam method and interpretation

Define pp or qq, write and plot the lines, find the optimum by equating two lines, state every probability, and interpret the value.

Follow the same steps every time.

Check your answer by substituting pp into every line: the lines for unused columns must give at least the value.

A value of 3.4, for example, means Rose wins 3.4 on average per play in the long run, not exactly 3.4 each time. A positive value favours Rose.

  1. 1

    Define

    state p or q and for which strategy

  2. 2

    Lines

    write each in simplified form and plot or sketch them

  3. 3

    Optimum

    highest point of the lower boundary, or lowest point of the upper boundary; equate the two lines

  4. 4

    State strategies and value

    both players, with zero probabilities

  5. 5

    Interpret

    the value is the average gain per play in the long run

Exam method for the graphical method

What does a value of 3.4 mean in context?

Try an exam question

Rose's pay-off matrix is (3104)\begin{pmatrix}3&1\\0&4\end{pmatrix}. Rose plays row 1 with probability pp. Find her optimal value of pp and the value of the game.

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Which expression is Rose's expected pay-off when Colin plays column 1?

1 mark, marked against the mark scheme

Revision notes