On the optimal portfolio problem with partial information and related mean field games with relative performance criteria

Panagiotis Souganidis; Thaleia Zariphopoulou

arxiv: 2605.14519 · v2 · pith:RGXDALKTnew · submitted 2026-05-14 · 🧮 math.OC

On the optimal portfolio problem with partial information and related mean field games with relative performance criteria

Panagiotis Souganidis , Thaleia Zariphopoulou This is my paper

Pith reviewed 2026-05-15 01:38 UTC · model grok-4.3

classification 🧮 math.OC

keywords optimal portfoliopartial informationmean field gamerelative performancenonlocal PDEvalue functionindifference pricing

0 comments

The pith

Partial information on stock drift allows closed-form optimal portfolios for general utilities, with mean-field relative performance games reducing to a nonlocal quasilinear PDE.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper solves the single-agent optimal portfolio problem under partial information about the stock drift for general utility functions, obtaining regularity of the value function and closed-form expressions for the optimal investment processes. It then analyzes N-player games where agents interact through the distribution of their peers' wealth and passes to the mean-field limit to obtain a game with field equilibrium. The value of this equilibrium is represented as the single-player value plus a function that solves a nonlocal quasilinear partial differential equation in the space of measures. This framework applies particularly to separable couplings, and yields explicit solutions when the couplings depend only on the average peer wealth. A sympathetic reader would care because these reductions make large-scale competitive portfolio problems tractable by linking them to familiar single-agent optimization and PDE techniques.

Core claim

We solve the single agent problem for general utilities using a new approach that yields regularity of the value function and closed form expressions for the optimal processes. We consider a N player game in which players interact through the law of peer's wealth and study its mean field limit. This leads to a game with field equilibrium. We analyze the cases of separable couplings and general utilities, and represent the value of the game as a compilation of the single player problem and a function solving a non local quasilinear pde in the space of measures. When the couplings depend only on the average of peer's wealth, we derive explicit solutions and various regularity results for the 0

What carries the argument

The representation of the mean-field game value as the single-player value function combined with a solution to a nonlocal quasilinear PDE in the space of measures.

If this is right

Regularity of the value function and closed-form optimal processes for general utilities in the single-agent partial information setting.
Mean-field equilibria for relative performance games characterized by the nonlocal PDE for separable couplings.
Explicit solutions and regularity results when couplings depend solely on the average of peers' wealth.
Interpretation of the game values through elements of indifference valuation and arbitrage-free pricing.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The filtering of the hidden drift process could be combined with other stochastic control techniques for broader classes of market models.
Existence and uniqueness results for the nonlocal PDE might enable numerical approximations for computing equilibria in finite but large player games.
Similar reductions could apply to mean-field games with other types of relative criteria beyond wealth comparisons.

Load-bearing premise

The partial information on the stock drift and the form of the performance couplings (separable or average-based) permit the mean-field limit to be taken and ensure the nonlocal PDE admits sufficiently regular solutions.

What would settle it

Numerical computation showing that for a power utility function the candidate optimal portfolio process does not achieve the claimed value under the filtered probability measure, or that the proposed value function fails to solve the nonlocal PDE.

read the original abstract

We study optimal portfolio choice models in markets with partial information about the stock's drift. We solve the single agent problem for general utilities using a new approach that yields regularity of the value function and closed form expressions for the optimal processes. We consider a N player game in which players interact through the law of peer's wealth and study its mean field limit. This leads to a a game with field equilibrium. We analyze the cases of separable couplings and general utilities, and represent the value of the game as a compilation of the single player problem and a function solving a non local quasilinear pde in the space of measures. We interpret the findings using elements from indifference valuation and arbitrage free pricing, Finally, when the couplings depend only on the average of peer's wealth, we derive explicit solutions and various regularity results for representative cases.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a clean reduction of relative-performance mean-field portfolio games to a single-agent problem plus a nonlocal quasilinear PDE in measure space, with closed forms only when couplings depend on average wealth.

read the letter

The core advance is the single-agent treatment under partial information on the drift. They claim a new approach that produces regularity of the value function and explicit optimal processes for general utilities. That part looks like the strongest piece, since it moves beyond the usual filtering-plus-HJB route and directly yields usable expressions. The mean-field reduction is also neat: the game value is written as the single-player value plus a function that solves a nonlocal PDE on the space of measures. For separable couplings this seems to work without circularity, and when the interaction is only through average wealth they get explicit solutions plus regularity results. That is genuinely useful for anyone modeling competitive portfolio choice with relative performance criteria. The soft spot is the PDE side for general couplings. The abstract only claims analysis for separable cases and explicit forms for the average-wealth case; there is no indication of a full existence-uniqueness-regularity theorem that would cover broader interactions. If the nonlocal quasilinear PDE lacks classical solutions or the verification step fails to go through, the representation collapses. The stress-test note on this point holds up from what is shown. This is for specialists in stochastic control and mean-field games in finance. A reader already working on incomplete-information portfolio problems or relative-performance equilibria will find the framework worth looking at, even if they have to supply their own PDE analysis. I would send it to peer review because the single-agent closed forms and the reduction structure are worth referee scrutiny, though the authors should be asked to strengthen the general PDE results.

Referee Report

2 major / 1 minor

Summary. The paper studies optimal portfolio choice under partial information on the stock drift. It solves the single-agent problem for general utilities via a new approach that yields value-function regularity and closed-form optimal processes. It then considers an N-player game with relative performance via the law of peers' wealth, passes to the mean-field limit, and represents the game value as the single-player value plus a function solving a nonlocal quasilinear PDE on the space of measures. Explicit solutions and regularity results are derived when couplings depend only on average wealth; the findings are interpreted via indifference valuation and arbitrage-free pricing.

Significance. If the PDE analysis is completed rigorously, the decomposition of the mean-field equilibrium value into a standard single-agent problem plus an independent nonlocal PDE would provide a useful structural result for relative-performance mean-field games in finance. The new single-agent method and the explicit average-wealth cases could serve as templates for other partial-information control problems.

major comments (2)

Abstract and the mean-field section: the central representation of the game value as the single-player value plus a solution to the nonlocal quasilinear PDE requires existence, uniqueness, and C^{1,2} regularity of solutions to that PDE for general separable couplings. No such theorem is stated or referenced, and the abstract indicates that only separable couplings and average-wealth cases are treated explicitly; without this, the verification theorem and equilibrium property cannot be established.
Abstract: the claim of closed-form optimal processes and regularity for the single-agent problem with general utilities is asserted without derivation details or verification steps in the provided text. A concrete outline of the new approach (e.g., the key transformation or ansatz used) is needed to confirm that the regularity propagates to the mean-field representation.

minor comments (1)

Abstract: the sentence 'This leads to a a game with field equilibrium' contains a repeated article.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. We address each major comment point by point below, indicating planned revisions where appropriate to strengthen the presentation and rigor.

read point-by-point responses

Referee: [—] Abstract and the mean-field section: the central representation of the game value as the single-player value plus a solution to the nonlocal quasilinear PDE requires existence, uniqueness, and C^{1,2} regularity of solutions to that PDE for general separable couplings. No such theorem is stated or referenced, and the abstract indicates that only separable couplings and average-wealth cases are treated explicitly; without this, the verification theorem and equilibrium property cannot be established.

Authors: We agree that a complete verification theorem for the mean-field equilibrium requires existence, uniqueness, and C^{1,2} regularity of the nonlocal quasilinear PDE. In the manuscript we fully establish these properties, together with explicit solutions, when the couplings depend only on average peer wealth. For general separable couplings the representation is derived under the standing assumption that the PDE admits solutions of sufficient regularity (a common hypothesis in the mean-field-games literature). We will revise the abstract and the mean-field section to make this distinction explicit, add a precise statement of the regularity assumptions needed for the general separable case, and include a short remark on how the verification proceeds once those assumptions hold. This is a partial revision. revision: partial
Referee: [—] Abstract: the claim of closed-form optimal processes and regularity for the single-agent problem with general utilities is asserted without derivation details or verification steps in the provided text. A concrete outline of the new approach (e.g., the key transformation or ansatz used) is needed to confirm that the regularity propagates to the mean-field representation.

Authors: The new approach for the single-agent problem is developed in detail in Section 2 of the manuscript. It consists of a filtering step that produces an equivalent full-information problem whose effective drift is the conditional expectation of the original drift; the associated HJB equation then yields both the closed-form optimal portfolio process and the C^{1,2} regularity of the value function for arbitrary utilities. To address the referee’s request we will insert a concise outline of this filtering transformation into the revised abstract and introduction, thereby clarifying how the regularity carries over to the subsequent mean-field analysis. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper first solves the single-agent partial-information portfolio problem independently for general utilities, obtaining regularity and closed-form optimal processes via a new approach. It then represents the mean-field game value explicitly as the single-player value plus an independent function solving a nonlocal quasilinear PDE on the space of measures. No equation or step reduces the target equilibrium to a self-definition, a fitted input renamed as prediction, or a load-bearing self-citation chain; the PDE is introduced as a separate object whose solvability (under the stated separable or average-based couplings) supports the representation without circular dependence on the equilibrium itself. The derivation remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The abstract invokes standard stochastic control and mean-field game frameworks from prior literature without introducing new free parameters, ad-hoc entities, or non-standard axioms beyond domain assumptions on utilities and market filtrations.

axioms (1)

domain assumption Standard assumptions on utility functions being concave and increasing, and market dynamics driven by Brownian motion with partial observation of drift.
Required to set up the HJB equations and mean-field limit.

pith-pipeline@v0.9.0 · 5444 in / 1291 out tokens · 47032 ms · 2026-05-15T01:38:11.707438+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We derive a new master system, comprised by the master equation and an optimality condition... non-local quasilinear pde in the space of measures
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel echoes

?

echoes
ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

H solves the linear pde Ht + ½ c² Hzz + c Hzy + ½ Hyy = 0

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.