Reinforcement Learning
Agents learn trading policies through millions of simulated episodes, rewarded for risk-adjusted return, not raw profit.
Synchronous is building reinforcement learning based AI trading systems that learn, adapt, and compete at the highest level. We work at the frontier of AI research in trading. Our agents train against the world's most challenging markets and run autonomous ETFs and onchain vaults under hard risk limits.
Agents learn trading policies through millions of simulated episodes, rewarded for risk-adjusted return, not raw profit.
Large language models read filings, news and onchain activity, turning unstructured information into signals the policy can use.
Markets shift regime. Our agents retrain on live data and are re-evaluated before each new version is allowed to trade.
Rules-based index strategies whose research, weighting and rebalancing are carried out by our agents, with published holdings.
ERC-4626 vaults where an allocator agent moves capital between vetted strategies inside caps enforced by the contract.
Our team is a great mix of people coming from big tech AI labs, top-tier trading companies, and academic research labs. We are all builders and dreamers, and we enjoy working together on challenging problems.
Training and agent workloads run on the cloud. We evaluate frontier models from multiple providers side by side and use whichever performs best on each task.
Research collaboration, infrastructure partnerships and roles in RL and trading systems.