Skip to content
QuantMax
QuantMax
  • Overview
  • Curriculum
    • FLUMental maths and numerical fluency
    • COMBCounting and combinatorics
    • PROBProbability
    • GAMEGames, decision theory and puzzles
      • 1Expected-value games

        • Pricing a game: EV, re-rolls and when to stop
      • 2Game theory

        • Game theory: dominance, mixing and the indifference condition
      • 3Poker and decision theory

        • Poker for traders: pot odds, ranges and bluffing frequency
      • 4Logic puzzles and multi-agent problems

        • Logic puzzles: information bounds and common knowledge
      • 5Classic brainteasers

        • Measuring and timing puzzles
        • Weighing, searching and strategy puzzles
      • 6Combinatorial games

        • Nim, symmetry strategies and who wins
    • MMMarket making
    • MKTMarkets and products

Practise

  • Question bank
  • Mental arithmetic
  • Market simulator
  • Arbitrage trees
  • Horse racing
  • Bid book
  • Screening tests
  • Mock papers

Reference

  • Formula reference
  • Search

Your record

  • Review queue
  • Progress
  • Leaderboard
  • Profile
  • Invite friends
AccountSend feedback
  1. Curriculum
  2. /Trading and market making
  3. /Games, decision theory and puzzles
  4. /Expected-value games

Pricing a game: EV, re-rolls and when to stop

GAME · Chapter 1·13 min read·Asked at Jane Street, SIG, Optiver, IMC

After this lesson you should be able to

  • Price a game with an option to continue by backward induction.
  • Solve a self-referential expectation of the form X=f(X)X = f(X)X=f(X).
  • State the optimal-stopping threshold rule and the secretary problem’s answer.

A trading interview is full of games whose value you are asked to quote. Almost all of them reduce to one idea: at every decision point, compare what you have against what continuing is worth, and take the larger. That comparison is the whole of optimal stopping, and it is also exactly how an American option is priced.

Proposition 1.1

The continuation value

Define VVV as the value of the game from here on if you keep playing optimally. At any point where you may stop, the value is max⁡(what you hold now, V)\max(\text{what you hold now},\ V)max(what you hold now, V). Compute VVV first, then the policy — "accept anything above VVV" — falls out for free rather than being guessed.

Holds when

  • With finitely many rounds, work backwards from the last one.
  • With infinitely many, VVV satisfies an equation in terms of itself.
  • The threshold always equals the continuation value, never the mean of the distribution.
Roll 1 → re-roll3.5Roll 2 → re-roll3.5Roll 3 → re-roll3.5Roll 4 → keep4Roll 5 → keep5Roll 6 → keep6
Figure 1.2 · The continuation value divides keep from re-roll. With one re-roll available, a fresh die is worth 3.5. Faces 1–3 should be exchanged for that continuation value; faces 4–6 are worth keeping. Averaging the six displayed values gives 4.25 before the first roll.

Example 1.3

One re-roll

You roll a fair die and may re-roll once, keeping the second value if you do. What is the game worth?

Show the worked solutionHide the worked solution

Worked solution

  1. Formula
    V=16∑f=16max⁡(f, Vcont),Vcont=3.5V = \frac{1}{6}\sum_{f=1}^{6}\max(f,\ V_{\text{cont}}), \qquad V_{\text{cont}} = 3.5V=61​f=1∑6​max(f, Vcont​),Vcont​=3.5
  2. Substitute
    max⁡(f,3.5)=3.5,3.5,3.5,4,5,6\max(f, 3.5) = 3.5, 3.5, 3.5, 4, 5, 6max(f,3.5)=3.5,3.5,3.5,4,5,6
  3. Solve
    V=25.56V = \frac{25.5}{6}V=625.5​
  4. Answer
    V=4.25V = 4.25V=4.25

Sanity check. Re-roll on 1, 2 or 3; keep 4, 5 or 6. The threshold is 3.5 because that is what a fresh roll is worth, not because it is the midpoint.

Derivation 1.4

Infinite horizon: solving X=f(X)X = f(X)X=f(X)

When the game can go on forever, condition on the first step and write the value in terms of itself.

  1. V=16∑f=16max⁡(f, V)V = \frac{1}{6}\sum_{f=1}^{6}\max(f,\ V)V=61​f=1∑6​max(f, V)

    Unlimited re-rolls: the continuation value is the game itself.

  2. Guess 4<V<5⇒V=4V+5+66\text{Guess } 4 < V < 5 \Rightarrow V = \frac{4V + 5 + 6}{6}Guess 4<V<5⇒V=64V+5+6​

    Faces 1 to 4 are re-rolled; 5 and 6 are kept.

  3. 6V=4V+11⇒2V=116V = 4V + 11 \Rightarrow 2V = 116V=4V+11⇒2V=11
V=5.5V = 5.5V=5.5

Proposition 1.5

Always verify the bracket

The step above *assumed* the threshold sat between 4 and 5 in order to write the equation. Having solved it, check: V=5.5V = 5.5V=5.5 means you re-roll anything at or below 5, so faces 1 to 5 are re-rolled and only a 6 is kept. That contradicts the assumption, so the equation must be rewritten.

Holds when

  • Correct version: V=5V+66⇒V=6V = \frac{5V + 6}{6} \Rightarrow V = 6V=65V+6​⇒V=6.
  • Which is right: with unlimited free re-rolls you simply wait for a six, so the game is worth 6.
  • The lesson is that a self-referential equation encodes a policy, and the policy must be consistent with the answer.

The rest of this lesson is in Premium

You have read the opening. 10 more sections follow, including 4 worked examples and 3 quick checks.

Start the free 7-day trialSign in

Nothing is charged for 7 days, and you can cancel before then. Or read Arithmetic that survives a clock in full, free.

Game theory: dominance, mixing and the indifference condition →
On this page
  • The continuation value
  • The continuation value divides keep from re-roll
  • Worked example — one re-roll
  • Infinite horizon: solving X=f(X)X = f(X)X=f(X)
  • Always verify the bracket

QuantMax · 141 lessons · 1342 questions · c5c0caa

  • Premium
  • Arbitrage trees
  • Horse racing
  • Invite friends
  • Account
  • About QuantMax

Firm names identify publicly reported question patterns and nothing more. QuantMax is not affiliated with, endorsed by, or recruiting for any firm named in the curriculum. Everything you do in lessons and the question bank is kept to your account.