Speaker "Revant Nayar" Details Back



Group Theory, Chaos and Financial Time Series


In this talk we show how group and chaos theoretic techniques can be used jointly to capture nonlinearity in financial time series in a much more efficient, robust and transparent manner than AI, SDEs and other canonical techniques. We begin by presenting diagnostics of ergodicity and disorder in time series, focusing on mutual information and entropies to capture fully non perturbative and non linear properties of the stochastic process. We remark on the emergence and breaking of symmetry groups in financial markets both empirically and from the SDE perspectives, and show how remarkably well these alone can capture time series dynamics without any input from underlying PDEs or SDEs. We end by establishing a link with neural networks.
Who is this presentation for?
Quants, data scientists, mathematicians
Prerequisite knowledge:
Prior experience with PDEs/SDEs/stochastic processes will be helpful
What you'll learn?


Revant Nayar is a former Research Affiliate at Princeton University and IAS, and now CTO at FMI Technologies. He has published four papers and given talks at many leading conferences and universities, with a special emphasis on physics-inspired alternatives to AI. He features in Wealth and Finance Magazine as well as CIO Outlook Capital Markets, as one of the chief developers of Field Machine Intelligence, which is a robust alternative to AI tailored for financial time series.