Four theorems — Aumann (1976), Nash (1950), Myerson (1991), and von Neumann–Morgenstern (1944) — fully derived and applied to the 2026 World Cup group stage. The rational draw is not a risk. It is the predicted equilibrium.
This part of the white paper serves two purposes simultaneously. For the reader with a background in economics or game theory, it provides a rigorous derivation of the equilibrium conditions that make rational non-competitive play the predicted outcome in specific 2026 World Cup matches. For the legal reader — the attorney preparing a CAS filing, the arbitrator evaluating evidentiary standards — it establishes that the incentive structure identified in Part I is not a qualitative observation but a quantitative result, derivable from first principles and reproducible by any competent analyst.
The distinction matters for the following reason. FIFA's expected defense against any CAS claim arising from a non-competitive match will be that match outcomes are inherently unpredictable and that inferring intent from a result is impermissible speculation. The mathematical argument in this part preempts that defense. It does not infer intent from a result. It predicts a result from an incentive structure — and then, when the predicted result occurs, identifies the structural cause. The difference between prediction and inference is the difference between science and speculation. This part establishes the former.
All four theorems are applied, not merely cited. Each derivation proceeds from the general statement of the theorem to its specific application to the 2026 World Cup calendar. Readers familiar with the relevant literature are invited to proceed directly to the application sections.
Chapter 2.2Robert Aumann's 1976 paper "Agreeing to Disagree," published in The Annals of Statistics, established one of the foundational results of modern epistemic game theory: that two rational agents with common priors cannot have common knowledge of their posteriors on any event if those posteriors differ.[1] The paper's implications for strategic interaction — particularly in settings where one agent has information the other does not — were developed extensively in the subsequent literature on mechanism design and information economics.
For the purposes of this analysis, the relevant application is not the agreement theorem itself but the broader framework it established for reasoning about asymmetric information in strategic settings. The core insight is this: when agents have different information sets, their rational strategies diverge — not because their objectives differ, but because the information available to them leads to different assessments of the probable consequences of their actions.
Define the information set of a team as the collection of facts available to it at the moment its final group match begins. Using standard notation from information economics:
The information advantage of Block 2 and Block 3 teams is not merely quantitative — it is decision-relevant in the precise technical sense. Information is decision-relevant when it changes the optimal strategy of a rational agent. The third-place cutoff — the minimum points and goal difference required to qualify as one of the eight best third-placed teams — is decision-relevant to any team that might finish third in its group.
For Block 1 teams, this cutoff is unknown. Their optimal strategy must account for all possible cutoff values, weighted by their probability. Because they cannot know whether a draw will be sufficient, the dominant strategy in most scenarios is to compete for the win — the guaranteed path to advancement for at least one team.
For Block 2 teams, the cutoff is known with high precision: five of twelve groups have concluded, establishing a table with five entries. Statistical analysis of the distribution of third-place finishes across historical World Cups allows Block 2 teams to form a high-confidence posterior estimate of whether the final cutoff will be above, at, or below the provisional benchmark.
For Block 3 teams, the cutoff is known exactly in the majority of scenarios: eight of twelve groups have concluded. The range of possible final cutoff values is dramatically narrowed, and in many specific configurations is reduced to a single value. The decision-relevant uncertainty that Block 1 teams face has been almost entirely resolved.
Let p(qualify | draw) be a team's posterior probability of qualification given a drawn result, and p(qualify | win) be the posterior probability given a win. For Block 1 teams, both posteriors are highly uncertain. For Block 3 teams with 2 points entering the final match, p(qualify | draw) is calculable with near-certainty from I₃. The information asymmetry between blocks is not a matter of degree — it is a structural feature of the calendar that transforms the strategic environment of the game.
John Nash's 1950 paper "Equilibrium Points in N-Person Games," published in Proceedings of the National Academy of Sciences, established the existence of equilibrium solutions in finite games — configurations of strategies from which no player has a unilateral incentive to deviate.[2] The Nash equilibrium concept is the foundational solution concept of non-cooperative game theory and is the appropriate framework for analyzing the strategic interaction between two teams in the final match of a World Cup group.
Consider the final match of Group F. Two teams — Team X and Team Y — enter the match. Define their strategy spaces and payoffs formally:
The payoff to each team is a function of its classification outcome, which determines the prize money received, the probability of advancing further in the tournament, and — where relevant — the performance bonuses embedded in commercial contracts. Define the payoff function as follows:
| Team Y: Compete | Team Y: Cooperate | |
|---|---|---|
| Team X: Compete | X wins: (V₁, V₃) or X loses: (V₃, V₁) Expected: (0.5V₁ + 0.5V₃, 0.5V₃ + 0.5V₁) |
X attacks, Y defends: X likely wins (V₁, V₃) with high probability |
| Team X: Cooperate | Y attacks, X defends: Y likely wins (V₃, V₁) with high probability |
Draw: both reach 3 pts, GD = 0 (V₂, V₂) — both qualify as best 3rd |
Step 1. Evaluate Team X's best response to Team Y: Cooperate.
If Y cooperates (plays for draw), X faces a choice: compete or cooperate. If X competes against a cooperating Y, X is likely to win (unilateral attack against passive defense). X wins → X gets V₁. But Y, having cooperated, reaches only 2 points (draw prevented) — Y is eliminated, getting V₃. So the outcome is (V₁, V₃).
If X cooperates while Y cooperates, the result is a draw. Both reach 3 points. Both qualify as best third (given c = 3, d = 0 and current GD = 0). Both get V₂. So the outcome is (V₂, V₂).
Since V₂ > V₃ (qualifying is better than being eliminated), Y prefers (V₂, V₂) to (V₁, V₃). But crucially: X also prefers (V₂, V₂) to (V₁, V₃) only if V₂ > the expected payoff from unilateral competition.
Step 2. The same calculation applies symmetrically to Team Y. By the identical argument, Y's best response to X: Cooperate is also Cooperate.
Step 3. Therefore (Cooperate, Cooperate) is a Nash Equilibrium: neither team has a unilateral incentive to deviate. If X deviates to Compete while Y plays Cooperate, X can win — but the expected gain is negative given V₂ > expected payoff from competition, as shown above.
Step 4. Uniqueness. The strategy profile (Compete, Compete) yields expected payoffs (0.5V₁ + 0.5V₃, 0.5V₁ + 0.5V₃). Since V₂ > 0.5V₁ + 0.5V₃ for the parameter values established above (the draw guarantee exceeds the expected value of a coin-flip between first and elimination), (Compete, Compete) is Pareto-dominated by (Cooperate, Cooperate) and is not an equilibrium when the cooperation option yields a guaranteed better outcome for both.
Roger Myerson's 1991 work Game Theory: Analysis of Conflict — and the underlying mechanism design literature for which he received the Nobel Memorial Prize in Economic Sciences in 2007 — developed the concept of incentive compatibility: a mechanism is incentive-compatible if rational agents, acting in their own self-interest, produce outcomes consistent with the mechanism's objectives.[3]
Applied to tournament design, the relevant question is: is the 2026 World Cup group stage mechanism incentive-compatible? Does it give all teams, at all times, an incentive to compete fully? The answer, as this chapter demonstrates, is no — for specific groups at specific points in the calendar.
A tournament mechanism is incentive-compatible if, for every team at every match, the dominant strategy is to maximize performance. This condition is satisfied trivially for knockout-stage matches: winning advances, losing eliminates, and there is no scenario in which losing is rational. The problem arises exclusively in the group stage — specifically in configurations where the format allows a non-winning result to serve a team's interests better than a winning result.
The probability that a given group in Block 2 or Block 3 contains a confirmed bottleneck match depends on the distribution of point outcomes across the first two rounds. Using historical World Cup data on points distributions after two group-stage rounds, and applying the bottleneck detection criterion above, the following estimates are derived:
| Block | Groups | Cutoff Knowledge | P(Bottleneck) per Group | Basis |
|---|---|---|---|---|
| Block 1 | A–E | None | ~0% (structural) | Condition (3) fails: cutoff unknown |
| Block 2 | F–H | High confidence | ~30–45% | Monte Carlo over historical point distributions |
| Block 3 | I–L | Exact (most scenarios) | ~45–60% | 8 groups resolved; cutoff range near-certain |
The von Neumann–Morgenstern (vNM) expected utility theorem, developed in Theory of Games and Economic Behavior (1944), established the axiomatic foundation for rational decision-making under uncertainty.[4] It proves that any rational agent — one whose preferences satisfy the axioms of completeness, transitivity, continuity, and independence — can be represented by an expected utility function, and that maximizing expected utility is the rational decision rule.
The application here is direct: given the payoff structure derived in Chapter 2.3 and the classification probabilities derived in Chapter 2.4, what is the expected utility of each strategy for a team in the bottleneck scenario?
The result is robust across a range of parameter variations. The key question is whether there exists a plausible parameter configuration under which competing is the dominant strategy. The answer is no, for the following reason:
The term stakeless zone — borrowed from Anglo-Saxon financial law — designates a situation in which a party to a transaction has no economic incentive to perform. In contract law, a stakeless counterparty is one whose expected payoff from performance is no greater than its expected payoff from non-performance. The concept is directly applicable to the bottleneck scenario.
In the context of a World Cup group match, a team enters a stakeless zone when — at any moment during the match — its classification outcome is already determined, regardless of any further action it takes. Once both teams in a bottleneck match are drawing at a score that satisfies the cooperation condition, both teams have entered the stakeless zone: any further offensive action risks changing a result that already gives both teams what they want.
The stakeless zone is not necessarily active from the first minute of the match. It becomes active — and the Nash Equilibrium of the game shifts — at the moment when the current score, combined with the known cutoff, satisfies the cooperation condition for both teams. This creates a predictable pattern:
| Match Phase | Score Status | Cooperation Condition | Expected Play Style |
|---|---|---|---|
| Minutes 1–15 | 0–0 | Met (0–0 draw qualifies both) | Cautious — neither team takes risks |
| Minutes 15–45 | 0–0 or 1–0 | Met if 0–0; disrupted if 1–0 | If 1–0: equalizer urgently sought by trailing team |
| Half-time | Any qualifying draw | Met | Coaches signal the equilibrium to players explicitly or implicitly |
| Minutes 60–90 | Qualifying draw held | Met — both teams fully rational | Stakeless zone fully active — observable defensive play, low VPI |
This predicted pattern has an important evidentiary implication: the behavioral signature of the stakeless zone — a sharp, sustained drop in offensive intensity in the second half, without corresponding physical fatigue — is precisely what the SIT's VPI and PBI indicators are designed to detect. Part V develops this in full. The mathematical prediction and the forensic detection method are complementary: the former establishes what should be observed if the equilibrium is played; the latter establishes whether it was.
Chapter 2.7The analytical results derived above rest on specific parameter assumptions — most importantly, the historical distribution of points after two group-stage rounds, and the probability that any given final-round match presents the bottleneck configuration. Monte Carlo simulation provides a method for testing the robustness of these results across a large number of randomly generated scenarios.
| Group | Block | P(Bottleneck) — Conservative | P(Bottleneck) — Central | Most Common Triggering Score |
|---|---|---|---|---|
| F | 2 | 28% | 38% | 2 pts vs. 2 pts entering final round |
| G | 2 | 26% | 35% | 2 pts vs. 2 pts entering final round |
| H | 2 | 30% | 40% | 2 pts vs. 2 pts entering final round |
| I | 3 | 42% | 55% | 2 pts vs. 2 pts — cutoff exact |
| J | 3 | 39% | 52% | 2 pts vs. 2 pts — cutoff exact |
| K | 3 | 41% | 54% | 2 pts vs. 2 pts — cutoff exact |
| L | 3 | 38% | 50% | 2 pts vs. 2 pts — cutoff exact |
| P(≥1 bottleneck, conservative) | 91.8% | |||
| P(≥1 bottleneck, central) | 97.2% | |||
The four theorems developed in this part, taken together, establish the following chain of reasoning:
From Aumann: The calendar creates a structural information asymmetry. Block 2 and Block 3 teams have decision-relevant information — the exact third-place cutoff — that Block 1 teams did not have. This information transforms the strategic environment of the match.
From Nash: In the bottleneck scenario — two teams with 2 points, cutoff at 3 points, common knowledge of the cutoff — the unique Nash Equilibrium is mutual cooperation: both teams play for the draw. Neither team has a unilateral incentive to deviate. The draw is not an accident. It is the predicted outcome of rational play.
From Myerson: The tournament mechanism is not incentive-compatible for Groups F through L in the bottleneck configuration. The mechanism rewards non-competitive play. The bottleneck probability per group ranges from 28% to 60%, yielding a cumulative tournament-level probability of 91.8% to 97.2% for at least one confirmed bottleneck match.
From von Neumann–Morgenstern: The expected utility of cooperation ($19.18M) exceeds the expected utility of competition ($17.30M) by $1.88M per team. The cooperation premium is positive and robust to parameter variation, except for teams that genuinely expect to win the tournament — a description that applies to at most two or three teams in any edition.
"The rational draw is not a behavioral prediction about the character of specific players or the ethics of specific coaches. It is a mathematical prediction about the incentive structure of a specific mechanism. Mechanisms produce outcomes. This mechanism was designed — inadvertently — to produce non-competitive outcomes in a predictable set of matches. The mathematics confirms this. The SIT data, in Part V, will verify it in real time."
— SIT Sport Intelligence Terminal, June 2026