Financial hedging in energy market by cross-learning machines
- 14 Downloads
The recent price volatility in the energy market highlights the importance of financial hedging and the need of its incorporation into an investor’s set of portfolio strategies. Realizing the rapid advancement of artificial intelligence, our study empirically examines the use of machine learning models for hedging price risk associated with the holding of energy-based financial product. From a technical perspective, kernel regression and support vector machine are trained to estimate the time-varying optimal hedge ratio given the observed price movement and other factors. The estimated hedge ratio is then employed to guide the price hedging strategy of the crude oil contracts traded in the commodity exchanges. The two machine approaches’ hedging effectiveness against price risk is also compared with those of the un-hedged portfolio as well as a well-studied econometric time series approach. Our results indicate that the two forms of learning machines substantially outperform time series model and no-hedging over the out-of-sample period but neither machine dominates another across all testing scenarios. Given these mixed results between kernel regression and support vector machine, we propose and develop a kernel-supervised support vector machine to take advantage of the complementary nature of the two machine learning approaches and enhance the supervised learning process through hierarchical/sequential infusion of information. Cross-learning empirical testing shows that the proposed cross-learning machine is more effective in hedging than individual kernel regression and support vector machine. Furthermore, our study evaluates the impact of incorporating the coefficient of absolute risk aversion and the transaction costs to machine learning models. In general, cross-learning machine outperforms others in all tested scenarios.
KeywordsEnergy financial market Time-varying commodity hedging Machine learning Cross-training algorithm Kernel regression Support vector machine
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 9.Kroner KF, Sultan J (1991) Exchange rate volatility and time varying hedge ratios. In: Rhee SG, Chang RP (eds) Pacific-basin capital markets research. Elsevier Science Publishers, North-Holland, pp 397–412Google Scholar
- 16.Lahmiri S (2013) Hybrid systems for Brent volatility data forecasting: a comparative study. Uncertain Supply Chain Manag 1:145–152Google Scholar
- 26.Pan Z, Sun X (2014) Hedging strategy using copula and nonparametric methods: evidence from China securities index futures. Int J Econ Finan Issues 4:107–121Google Scholar
- 29.Drucker H, Burges CJC, Kaufman L, Smola AJ, Vapnik VN (1997) Support vector regression machines. In: Advances in Neural Information Processing Systems 9, NIPS 1996, pp 155–161, MIT PressGoogle Scholar
- 35.Xie C, Mao Z, Wang G (2015) Forecasting RMB exchange rate based on a nonlinear combination model of ARFIMA, SVM, and BPNN. Math Probl Eng 2015:1–10Google Scholar
- 39.Ahmadi E, Jasemi M, Monplaisir L, Navavi M, Magnoodi A, Jam P (2018) New efficient hybrid candlestick technical analysis model for stock market timing on the basis of the Support Vector Machine and Heuristic Algorithms of Imperialist Competition and Genetic. Expert Syst Appl 94:21–31CrossRefGoogle Scholar
- 41.Lahmiri S (2014) Entropy-based technical analysis indicators selection for international stock markets fluctuations prediction using support vector machines. Fluct Noise Lett 13:1450013.1–1450013.16Google Scholar
- 42.Casdagli M (1992) A dynamical systems approach to modeling input-output systems. In: Casdagli M, Eubank S (eds) Nonlinear modeling and forecasting vol XII of SFI Studies in the sciences of complexity. Addison-Wesley, Reading, p 265Google Scholar
- 50.Lahmiri S, Boukadoum M (2015) An ensemble system based on hybrid EGARCH-ANN with different distributional assumptions to predict S&P 500 intraday volatility. Fluct Noise Lett 14:1550001.1–1550001.10Google Scholar