Automated trading with machine learning on big data

Financial markets are now extremely efficient,nevertheless there are still many investment funds that generatealpha systematically beating markets' return benchmarks. Theemergence of big data gave professional traders the newterritory, leverage and evidence and renewed opportunitiesof their profitable exploitation by Machine Learning (ML)models, increasingly taking over the trading floor by 24/7automated trading in response to the continuously fed datastreams. Rapidly increasing data sizes and strictly real-timerequirements of the trading models render large subset ofML methods intractable, overcomplex and impossible to applyin practise. In this work we demonstrate how to efficientlyapproach the problem of automated trading with large portfoliostrategy that continuously consumes streams of data acrossmultiple diverse markets. We demonstrate a simple scalabletrading model that learns to generate profit from multiple intermarketprice predictions and markets' correlation structure.We also introduce the stochastic trade diffusion technique tomaximise trading turnover while reducing strategy's exposureto market impact and construct the efficient risk-mitigatingportfolio that backtests with the strong positive return.