Search

We identify a source of numerical instability of quadratic programming problems that is hidden in its linear equality constraints. We propose a new theoretical approach to rewrite the original optimization problem in an equivalent reformulation using the singular value decomposition and substituting the ill-conditioned original matrix of the restrictions with a suitable optimally conditioned one. The proposed novel approach is showed, both empirically and theoretically, to solve ill-conditioning related numerical issues, not only when they depend on bad scaling and are relative easy to handle, but also when they result from almost collinearity or when numerically rank-deficient matrices are involved. Furthermore, our strategy looks very promising even when additional inequality constraints are considered in the optimization problem, as it occurs in several practical applications. In this framework, even if no closed form solution is available, we show, through empirical evidence, how the equivalent reformulation of the original problem greatly improves the performances of MatLab®’s quadratic programming solver and Gurobi®. The experimental validation is provided through numerical examples performed on real financial data in the portfolio optimization context.

In the financial framework, the concepts of connectedness and diversification have been introduced and developed respectively in the context of systemic risk and portfolio theory. In this paper we propose a theoretical approach to bring to light the relation between connectedness and diversification. Starting from the respective axiomatic definitions, we prove that a class of proper measures of connectedness verifies, after a suitable functional transformation, the axiomatic requirements for a measure of diversification. The core idea of the paper is that connectedness and diversification are so deeply related that it is possible to pass from one concept to the other. In order to exploit such correspondence, we introduce a function, depending on the classical notion of rank of a matrix, that transforms a suitable proper measure of connectedness in a measure of diversification. We point out general properties of the proposed transformation function and apply it to a selection of measures of connectedness, such as the well-known Variance Inflation Factor.

The concept of connectedness has been widely used in financial applications, in particular for systemic risk detection. Despite its popularity, at the state of the art, a rigorous definition of connectedness is still missing. In this paper we propose a general definition of connectedness introducing the notion of proper measures of connectedness (PMCs). Based on the classical concept of mean introduced by Chisini, we define a family of PMCs and prove some useful properties. Further, we investigate whether the most popular measures of connectedness available in the literature are consistent with the proposed theoretical framework. We also compare different measures in terms of forecasting performances on real financial data. The empirical evidence shows the forecasting superiority of the PMCs compared to the measures that do not satisfy the theoretical properties. Moreover, the empirical results support the evidence that the PMCs can be useful to detect in advance financial bubbles, crises, and, in general, for systemic risk detection.

This paper presents a novel methodological approach called the Model of Models (MoM). MoM concerns the selection of the best model for a given partition of the data derived from the realization of the independent variables. Compared to ensemble techniques and model averaging approaches proposed in the literature, MoM does not require a selection of which models to include in the pool of models and it works without resorting to the combination of model predictions. MoM works on parametric and non parametric predictive models as well as any other dependent or independent variables. In the case of a partition of the data, the theoretical proposal derives the properties of MoM. The implementation of MoM, when no partition of the data is available in advance, is performed using a new algorithm termed as MoMa. In order to show how MoM works, empirical evidence is provided on simulated data sets. The proved theoretical results coupled with the empirical evidence gathered from simulated data demonstrate that MoM is a good strategy to deal with model choice and model uncertainty.