Omar Rojas

Adaptive market efficiency of agricultural commodity futures contracts

In this paper we investigate the adaptive market efficiency of the agricultural commodity futures market, using a sample of eight futures contracts. Using a battery of nonlinear tests, we uncover the nonlinear serial dependence in the returns series. We run the Hinich portmanteau bicorrelation test to uncover the moments in which the nonlinear serial dependence, and therefore adaptive market efficiency, occurs for our sample.

Production Planning for a Company in the Industry of Compact Discs Mass Replications

This chapter addresses a production planning problem for a company that mass-replicates compact discs. In the current situation, the staff of the company creates a production plan for the long term, with low flexibility to make a new plan when new orders arrive. The combination of attributes of the orders and the available machines for the processes generate a high complexity to determine the appropriate production routes and sequencing. To reduce complexity, the computation of a priority index is proposed to combine different attributes of the orders. To optimize the utilization of the production capacities, two approaches were proposed: a Simulation Model and a Linear Programming Model. The priority index is used in both models to promote early scheduling of certain orders to the machines during the planning horizon. The results show that the models proposed deliver production plans in a short time, with a better utilization of the production capacity, and with a focus on improving service level when compared to the current methodology in the company. Additionally, the Linear Programming Model is integrated into an intelligent decision-support system to guarantee the data transfer from the information system of the company and fast execution as often as needed.

Design of experiments and response surface methodology to tune machine learning hyperparameters, with a random forest case-study

Abstract Most machine learning algorithms possess hyperparameters. For example, an artificial neural network requires the determination of the number of hidden layers, nodes, and many other parameters related to the model fitting process. Despite this, there is still no clear consensus on how to tune them. The most popular methodology is an exhaustive grid search, which can be highly inefficient and sometimes infeasible. Another common solution is to change one hyperparameter at a time and measure its effect on the model’s performance. However, this can also be inefficient and does not guarantee optimal results since it ignores interactions between the hyperparameters. In this paper, we propose to use the Design of Experiments (DOE) methodology (factorial designs) for screening and Response Surface Methodology (RSM) to tune a machine learning algorithm’s hyperparameters. An application of our methodology is presented with a detailed discussion of the results of a random forest case-study using a publicly available dataset. Benefits include fewer training runs, better parameter selection, and a disciplined approach based on statistical theory.

A Bayesian approach to model changes in volatility in the Mexican stock exchange index

ABSTRACT We model the changes in volatility in the Mexican Stock Exchange Index using a Bayesian approach. We study the time series with a wide set of models characterized by a Markov switching heterogeneity. The advantage of this approach is that it allows for a broader spectrum of possible models since the estimation of the moments of the parameters is done using the finite mixture distribution MCMC method, without relying on assumptions about large sampling and mathematical optimization. This is particularly relevant for emerging markets’ financial data because of its special characteristics, like being more susceptible to jumps and changes in volatility caused by exchange rate swings, financial crises and oil and commodity prices. For model comparison, we use the marginal likelihood approach and the bridge sampling technique. The best representation of the data is given by a switching model with three states rather than any other autoregressive linear or non-linear model. The periods of volatility found by the model coincide with different financial crisis. Whereas other studies of volatility for the same market impose the Markovian model that captures changes in volatility, we let our model to be defined in an endogenous way.

Causality patterns for Brent, WTI, and Argus oil prices

ABSTRACT Causality patterns are analysed for daily Brent, West Texas Intermediate (WTI), and Argus Sour Crude Index (Argus) oil prices, Argus is the reference price for exports from Saudi Arabia, Kuwait and Iraq. Nonparametric Granger causality testing uncovers bi-directional causal links between Brent and WTI prices at multiple lags. Unidirectional causality from both Brent to Argus and WTI to Argus is also documented. If the current Saudi Arabia attempt to increase market share is successful, variations in Argus prices may start preceding movements in Brent and WTI, also.