Research on alternative markets.
PRLX is an independent lab that builds and tests high-frequency strategies on Solana, prediction markets and agentic systems.
We develop and test trading strategies for markets whose structure is still being worked out.
PRLX is a small, independent research team. Our work covers Solana market structure and latency, market making in prediction markets, and whether AI agents can develop trading strategies of their own. We trade only our own capital.
The name comes from parallax error: the wrong reading you get when you look at a gauge from the side. Every venue, validator and data feed sees the price from a different place, a little late, and much of what we study is the size of that gap.
Solana latency
Where transactions land inside a 400 ms slot, how far a seat sits from the leader, and what a few milliseconds are worth.
Prediction markets
Market making in short-dated event markets: queue position, maker incentives, and how quoted odds react when a faster market moves first.
Agentic trading
Whether self-learning AI agents can develop strategies of their own, studied as a firm of agents that train, get hired, are promoted or let go, and teach the next cohort.
Emerging venues
New chains and sequencer designs, and who gets to trade first on them.
How a strategy gets built.
Every idea passes the same six gates, in order. Most stop early, and that is the point: each gate is cheaper than the one after it.
Hypothesis
A mechanism in writing, with the pass mark set before any data is read.
Replay
Tested on recorded market data, at a speed we can actually reach.
Latency model
Where our orders would really land, measured against the leader.
Paper
Live data and simulated orders, checked against the replay.
Probes
Real transactions that land on chain but never trade, to check our seat.
Small size
Capital only after every earlier gate, starting small.