Completed

Abstract

ODERIM “Outlier Detection/Estimation and mitigation for RIsk Management and control, based on Advanced SSP methods, with a focus on extreme situations”.

Abstract

Many institutional traders split large orders into smaller orders sent over some time period. This schedule may be optimized to reduce price impact. I have developed performance metrics to assess how effective funds are at (i) executing these smaller orders, (ii) deciding when to wait for orders to be filled (i.e. market timing), and (iii) scheduling the smaller orders. The performance metrics have sound theoretical backing and let us separate trading-related performance from noise. I propose to use data on orders and trades for a selection of investment funds to characterize these skills. For the initial work, I will study: (i) the relative magnitudes of these skills, (ii) how these skills vary across funds, (iii) what fraction of firms seem to possess superior trading-related skills, (iv) how firms’ skills change over time due to learning, and (v) the savings in transactions costs which accrue to investors. For possible further work, I suspect this data would help answer further questions including: (vi) how firms’ trading-related performance changes with macroeconomic factors, (vii) whether changes in trading-related skills result in fund inflows, (viii) the value of these inflows to the funds.

Abstract

The object of this project is to study default dependence and contagion amongst nonagency securitized mortgages in the US over the period 1998-2011. We will use a Cox proportional hazard model in a competing risk framework for default (and prepayment) and a copula model for the dependence amongst individual hazards. Dependence between defaults can occur because of geographical proximity, common economic conditions, which may be of either a local or economy-wide nature, the business cycle, interest rates, etc. We want to quantify the amount of this default dependence and investigate the reasons why such dependence occurs.

Working paper

Contagion in Subprime Mortgage Defaults : a Composite Likelihood Approach

Abstract

The focus of this study is to develop theories that can underpin information mining on the web to produce reliable information and to assess the impact of existing methods on the behaviour of market prices using techniques that are based upon the concept of entropy.

The framework for the analysis will be provided by information theory. The major metrics will be constructed on the application of concepts related to Shannon entropy and cross entropy. The data for the study will be drawn from Thomson Reuters market data provided by The Securities Industry Research Centre of the Asia Pacific (SIRCA).

Working papers:

• An entropy based analysis of the relationship between the DOW JONES Index and the TRNA Sentiment series,
• Daily Market News Sentiment and Stock Prices.

Working paper: High-Frequency Trading around Large Institutional Orders, forthcoming in the Journal of Finance, see Publications.

Working paper: The Booms and Busts of Beta Arbitrage

Working Paper: Momentum Strategies in Futures Markets and Trend-following Funds

Working paper: Identifying Contation. Paper published in Journal of Applied Econometrics, see Publications.

In a framework where the CAPM holds conditionally, we develop a model for the cross-section with fairly general dividend dynamics and a stochastic discount factor that accounts for standard asset pricing moments. Our model, which successfully captures unconditional betas across industries, features a cyclical component that induces significant variations in conditional betas. These variations exhibit non-linear patterns in relation to the business cycle. Using the time-series of the betas implied by our model, we  find that the conditional CAPM explains the cross section of industry average quarterly returns over the period from 1927-2021.

Paper : "Cyclical Beta"