High Frequency Trading

Low latency market access via a 100% in-house application stack – from FPGA to Deep Neural Nets.

Statistical Arbitrage within and across exchanges, in which we we predict and capitalize on small mis-pricings in one or more assets.

Market Making, in which we provide liquidity for other market participants and tighten the bid-ask spread while earning the small difference between the bid-price and ask-price.

Positional Bets

Identify and take positions using deep reinforcement learning – trades from data with no humans in the loop.

Trading strategies are "generic", which means they learn how to trade particular asset classes and exchanges by themselves.

We do this without taking outsized risks on any single position – and we never take tail risks which can cause ruin.

Long Term Portfolios

Spark ideas that lead to long term holdings which are highly scalable. These trade infrequently – at most daily, and are less fragile wrt execution costs.

We have a mix of fundamental portfolios (read: no ML) and quant portfolios (read: some ML). Both have significantly outperformed the market.