ATechLab develops quantitative, low latency trading strategies .

We develop and exploit a variety of technological methods including not only classical quantitative analysis but also deep learning. Our strategies use learning techniques for several purposes, but not the least for generating stable alphas. We use data to train those alphas in order to make price direction predictions on different time and event horizons. The data helps us to find fundamentally stable relations between demand/supply market forces in the past and the near future.
Our trading activities comprise fully autonomous strategies, all decisions are made purely based on algorithms. Those algorithmic decisions in short are based on the price of alpha, but those alphas themselves are a complex product of predicting market state quantities such as volatility, expected traded volume, time to next trade etc. Moreover, we track the microstructure of the market that helps us to keep a tight spread, which facilitates an efficiency of the markets. The ML allows confidently forecasting most target values of interest, and yet to spend a little time engineering inputs for predictive models, as the learning algorithms do this semi-automatically.