Research

Fareground research.

01 · The question

Can agents in simulated environments predict real-world outcomes?

Fareground uses multi-agent simulations to study real systems before their outcomes are known. We model the actors, information, incentives, and constraints that shape a scenario, then observe what emerges when agents make decisions inside it.

The question is not whether a simulation can produce a plausible story. It is whether its aggregate result contains useful signal about what happens next — and under which conditions that signal is reliable. We compare simulations with observed outcomes and simpler baselines to learn where the method works, where it fails, and how it can improve.

02 · Method

How we work.

01Frame the questionCollect the relevant data, define the scenario, and identify the observable outcome the study will measure.
02Build and groundEncode the actors, information, rules, incentives, and possible actions, then calibrate the environment against known evidence.
03Run simulationsExecute repeated trials across seeds, agent populations, models, strategies, and controlled variations.
04Analyze the systemAggregate outcomes and study how agents reason, coordinate, adapt, and influence one another.
05Publish the evidenceRelease the findings with the runnable environment, configuration, evaluation, and stated limitations.
03 · Themes

What we study.

Marketing & demandForecast campaign response, message resonance, audience movement, conversion, and demand.
Product, pricing & adoptionTest launches, packaging, pricing, positioning, retention, and competitive response.
Competitive strategyModel competitor moves, market entry, strategic reactions, partnerships, and investment choices.
Markets & investmentStudy price formation, portfolio behavior, liquidity, capital allocation, and market stress.
Fraud & abuseSimulate evolving fraud rings, collusion, scams, laundering patterns, and defensive interventions.
CybersecurityRehearse attacker and defender behavior, incident propagation, controls, and recovery decisions.
Supply chain & operationsTest sourcing, inventory, routing, capacity, disruption, and cascading operational effects.
Negotiation & dealmakingExplore bargaining strategies, information asymmetry, concessions, coalitions, and agreement.
Policy & regulationEstimate stakeholder response, distributional effects, compliance behavior, and unintended consequences.
Media & informationModel attention, narrative formation, news cycles, misinformation, and belief propagation.
Legal & litigationExplore case strategy, settlement dynamics, evidentiary narratives, and likely procedural outcomes.
Agent evaluation & competitionCompare models and strategies through repeated matches, behavioral traces, and adaptive benchmarks.
04 · Results

What we've found.

The research record brings together active studies, runnable environments, resolved predictions, and publications as the work develops.