The Gold Standard of Zero-Sum Games | Separating Expected Game Value from One-Round Payoff Variability

Ask whether the reader needs the minimax expected value, a realised score, probability of loss, variance, a lower quantile or exploitability. One word such as performance cannot select all of them.

Write an evidence ledger before calculating: target quantity, population or object, unit, boundary, observation rule, transformation and decision limit. Mark every input as observed, stipulated, fitted or assumed.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘define the payoff question’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Next chapter

A realised payoff is one draw after actions are selected. Game value is an expectation under optimal mixed play and the specified information structure. A round need not equal the value.

Draw the mechanism in order. Separate the system state, the measurement, the mathematical summary and the claim. Then name one rival explanation and the additional observation that would distinguish it.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘separate game value from realised payoff’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

List every payoff and its probability under the strategy pair. Probabilities must sum to one and remain attached to the relevant action pair.

Keep the comparison fair by holding the population, interval, protocol and denominator constant. If one changes, state that the new number answers a different question even when the label looks familiar.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘write the outcome distribution’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

For payoffs +4 and −2 with probability one half each, E[X]=0.5×4+0.5×(−2)=1. The unit is payoff points per round in expectation.

Use two checks that can fail differently: reconstruct a total, test a limiting case, repeat the result from definitions or inspect a shared control. Two calculations that reuse one assumption are only one check.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘calculate the expected payoff’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Variance is E[(X−1)²]. The deviations are +3 and −3, so variance is 0.5×9+0.5×9=9 point² and standard deviation is 3 points.

Separate measurement from decision. A correct number can still be too local, uncertain or method-dependent for the intended action. Report the number with its conditions and say what stronger evidence would be needed.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘calculate variance around the same mean’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Distribution A is +4/−2 equally likely: mean 1, variance 9. Distribution B is +2/0 equally likely: mean 1, variance 1.

Preserve identity and order. Equal totals can hide different mechanisms when records come from different areas, cohorts, beam fields or payoff outcomes. Aggregate only after the reader’s question no longer depends on that identity.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘work the 9 versus 1 variance comparison’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Both averages are one point, but A produces a wider one-round spread and includes losses. The equal expectation settles neither short-horizon reliability nor a risk-sensitive preference.

Run a changed-context transfer check. Change one consequential condition while holding the remaining basis fixed, calculate again and explain why the result moves or stays stable.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘explain why equal expectation does not mean equal experience’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

An equilibrium probability is a prescribed distribution; observed frequencies in a finite sample fluctuate. Ten rounds need not reproduce the exact mixing ratio, and frequency mismatch alone does not establish exploitability.

Put the boundary of inference beside the answer. Name the system, representation and interval to which the result applies, then name the nearest tempting conclusion the evidence does not establish.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘keep equilibrium from empirical frequency’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Adding a constant to every payoff shifts the mean but leaves variance unchanged. Multiplying every payoff by c multiplies variance by c². State the unit and transformation before comparing dispersion.

Use uncertainty as structure. Vary the input most capable of changing the conclusion over a defensible range and show whether the ranking or decision changes.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘preserve payoff units and affine transformations’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Distribution A has variance 9 and a 50% probability of a negative payoff. Another distribution can have the same variance with a different loss probability, so the two summaries are not interchangeable.

Finish with a reproducible evidence package: source records, definitions, exclusions, identities, calculations, validation result and revision rule.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘distinguish variance from downside probability’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

The expected value can be positive while the probability of a negative round is substantial. Record P(X<0) separately when the question asks about a loss event.

Write an evidence ledger before calculating: target quantity, population or object, unit, boundary, observation rule, transformation and decision limit. Mark every input as observed, stipulated, fitted or assumed.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘calculate the chance of a negative round’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

For independent repeated rounds, the average payoff’s variance falls with the number of rounds. This concerns sampling variation around a fixed strategy pair; it does not repair strategic exploitation or changing opponents.

Draw the mechanism in order. Separate the system state, the measurement, the mathematical summary and the claim. Then name one rival explanation and the additional observation that would distinguish it.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘use repeated-round averages carefully’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Rounds can share hidden states, adaptive opponents or common shocks. Dependence can prevent the usual 1/n variance reduction. Preserve the time and information structure.

Keep the comparison fair by holding the population, interval, protocol and denominator constant. If one changes, state that the new number answers a different question even when the label looks familiar.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘recognise dependence across rounds’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Exploitability asks how much a best response can gain against a strategy. Volatility asks how realised payoffs vary under a specified opponent and distribution. A stable but exploitable strategy remains exploitable.

Use two checks that can fail differently: reconstruct a total, test a limiting case, repeat the result from definitions or inspect a shared control. Two calculations that reuse one assumption are only one check.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘keep exploitability separate from volatility’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

A game can have multiple equilibrium strategy pairs with the same value but different outcome distributions. Comparing their variance is an added selection question, not a revision of the minimax theorem.

Separate measurement from decision. A correct number can still be too local, uncertain or method-dependent for the intended action. Report the number with its conditions and say what stronger evidence would be needed.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘compare alternative equilibrium selections’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Classical zero-sum value does not encode a player’s utility for variance unless the payoff definition already does so. Adding a variance penalty creates a different objective that must be declared.

Preserve identity and order. Equal totals can hide different mechanisms when records come from different areas, cohorts, beam fields or payoff outcomes. Aggregate only after the reader’s question no longer depends on that identity.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘avoid adding risk preference to a zero-sum claim silently’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Variance may hide skew or extreme tails. Report relevant outcome bounds, loss probability or quantiles when the decision depends on them.

Run a changed-context transfer check. Change one consequential condition while holding the remaining basis fixed, calculate again and explain why the result moves or stays stable.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘use quantiles and bounds when variance is insufficient’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Multiply all A payoffs by 10: outcomes become +40 and −20, mean 10 and variance 900. Strategic ordering under positive scaling can remain while numeric variability changes by the square.

Put the boundary of inference beside the answer. Name the system, representation and interval to which the result applies, then name the nearest tempting conclusion the evidence does not establish.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘test a changed payoff scale’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

For outcomes −3, +1 and +5 with probabilities 0.25, 0.50 and 0.25, the mean is 1 and variance is 8. The same mean again coexists with another dispersion pattern.

Use uncertainty as structure. Vary the input most capable of changing the conclusion over a defensible range and show whether the ranking or decision changes.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘transfer to a three-outcome distribution’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Replace ‘the game value is one, so each round returns one’ with ‘the equilibrium expectation is one; in distribution A, one-round outcomes are +4 or −2 and the variance is nine’. Preserve the horizon and opponent.

Finish with a reproducible evidence package: source records, definitions, exclusions, identities, calculations, validation result and revision rule.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘repair the claim that value predicts every round’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Does minimax minimise variance? No. It optimises the guaranteed expected payoff under the model.

Can observed volatility prove a strategy is exploitable? No. Exploitability needs best-response analysis.

Does a positive value prevent losing rounds? No. The outcome distribution can include negative payoffs.

Write an evidence ledger before calculating: target quantity, population or object, unit, boundary, observation rule, transformation and decision limit. Mark every input as observed, stipulated, fitted or assumed.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘answer focused reader questions’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Next chapter

Use the minimax sources for the expected-value foundation, then follow the linked branches for exploitability, scenario robustness and broader probability reasoning.

Draw the mechanism in order. Separate the system state, the measurement, the mathematical summary and the claim. Then name one rival explanation and the additional observation that would distinguish it.

Apply the rule to the central fictional case with unrounded inputs, then label the result with its population, unit and protocol. Name one alternative explanation and the observation that would distinguish it.

For Zero-Sum Games, complete seven fields: target, object or population, unit, boundary, protocol, result and limit. For ‘continue through verified game routes’, a plausible number can otherwise hide a changed denominator or undocumented condition.

Change one condition that matters while holding the remaining basis fixed. Recalculate or re-inspect the evidence, state what stays comparable and name the stronger conclusion that still does not follow.

Close with two lines. The first states what the evidence establishes under the named conditions. The second states the nearest tempting conclusion that remains unsupported. Ask another reader to identify the condition that would make the first line false.

Contents · Previous chapter · Return to Zero-Sum Games