01

White-Glove Microstructure

Generative Order Flow Studios — We don't forecast prices; we simulate markets into submission.

Core Research Questions

  • How do we generate realistic order flow that preserves stylized facts (spread dynamics, queue position, cancellations, volatility clustering) across regimes?
  • Can generative simulators be conditioned on policy changes (tick size, fee changes, auction rules) to predict microstructure impact?
  • How do we validate "market realism" beyond RMSE (distributional tests, agent-based stress testing, tail fidelity)?

Key Methodologies

  • Diffusion/transformer/state-space generative models for event streams and LOB states
  • Causal evaluation + distributional testing; calibration to microstructure invariants
  • Agent-based adversaries to probe exploitability
Trading/Investment Impact

Higher-fidelity backtesting and stress testing for execution, short-horizon signals, and market impact models. Synthetic scenario generation for liquidity shocks and "unknown unknowns."

02

Compliance-by-Construction Agents

The "Licensed-to-Trade" RL Stack — Trading agents that are born compliant: constraints are first-class, not after-the-fact kill switches.

Core Research Questions

  • How do we translate regulations, exchange rules, and internal policies into machine-checkable constraints (position limits, market manipulation patterns, best execution)?
  • Can we build constrained RL/safe RL that optimizes under hard constraints and produces usable rationales?
  • How do we prove (or strongly evidence) that an agent won't engage in prohibited behaviors under distribution shift?

Key Methodologies

  • Constrained optimization/safe RL (Lagrangian, shielded policies, offline RL with constraints)
  • LLMs for rule parsing → formal constraint representations; program analysis style verification where feasible
  • Auditability: counterfactual explanations + reproducible decision logs
Trading/Investment Impact

Faster deployment of systematic strategies with lower governance friction. Reduced tail risk from "model goes rogue" behaviors; improved survivability during audits/incidents.

03

The Invisible Exchange

Privacy-Preserving Execution & Auction Design — Dark pools, but mathematically honest.

Core Research Questions

  • Which privacy model actually matters for trading: hiding intent pre-trade, minimizing information leakage post-trade, or preventing operator abuse?
  • Can we design auctions/matching (batch auctions, sealed-bid, intent auctions) that reduce adverse selection while remaining performant?
  • What is legal enforceability and surveillance posture of privacy-preserving venues?

Key Methodologies

  • Secure multi-party computation, zero-knowledge proofs, threshold encryption
  • Formal threat models (operator, participant collusion, sequencer/block proposer)
  • Empirical microstructure evaluation: does privacy improve or worsen realized spreads/adverse selection?
Trading/Investment Impact

Better execution for large orders; reduced signaling and slippage. MEV-aware strategy design for digital assets; potentially exploitable structure in new auction mechanisms.

04

Time Lords of Finance

Telecom-Grade Synchronization & Deterministic Latency — Treat time as an asset.

Core Research Questions

  • How do clock quality, asymmetry, packet delay variation, and boundary clock behavior bias market microstructure measurements?
  • Can we design time-quality-aware trading and risk systems (signals that degrade gracefully when time sync degrades)?
  • What is the economic value of deterministic networking vs raw speed?

Key Methodologies

  • PTP/time-sync monitoring, causal timestamp correction, uncertainty quantification
  • Network measurement + queueing models; co-design of feed handlers and time-aware analytics
  • Robust statistics to prevent timing artifacts being mistaken for alpha
Trading/Investment Impact

Cleaner event ordering → better microstructure models, latency attribution, and execution analytics. New "ops alpha": fewer phantom signals, fewer false positives in surveillance, tighter risk controls.

05

Data With Provenance

Synthetic Alternative Data That Survives Court — We only trade on data we can defend.

Core Research Questions

  • How do we quantify and enforce rights: licensing, consent, purpose limitation, retention, cross-border transfer?
  • When does synthetic data meaningfully reduce privacy/IP risk without destroying predictive signal?
  • Can we build a "data lineage → model lineage → decision lineage" chain suitable for internal and external review?

Key Methodologies

  • Differential privacy, federated learning, privacy audits (membership inference), watermarking
  • Data provenance systems + model cards; automated policy enforcement in pipelines
  • Synthetic data evaluation that prioritizes decision-equivalence, not just distribution similarity
Trading/Investment Impact

More alternative data access with lower legal/reg blowback. Faster iteration because datasets are governable and reusable.

06

Sanctions, Supply Chains, and Signal

Quantifying Legal Shockwaves — Trade law as a market factor model.

Core Research Questions

  • How do regulatory events propagate through supply networks into prices and vol (and with what lags)?
  • Can we build causal estimators that separate "headline reaction" from real constraint tightening (shipping insurance, port throughput, telecom outages)?
  • How do we forecast enforcement intensity (not just policy announcements)?

Key Methodologies

  • Causal inference (diff-in-diff, synthetic controls), graph models of supply chains, regime detection
  • NLP on legal/regulatory documents with careful grounding; event databases with provenance
  • Scenario generation tied to plausible legal pathways (what can actually be enforced)
Trading/Investment Impact

Event-driven and medium-horizon strategies in commodities/FX/equities/credit. Better risk management around geopolitical tails and sudden liquidity regime changes.

Interested in Collaboration?

Our research lines benefit from diverse perspectives and expertise. We welcome partnerships with academic institutions, industry leaders, and regulatory bodies.