John F. O. Bilson

The profitability of currency speculation

This paper presents the results of a post-sample simulation of a speculative strategy using a portfolio of foreign currency forward contracts.The main new features of the speculative strategy are (a)the use of Kalman filters to update the forecasting equation, (b) the allowance for transactions,costs and margin requirements and (c) the endogenous determination of the leveraging of the portfolio. While the forecasting model tended to overestimate profit and underestimate risk, the strategy was still profitable over a three year period and it was possible to reject the hypothesis that the sum of profits was zero. Furthermore, the currency portfolio was found to have an extremely low market risk. Combinations of the speculative currency portfolio with traditional portfolios of U.S. equities resulted in considerable improvements in risk-adjusted returns on capital.

Trading Model Uncertainty and Statistical Process Control

This article examines the use of statistical process control (SPC) as a methodology for monitoring trading model uncertainty. Traditional quantitative risk management methods do not incorporate the inherent process control problems financial modeling. The reference adaptations apply to the a priori model design process and a posteriori model control. To build and monitor trading models, SPC can be used in conjunction with classical financial metrics to better control market risk.

CTA Performance Persistence: 1994-2010

This article tests the performance persistence hypothesis for Commodity Trading Advisors (CTAs), considering the impact of incubation and backfill bias. The authors apply the Fama-MacBeth approach and quintile analysis, and conclude that ranking CTAs using the t-statistic of alpha with respect to a CTA benchmark is predictive of future returns. The authors provide evidence that the identified strong persistence of the best-performing funds may be driven solely by the incubation and backfill biases. They find that the worst-performing funds have a higher probability of liquidation than those of the other quintiles, and the top-performing funds have a higher conditional probability of staying top performers versus becoming worst performers than that of the worst performing funds.

An Empirical Investigation of Eastern European Bond Markets

ABSTRACT We examine the value of Eastern European emerging bond markets to global fixed income managers. In an environment where bonds from traditional developed markets are offering modest yields, emerging market bonds with attractive yields are becoming more popular with institutional managers. Furthermore, the returns on these bonds exhibit low correlations with traditional fixed income investments and thus offer opportunities for portfolio diversification. We develop a multifactor forecasting model and estimate its parameters using a dynamic Kalman filter procedure. The forecasts are then used to construct optimal mean–variance portfolios with and without emerging market bonds. We find that the portfolios that include emerging market bonds have significantly higher Sharpe ratios.

Trading asymmetric trend and volatility by leverage trend GARCH in Taiwan stock index

This article develops a leverage trend Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model by incorporating asymmetric trend of returns of the exponential autoregressive and asymmetric volatility of GARCH models to study the asymmetric effects. Using in-sample daily data of Taiex over the period 4 January 1980 to 25 August 1997 and postsample daily data over the period 26 August 1997 to 10 September 2007, the evidence reveals that a curvaceous risk–return relationship and both asymmetric volatility and asymmetric trend of returns are significant in Taiex. The episode of asymmetric trend of returns is that the positive information creates a higher return trend than the negative information of the same amount, while similarly to most studies, the evidence of asymmetric volatility appears that the negative information makes a higher volatility than the positive information of the same size. Most remarkably, we evidence that the volatility asymmetry effect is a conservative trading factor and the return trend asymmetry effect is an active trading factor. In comparison of post-sample performance using rolling-window technique, the leverage trend GARCH model indeed outperforms the other three models with single asymmetry adjusted or without asymmetry adjusted, while the asymmetry nonadjusted model performs the worst. It implies that the return trend asymmetry (active trading) and the volatility asymmetry effects (conservative trading) tend to compensate, but not offset each other.

An Empirical Bond Portfolio Study: Evidence from the Asian Emerging Bond Market

Most of the research on the benefits of diversification into emerging markets (EMs) has focused on equity markets. In this research, the focus is on investments in fixed income instruments. Specifically, the research explores the performance benefits of developed markets (DMs) combined with the Asian emerging market (AEM). This research indentifies the linkage and leverage of government bond portfolios between DMs and the AEM. It first identifies the potential diversification and describes the financial integration for incorporating AEM government bonds into DMs government bond portfolios. In the second phase, it constructs the dynamic linear regression models and conducts the mean-variance tests to demonstrate the incremental benefit of the strategy. In the last phase, a robust test examines the strength of bond portfolio performance between DMs with the AEM and the government bond index.

The Distribution of Returns and Risk Forecasting

Publisher Summary The objective of this chapter is to revive the simple analytic tradition of value-at-risk (VaR) calculation while taking account of nonlinear payoff functions and departures from normality. The basic idea behind the approach is that there are fundamental risk factors that are multivariate normal and the departures from normality observed in actual return series are the consequence of a nonlinear relationship between the actual returns and the fundamental risk factors. This approach is used to conduct a risk analysis of a simple futures portfolio. The risk analysis includes estimates of the VaR, the components of the VaR, and the conditional expected loss (CeL) of the portfolio. All commercial applications of the VaR methodology are based on advanced simulation models. These models have the advantage of being able to use exact valuations rather than using delta and delta-gamma approximations, and they can employ distributions of returns based on historical data rather than the normal approximation.