Umeå University's logo

This article focuses on predicting trends in Commodity Trading Advisors (CTAs), also known as trend-following hedge funds. The article applies a Hidden Markov Model (HMM) for classifying trends. By incorporating additional features, a regularized logistic regression model is used to enhance prediction capability. The model demonstrates success in identifying positive trends in CTA funds, with particular emphasis on precision and risk-adjusted return metrics. In the context of regularized regression models, techniques for statistical inference such as bootstrap resampling and Markov Chain Monte Carlo are applied to estimate the distribution of parameters. The findings suggest the model’s effectiveness in predicting favorable CTA performance and mitigating equity market drawdowns.

We investigate intra-day seasonal patterns in Brent crude oil futures traded at the Intercontinental Exchange in London. Our data cover tick data for futures of various maturities from January 2010 to October 2021, a database covering 130 Gigabytes of oil transactions. We convert these data into one-minute data and observe statistically significant intra-day seasonal patterns with peaks and bottoms at particular times of the day, depending on maturities and whether or not spreads are considered. In the second part of our analysis, we explore whether these systematic patterns can be exploited to create arbitrage-like long-short strategies, going long and short at particular times during the day. The answer is yes; even when accounting for realistic transaction costs and margin requirements, some of the proposed strategies can create consistent positive and significant CAPM alphas.

We study market efficiency for high-frequency Brent Crude oil futures prices across intra-daily sampling frequencies ranging from one minute to two hours using a sample period from 2006 to 2021. The efficiency dynamics of Brent crude oil futures prices are scrutinized over time using rolling estimation windows. We also propose to study intra-daily market efficiency using a signature plot across sampling frequency. Our results show substantial differences in market efficiency at the intra-daily level. At very high sampling frequencies (less than 10 min), the market is on average inefficient. For lower sampling frequencies, efficiency in Brent oil futures generally improves over the examined sample period.

We explore the impact of stochastic mortality, health, and parasites on animal-based commodities risk management, with a specific emphasis on salmon aquaculture. Our investigation delves into the stochastic nature of mortality, fish lice infestation, and treatment plans based on comprehensive historical data from Norway. Given that salmon lice pose a significant challenge to salmon aquaculture, with associated treatment costs estimated to be comparable to feeding expenses, lice removal is imperative to ensure the survival of the salmon and comply with the Norwegian government's stipulation of maintaining an upper threshold of 0.5 lice per fish. We propose a new model that considers the relationship between hosts and parasites and determines the number of treatments required as well as the overall cost of these treatments. An important aspect of our model is its incorporation of stochastic effectiveness for each removal. After calibrating the model to our dataset, our study examines the impact of the host-parasite relationship and the required interventions on the optimal harvesting decision and draws a comparison to models that make an assumption of deterministic mortality. Our results indicate that a gain of 1.5 % in farm value (per rotation) can be obtained by employing an optimal harvesting rule based on the stochastic host-parasite model.

Within the context of forecasting U.S. inflation, this study explores the predictive power of food commodities futures prices, focusing on enhancing the precision of both short- and long-term forecasts. We develop single commodity models for twelve different food commodities and also construct two aggregated models: a simple component model and a Principal Component Analysis (PCA)-based model, both utilizing price indices of the selected commodities. Monthly futures data from 1996 to 2023 for the nearest maturity dates are segmented into in-sample fitting and out-of-sample forecasting, covering forecast horizons of 3, 6, 9 and 12 months. Our findings indicate that aggregated models, particularly the PCA-based method, exhibit superior forecasting performance. Furthermore, to verify model robustness, we conduct parallel forecasts using spot prices and perform subsample analyses. Collectively, these results underscore the predictive power of commodity futures for forecasting food inflation.

We investigate a utility maximisation problem for a pension scheme which offers an income drawdown policy. Apart from market risk, we specifically focus on longevity basis risk which arises when the forces of mortality of the scheme's target population and the reference population of a longevity bond used for hedging, are not perfectly correlated. By modelling the forces of mortality of the reference and target populations as stochastic affine processes, we derive analytic solutions for the relevant investment strategy and benefit withdrawal rate. Our model also accounts for dependencies between mortality rate fluctuations and stock prices. Our extensive numerical results demonstrate that the longevity bond acts as an effective hedging instrument, even in the presence of longevity basis risk.

We determine forest lease value and optimal harvesting strategies under model parameter uncertainty within stochastic bio-economic models that account for catastrophe risk. Catastrophic events are modeled as a Poisson point process, with a two-factor stochastic convenience yield model capturing the lumber spot price dynamics. Using lumber futures and US wildfire data, we estimate model parameters through Kalman filtering and maximum likelihood estimation and specify the model parameter uncertainty set as the 95% confidence region. We numerically determine the forest lease value under catastrophe risk and parameter uncertainty using reflected backward stochastic differential equations (RBSDEs) and establish conservative and optimistic bounds for lease values and optimal stopping boundaries for harvesting. Numerical experiments further explore how parameter uncertainty, catastrophe intensity, and carbon sequestration impact the lease valuation and harvesting decision. In particular, we explore the costs arising from this form of uncertainty in the form of a reduction of the lease value. These are implicit costs which can be attributed to climate risk, and are likely to become more significant as forestry resources play a larger role in the energy transition. We conclude that in the presence of parameter uncertainty, it is better to lean toward a conservative strategy reflecting, to some extent, the worst case than being overly optimistic. Moreover, our results suggest that convenience yield plays a substantial role in determining optimal harvesting strategies within the two-factor model adopted in this study.

In corporate finance, we usually face the problem of pricing a type of asset, of which the cash flow is a linear function of the firm total cash flow up to the first time the latter exits from a given domain when their claimant receives a lump-sum payoff in addition. We call them standard assets. We provide a unified explicit pricing formula for all standard assets in a double-exponential jump-diffusion cash flow model with regime-switching. We explicitly derive prices of corporate securities involving recently introduced Total Loss Absorption Capacity (TLAC) Bonds in the mixed model. Our numerical analysis shows that the optimal coupon rate of TLAC bonds in the boom regime is substantially higher than that in the recession regime; thus, the firm should issue more TLAC bonds in a better economic environment.