One model of every venue
Normalized connectivity with explicit handling of each venue’s tick sizes, fee schedules, rate limits and settlement rules.
Approach
Disciplined pricing for uncertain outcomes.
A market maker is only as useful as its prices are good, its risk is controlled, its systems are reliable and its conduct is beyond question. This is how we approach each.
Pricing
Fair value, continuously.
Our quotes are anchored to probability models, not to the last trade. We treat every contract as a forecasting problem, and every quote as a forecast we will be held to.
- Information in, probabilities out
- Models ingest what moves each market (data releases, polls, game state, related prices and order flow) and produce calibrated probabilities with an estimate of their own uncertainty.
- Coherence across markets
- Related contracts must agree. Mutually exclusive outcomes must sum to one, strike ladders must be monotonic, and the same event on two venues should carry the same price, net of fees and settlement terms. We enforce these constraints in every quote.
- Scored, not assumed
- We evaluate forecasts with proper scoring rules, among them Brier score, log loss and reliability curves, and let the results, not intuition, decide which models trade.
Risk
Risk, measured at resolution.
Event contracts don’t drift to their final value. They jump. A position that looks modest at 50¢ can be worth a full dollar, or nothing, the moment an outcome is known.
- Terminal exposure
- We size inventory against what a position is worth at resolution, not just where it marks today.
- Aggregation by event
- Many contracts hinge on the same fact. We net exposure by underlying event and shared drivers, not ticker by ticker.
- Event-time awareness
- Quote size and width adapt ahead of scheduled information, such as data prints, debates and kick-offs, and react immediately to unscheduled shocks.
- Hard limits
- Pre-trade checks, position caps and kill switches run independently of strategy code, and they fail closed.
Technology
Engineered for continuity.
Probabilistic markets rarely sleep: news breaks overnight, games go to overtime, and some venues trade around the clock. Our systems are built to run continuously, and to fail safe when something upstream doesn’t.
Freshness budgets
Every input carries a maximum age. When a feed lags or a venue degrades, quotes are pulled automatically, before they can be picked off.
Research is production
Research and trading share code, and our simulation models queue position, fees and latency, so what we test is what we trade.
Everything replayable
Every quote, fill and decision is logged with its inputs, so any action can be explained after the fact.
Conduct
Fair markets are the product.
Prediction markets work only if participants trust both the prices and the process. We hold ourselves to standards that protect each.
- 01
The rules of the venue.
We comply with the rules of every venue we trade on and with the laws and regulations that govern them.
- 02
No manipulation.
No wash trading, spoofing or layering. Our controls are designed to prevent it, not merely to detect it.
- 03
No inside information.
We do not trade on material non-public information, and we take care to know where our information comes from.
- 04
Hands off outcomes.
We never attempt to influence the events we trade, or the processes by which markets resolve.
Work with us
A market maker that shows its work.
Venues, counterparties and due-diligence reviewers: we’d like to hear from you.