Efficiency, predictability and liquidity in the commodity futures markets
|Author:||Bc. Vojtěch Čermák|
|Year:||2015 - summer|
|Leaders:|| prof. PhDr. Ladislav Krištoufek Ph.D.
|Work type:|| Bachelors
|Awards and prizes:|
|Abstract:||This thesis examines efficiency of several CME commodity futures and its relation to market liquidity
over the ten years period. The goal is to find ARMA model that is better than white noise in terms of
forecasting power and carry out analysis of market liquidity if we find such model. This is done by
comparing selected ARMA models to white noise. In order to do that, we use Diebolt - Mariano test
on forecast errors obtained by pseudo out - of - sample analysis using rolling window with re -
estimation. Concern of furhter analysis are factors, that can influence the DM statistics.
Main findings are, that we are able to find such ARMA model for small enough time period within the
ten years period for almost all commodities. For most commodities, this sub period is not long
enough to violate efficient market hypothesis. Only for palladium and lean hog futures this period is
longer than one year. These two futures shows strong signs of inefficiency, as its predictability is not
out - weighted by liquidity restrictions.