The man-machine divide in trading is being continually tested, but the evidence so far seems to be that human and algorithmic based trading are two sides of the same coin, inexorably bound up with each other.

“Humans are great at big-picture long term analysis, but machines are able to do things we can’t,” said Tucker Balch, professor of computer science at Georgia Institute of Technology and founder of Lucena Research, a machine learning-based investment technology company. “A computer can examine all the details and relationships between thousands of securities across a vast market.”

Lucena Research develops advanced tools that leverage artificial intelligence for price prediction, hedging, and other complex portfolio management tasks.

“I’ve come to believe that the best portfolio management requires a combination of human and machine,” Balch said.

The critical test in applying machine-based learning and other forms of artificial intelligence to trading is whether it can generate alpha, or returns that exceed those that could be attributed solely to market forces.

QuantDesk, Lucena research’s flagship product, incorporates over 200 fundamental, technical, and proprietary time series indicators to exploit market opportunities with precision, and scientifically validate and assess investment decisions.

“By perpetually self-adjusting its predictive model-based data analytics, QuantDesk enables hedge funds, portfolio managers and wealth advisors of all sizes to leverage quantitative research of fundamental, technical and proprietary data and incorporate statistical forecasting into their proprietary investment strategies,” said Erez Katz, co-founder and CEO of Lucena.

QuantDesk incorporates data from InsiderInsights, a provider of proprietary data on insider trading, into its existing fundamental and technical data indicators in order to form a statistical price forecasting and confidence measures for the underlying stocks.

Lucena’s customers will be able to monitor and qualify insiders’ activity, adding another dimension to its predictive models for share price movements and other market opportunities.

A successful quantitative-based investment strategy must include both quality data coupled with computational technology that helps clients extract meaningful and actionable intelligence from that data.

“Many highly-traded equities are already at point where it’s hard to fund alpha through fundamental and technical indicators alone,” Katz said. “The true nature of machine learning is to identify which among the plethora of indicators are meaningful, and identify proprietary data sources that have not been overanalyzed. That’s why we partner proprietary data providers like InsiderInsights.”

A core technology of Lucena Research is a portfolio optimizer that provides recommendations for allocations to each asset in a portfolio, based on forecast alpha, volatility, and covariance between each asset and every other asset.
“The effectiveness of an optimizer’s recommendations depends significantly on the quality of the data it has to work with especially how predictive that data is for the future,” said Balch.

Overall, historic volatility and covariance are predictive of future volatility and covariance, but forecast alpha is usually less accurate. If forecast alphas are off by much they overwhelm the optimizer’s allocations and send it in the wrong direction.

“Our approach is to leverage the sweet spot of optimizer effectiveness by taking forecast alpha out of the picture,” Balch said. “We do that by telling the optimizer that our forecast alpha for all assets is the same (namely zero), and we ask it to minimize future volatility.”