This thesis introduces two algorithmic statistical arbitrage trading strategies based on the fixed hedge ratio of the Engle-Granger cointegration regression and the daily forward-looking hedge ratio of the Kalman filter. Both trading strategies have the objective to exploit short-term deviations from the stochastic long-term equilibrium between country-specific commodity currency futures and commodity futures of Australia, Canada, New Zealand and South Africa based on daily futures prices in the time period from 2005 until 2013. The empirical results suggest that the cointegration relationship between commodity currency futures and commodity futures is highly unstable and switches between a non-cointegrated and a cointegrated regime over time. The error correction models show that commodity futures are weakly exogenous and that commodity currency futures mainly react to short-term deviations from the long-term equilibrium. In addition, the Kalman filter reveals that the pair-specific hedge ratios are highly sensitive over time. The thesis demonstrates that both trading strategies are suitable to exploit statistical arbitrage opportunities based on different combinations between the trading threshold and convergence target. However, the profitability of both trading strategies declined out-of-sample owed to the regime switches in the cointegration relationship and the smaller size of the price anomalies. Further research should focus on the time-varying properties of the hedge ratios and the causes for the regime switches in the cointegration relationship including the implementation of non-linear, respectively regime switching models. Also the pair-specific holdings of the portfolios could be optimised and the performance of the trading strategies tested on high frequency data.