skip to content

Keynes Fund


Project Summary

Many financial markets (for example stock markets, FX, treasury bonds, commodities) involve electronic order book trading at very high frequency. This produces a vast amount of complex data, the order book itself reflects many different "messages" such as different order types, cancellations, and executions. This makes its analysis very challenging. At the same time, this data allows one to ask fundamental questions about market efficiency, competition between intermediaries, market manipulation, the efficacy of regulatory and policy changes and so on.

We aim to develop new econometric tools to study high-frequency data emerging from algorithmic high-frequency trading. The projects are based on the ReMeDI regime recently proposed by Li and Linton (2019), which provides effective tools to separate a time series into a permanent and transitory component, where the former is identified with the fundamental values of asset prices whereas the latter reflects either market microstructure is- sues or market inefficiency/mispricing. In the high-frequency environment, the temporary component can be highly persistent, which was the motivation for our new methodology. We plan to apply and extend the ReMeDI method to study the following problems.

  1. How to estimate the parameters of the fundamental values (e.g., volatility) using tick data?

    To answer this question, we aim to develop a new de-noise method using the ReMeDI approach. We will develop new estimators to measure the volatility of the fundamental values, and testing for the presence of jumps. We will also study the impact of sampling frequency on estimation. We aim to construct new estimators that are robust to data frequencies and such robustness is a direct consequence of the merit of the ReMeDI design.

  2. How to empirically measure high-frequency market liquidity? In particular, how to measure the liquidity of a portfolio?

    We will apply the two liquidity measures proposed by Li and Linton (2019) to measure high-frequency market liquidity. The two measures provide more accurate estimation than the classic Roll measure (Roll (1984)) when the order flows exhibit autocorrelation patterns. We will develop the multivariate ReMeDI estimators, and they can be applied to measure the liquidity of a portfolio.

  3. How to tell whether a time series of asset prices embed market frictions?

    We will develop a testing procedure based on the ReMeDI estimators to tell whether a martingale process is adequate to model the prices of a financial asset. It compares the relative sizes of the mispricing and fundamental values.

The projects well suit the objective and themes of the Keynes Fund, which promotes the study of the sources and consequences of the failure of market efficiency. Specifically, we focus on the high-frequency financial market operates at astonishing speed by computer programs. We will design econometric tools to empirically measure the volatility of fundaments.



Dr. Merrick Li and Prof. Oliver Linton


Dr Merrick Li is a Cambridge-INET Postdoctoral Research Fellow at the Faculty of Economics, University of Cambridge. His research expertise is in Analysis of High-Frequency Financial Data.


Professor Oliver Linton is the Professor of Political Economy in the Faculty of Economics and a Fellow of Trinity College, University of Cambridge. His research is focussed on nonparametric and semiparametric methods with an interest in Financial Econometrics.


Share this Project