WE DO THE WORK
BEFORE WE WRITE THE CHECK.
>Kosmos Labs is where our research, simulations and mechanism-design work lives. Every engagement starts with a model — and ships as running code. Below: a live index of what we’ve built, what we’re simulating now, and what we’re publishing next.
WHAT WE'VE BUILT
Three recent engagements. Each one starts as a question, ships as running code.
Possession-level liquidation engine
Built a discrete-event liquidation engine for an on-chain leveraged sports market. Possession-by-possession score processes feed real-time liquidation triggers (linear & exponential curves) — the simulation surfaces where the engine deleverages too early and where it leaves the protocol holding bad inventory. Stress-tested under thousands of game shapes before the engine shipped.
Yield-tokenization profit surface
Discrete-time simulator for a yield-tokenization protocol — validator accounts, PT/YT minting, secondary trading. Swept (deposits × velocity) parameter grids over a 24-month horizon to size the protocol-vs-staker-vs-validator profit split before launch. Cut three configurations that looked attractive on paper but starved the staker side.
Continuous clearing auction
Agent-based model of a continuous clearing auction with six heterogeneous agent types (early, informed, momentum, optimal, naïve, whale). Swept sentiment, information leakage and demand-intensity regimes to find supply schedules that resist whale extraction without underpricing the asset. The output: a front-loaded schedule that dominates back-loaded variants on staker surplus.
END-TO-END MODELLING
From hypothesis to durable growth, in three connected loops.
Mechanism & Quant Research
- Mechanism & market design
- Causal inference: panel, DiD, Bayesian
- Market microstructure & liquidity
- Experiment design: A/B, bandits
Simulation & Experimentation
- Agent-based models for user / validator behavior
- Monte Carlo for issuance, treasury & runway
- MEV & game-theory adversarial sims
- Sensitivity surfaces & policy search
Data Platform & Monitoring
- On-chain ETL & feature store
- Dashboards: retention, liquidity, unit econ
- Real-time anomaly detection
- Quarterly reviews & parameter tuning
Two ways in.
We take a small number of engagements per quarter. Pick the lane that fits — or write us and we’ll route you.
Advisory engagements.
Mechanism design, simulations, tokenomics, agent-based stress-tests. We ship code, not slide decks. Each engagement starts with a model and ends with the running implementation — the same way we work with our portfolio. Open to teams whether we’ve invested or not.
CONTACT RESEARCHCapital + advisory.
Building something we should back? Seed and pre-seed founders in crypto-economics get capital and the same research desk, bundled. The interactive simulators on this page are the kind of work that lands in the first 90 days of a portfolio engagement.
SEND YOUR DECK