OMI Annual Report 2017/18

OMI Annual Report 2017/18

Foreword from OMI Director, Prof. Steve Roberts:

This year has seen AI and machine learning even more prominent in the media. From playing Go to advances in autonomous systems, advances seem to arrive at an unprecedented rate. Our keen focus on principled AI and machine learning within the OMI means that we are able to follow these advances, develop and innovate new ones and work to make them deployable.

Our ability to produce world-leading science requires a commitment to bring talent into the OMI. We are delighted that over the past year and half we have been able to expand. We have appointed not only three new faculty (two senior researchers and a departmental lecturer), but also welcome four postdoctoral researchers into the institute. Over the last year, we have hosted ten interns and will be welcoming twelve new graduate students into the OMI to work on various aspects of machine learning for finance.

We have worked over the year to develop in-house hardware and software to accelerate our research. Most notably, our IT team have created a rapid development infrastructure, co-located with our data servers. This enables development and testing of models, techniques and deployments with fast, reliable access to our vast data repositories. Allied with hardware capability, we have spent time creating new software to distil the data feeds that come into the OMI, creating a data lake of derived measures (such as realised volatility) essential for research. This last summer we pioneered the development of a trade simulator. This enables us to test, for the first time, the effectiveness of algorithmic trading strategies in a digital twin of a live financial market. Further, it allows different algorithms to compete in the marketplace, paving the way to realistic multi-agent simulations.

This year has seen links throughout the University strengthen, welcoming new members from Computer Science, the Saïd Business School and the department of Economics. Our members have published broadly, producing innovative work at the heart of finance (such as developing algorithms to forecast prices from limit order books) to novel algorithms for machine learning on quantum computers - a transformation yet to come. We have been involved throughout the year at high-profile external events; OMI members have delivered keynote seminars and taken part in policy workshops at the heart of government and the finance industry. Our internal termly workshops have continued to be very well supported, building outreach and collaboration across the University.

We look forward to the challenges, inspiration and innovation the coming year will bring!

Steve Roberts, OMI Director