Some alternatives to measure volatility of energy price series

Resumen

In general, the volatility measures are computed through the standard deviation of the price series orthe log-return series. However, this approach does not reflect the actual degree of uncertaintyassociated to a price data series. In this work we propose the Permutation Entropy, TopologicalEntropy and the Modified Permutation as alternative volatility measures. Using simulated data, weshow that these measures are appropriate to quantify the degree of randomness or determinism of atime series, that is, to measure volatility of a price series. Moreover, volatility can be computed underno assumptions on the price series and even its nonlinear dynamic is taken into account.The new measures were applied to different electricity markets (Nord Pool, Omel, Ontario and fourAustralian markets) and the results showed their capability of relating the periods with high and lowvolatilities to historical events or climatic factors. Therefore, the proposed measures help to identifythe factors that can produce changes in the predictability of the price series (such as loads, climaticfactors, market regulations, etc.). Moreover, they can identify the time periods with a high degree ofdeterminism and find out whether seasonal behavior exists. For example, they could determine thatsummers are less volatile than the rest of the seasons, and this fact would be applied to improveforecasting price models.The proposed volatility measures might also be useful to compare the uncertainty associated withdifferent energies (electricity, oil, gas, etc.), to determine which of them is more predictable. As aconsequence, we believe that this new approach to measure volatility of the prices provides investors,customers and regulators an important feedback