RAND KWONG YEW LOW, University of Queensland Business School
Email: r.low@business.uq.edu.au
ROBERT W. FAFF, University of Queensland, Financial Research Network (FIRN)
Email: r.faff@business.uq.edu.au
KJERSTI AAS, Norwegian Computing Center
Email: kjersti.aas@nr.no

The existence of asymmetric dependence structures has been widely reported within equities throughout extant financial literature. However, many advanced mean-variance (MV) optimization models fail to incorporate this aspect explicitly within the portfolio management process. We demonstrate an application of the Clayton canonical vine copula (CVC) to incorporate asymmetric dependence in large portfolios to generate estimates of the expected return and covariances for several MV optimization models in a Bayesian approach to reduce estimation error in comparison to historical sampling windows as performed by DeMiguel et al. (2009b). We consistently find that several MV rules are improved across various data sets when asymmetric estimates are used. In addition, these enhancements lead to out performance of the naïve equally-weighted portfolio (1/N) by variations of the minimum-variance (MIN) optimization rule and combination portfolio strategies across various metrics. Although these improvements come at the cost of increased turnover, they show evidence in support of MV theoretical framework for investors.

CHARLES S. TAPIERO, NYU Poly - Department of Finance and Risk Engineering
Email: ctapiero@poly.edu

Convergence of financial theory and practice heralded by the seminal and fundamental economic research by Arrow and Debreu in the early 1950’s has led in the hands of financial engineers to an extraordinary financial innovation in the 70’s and ever since. The theoretical ability to price future assets has led to an explosive growth of financial trading, liquidity and to the growth of finance and its citadels. Options and credit derivative markets, securitization of non or partially liquid assets have provided extraordinary opportunities to “unearth� frozen assets for trade and profit.

The 2007-2009 financial crisis has evoked a greater awareness that traditional financial dogma is based on assumptions that have become increasingly difficult to justify. Globalization, the growth of "Too Big To Fail" (TBTF) firms, insiders trading, information and power asymmetries and the growth of regulation have, among other factors, conspired to render the assumption of complete markets to be unsustainable. Deviant behaviors in financial markets, non-transparency of transactions, complexity and dependence on a global scale have created a fragile and contagious global economy, where systemic risks are no longer an exception but a permanent threat.

By the same token, the explosive growth of information and data fueled by the internet and social media as well as an IT has created far greater dependence of financial systems and institutions on Information Technology (IT) emphasizing information as assets they seek to use for both financial management and competitive advantage. Technology, engineering and financial trends and developments have combined to yield an extraordinary growth of complexity, regulation and globalization, providing new opportunities and risks and undermine the traditional model based approaches to finance.

The purposes of this paper are to outline a number of factors, economic or otherwise that undermine the finance’s fundamental theories, their practice, regulation and their implications to the future of finance. In particular, we emphasize a strategic and multi-polar finance, beset by complexity, chaos and countervailing forces leading a multi-agent finance, computational and financial data-analytics driven rather than simple risk models of uncertainty.

A number of pricing models based on a strategic finance and a micro-matching of economic and financial states are also summarized and their practical implications drawn. In addition, topics such as Big Data finance and multi-agent financial modeling are outlined to provide elements for future theoretical and practical developments. This paper is a work in progress and therefore its intent is to attract greater attention to some elements (however selective and partial) that would contribute to potential transformations of a financial engineering future.

SUDIP RATAN CHANDRA, Indian Statistical Institute, Kolkata
Email: sudipratan@gmail.com
DIGANTA MUKHERJEE, Indian Statistical Institute, Kolkata
Email: digantaisi@yahoo.com

We propose a Mellin transform method to numerically compute the prices and greeks of single barrier options driven by a class of Levy processes under incomplete market with Follmer Schweizer minimal measure. We develop a partial integro-dierential equation (PIDE) for Barrier option and applied Mellin transform on it. Subsequently we perform the Inversion of these Mellin transform to construct an approximation of the prices and sensitivities of barrier options in a class of Levy models. The class includes many of the Levy models employed in quantitative finance such as the NIG, CGMY and Meixner models.

HAIM LEVY, Hebrew University of Jerusalem - Jerusalem School of Business Administration
Email: mshlevy@mscc.huji.ac.il
MOSHE LEVY, Hebrew University of Jerusalem - Jerusalem School of Business Administration
Email: mslm@mscc.huji.ac.il

The main problem of portfolio optimization is parameter estimation error. Various methods have been suggested to mitigate this problem, among which are shrinkage, resampling, Bayesian updating, naïve diversification, and imposing constraints on the portfolio weights. This study suggests two substantial extensions of the constrained optimization approach: the Variance-Based Constraints (VBC), and the Global Variance-Based Constraints (GVBC) methods. By the VBC method the constraint imposed on the weight of a given stock is inversely proportional to its standard deviation: the higher a stock’s sample standard deviation, the higher the potential estimation error of its parameters, and therefore the tighter the constraint imposed on its weight. GVBC employs a similar idea, but instead of imposing a sharp boundary constraint on each stock, a quadratic “cost� is assigned to deviations from the naive 1/N weight, and a single global constraint is imposed on the total cost of all deviations. Comparing ten optimization methods we find that the two new suggested methods typically yield the best performance, as measured by the Sharpe ratio. GVBC ranks first. These results are obtained for two different datasets, and are also robust to the number of assets under consideration.