← All quantitative finance open problems

Market microstructure / Construction problem

Endogenous Market Efficiency Under Long-Memory Order Flow

Connects nonequilibrium statistical physics and market microstructure by asking how adaptive liquidity transforms persistent institutional trading into diffusive prices, with impact and efficiency emerging from participant interactions.

OpenEditorially scopedFramework needs definitionCompiled 2026-11-09

Seed catalog · Showing the bundled record.

Problem statement

Research program: construct and analyze an endogenous market in which institutional order splitting produces persistent signed order flow with Cε(τ)∼cτ−γC_{\varepsilon}(\tau)\sim c\tau^{-\gamma}, c>0c>0 and 0<γ<10<\gamma<1, while strategically adapting liquidity providers generate asymptotically Brownian transaction prices. Derive quotes, inventory adjustment and price impact from agent objectives and interaction rules, without postulating an efficient-price process or tuning a prescribed impact kernel to cancel order-flow persistence. A central target is the functional limit (p⌊nt⌋−p0)/n⇒DBt(p_{\lfloor nt\rfloor}-p_0)/\sqrt n\Rightarrow\sqrt{D}B_t with D>0D>0, for the resulting zero-drift price observable. The new contribution must extend the assumptions or conclusions of existing propagator, nonlinear-impact, Hawkes and arbitrageur microfoundation models.

Assumptions and scope

  • Specify institutional metaorder sizes, arrival and splitting rules, and establish persistent order flow from these primitives; correlation decay alone does not determine the process law or imply a central limit theorem.
  • Fix liquidity providers' information, utility or profit criteria, inventory penalties, capital constraints, competition and learning rules, together with the equilibrium or adaptation concept.
  • Define transaction price versus midpoint or log-price, trade time versus calendar time, matching rules, tick size and spread. Derive zero drift under stated symmetry conditions or explicitly center by a proved drift before taking the scaling limit.
  • Seek an invariant regime for returns and the relative order-book/liquidity state, including liquidity-provider inventories where appropriate; the absolute level of a nondegenerate diffusing price is not stationary.
  • State initial-law, moment and dependence conditions and prove process-level convergence, for example in Skorokhod D([0,T]) with its J1 topology for each finite T, including tightness and a finite positive diffusion coefficient. Linear variance growth or white returns alone does not establish this limit.
  • For no-manipulation results, specify predictable self-financing round trips, information, admissibility, fees, impact accounting and terminal liquidation. A Brownian limit is distinct from exact martingale efficiency or nonnegative expected execution costs at finite horizons.
  • The restricted propagator reproduction must fix volume weights, convergence of historical sums or price increments, and any independent noise contribution; an added random-walk component can dominate a subdiffusive impact component.

What would settle it?

First milestone (known mathematics): audit the five cited papers at theorem level and formalize βc=(1−γ)/2\beta_c=(1-\gamma)/2 in a restricted discrete-time propagator model, recording its assumptions and noise convention. Original-research targets: T1, construct an economically specified endogenous market and prove existence of a stationary relative-state regime consistent with its equilibrium or adaptation rules; T2, derive the impact response and its scaling from agent strategies; T3, prove the stated nondegenerate Brownian functional limit, including tightness; T4, characterize a parameter/behavior class where diffusion persists, or exhibit failures; T5, establish nonnegative expected impact costs for admissible round trips under explicit accounting. Lean proofs can certify rigorously stated claims; reproducible stochastic simulations support conjecture discovery and counterexample searches. Empirical order-book comparisons assess modelling assumptions separately from the mathematical resolution.

Selected for a theoretical or computational target. A successful calibration or numerical benchmark alone does not establish the stated mathematical guarantee.

Status and evidence

Exploratory long-horizon research program, not a single established conjecture or an unexplained phenomenon in every model. Bouchaud et al. (2004), section 3.2, equation (23), already derive the critical propagator balance 2β+γ=12\beta+\gamma=1; reproducing it is not a novel result. Their equation (20) also contains an independent diffusive-noise contribution, so a subdiffusive impact component need not make total prices subdiffusive. Lillo–Farmer document compensation of persistent order signs by volume and liquidity fluctuations. Sato–Fujiwara–Kanazawa's v5 revision (5 November 2026) gives an exactly solvable nonlinear-impact model mapped to a Levy walk and reports diffusive prices under square-root impact. Muhle-Karbe et al. (2026) obtain Hawkes-based scaling limits and relations among flow, impact and rough volatility, using no-arbitrage constraints. Bonart (2026) microfounds diffusive impact with arbitrageurs under asymptotically complete aggregate private information and suitable convergence to a stationary limit. These results substantially overlap the motivation: novelty must be located through a theorem-level audit and explicit extensions, such as finite-information, inventory-constrained adaptive liquidity with proved equilibrium and a robust functional limit. Such extensions are candidate directions, not certified gaps after an exhaustive literature review.

Preliminary source check; not an expert-verified open status or an independent audit of proposed proofs. The compilation date is not a literature-review cutoff.

Sources and further reading

  1. Bouchaud, Gefen, Potters & Wyart: Fluctuations and response in financial markets: the subtle nature of random price changes (2004) ↗background · source checked 2026-11-09

    HTML sections 3.1–3.2: equations (12)–(13), (20) and (22)–(23); section 3.4 separates impact from independent noise.

    Model and scaling argument inspected directly. The known critical-exponent relation concerns a prescribed propagator; independent diffusive noise requires separate accounting.

  2. Lillo & Farmer: The long memory of the efficient market (2004) ↗background · source checked 2026-11-09

    arXiv v2 abstract: persistent order signs compensated by transaction-size and liquidity fluctuations.

    Empirical adaptive-liquidity motivation, not a general strategic-equilibrium functional limit theorem.

  3. Sato, Fujiwara & Kanazawa: Why financial prices exhibit Brownian motion despite predictable order flow: An exactly solvable nonlinear price impact model (2026 revision) ↗partial progress · source checked 2026-11-09

    arXiv v5 abstract and submission history; revised 5 November 2026.

    Reports a nonlinear Lillo–Mike–Farmer extension mapped to a solvable Levy walk, with diffusion under square-root impact. Full theorem scope and convergence topology remain to be audited.

  4. Muhle-Karbe, Ouazzani Chahdi, Rosenbaum & Szymanski: A unified theory of order flow, market impact, and volatility (2026 preprint) ↗partial progress · source checked 2026-11-09

    arXiv v2 abstract: core/reaction Hawkes flows, scaling limits and no-arbitrage implications.

    Existing microstructural scaling theory; do not advertise its covered regimes as newly unresolved.

  5. Bonart: Diffusive Market Impact: A Consistent Microfoundation (2026 preprint) ↗partial progress · source checked 2026-11-09

    arXiv v2 abstract: arbitrageur information, stationary-limit assumptions, causal all-pass impact law and positive costs for a selected phase.

    Directly overlapping endogenous-impact microfoundation. Proposed novelty requires a precise comparison with its information, convergence and impact-cost assumptions.