This study analyses momentum returns in 54 countries covering 34 years. It is the first study where optimising programmes are applied to momentum returns and portfolio selection. Momentum returns have remained a contentious topic and a substantial amount of research purports to support and negate this anomaly. Momentum studies have previously concentrated on finding the cause of this anomaly or determining whether abnormal returns are present only in a particular dataset. However, there is no clear consensus regarding how to construct/implement a momentum strategy. To date it has been unclear as to how many portfolios should be created, whether or not the portfolios should be equally-weighted or value-weighted and how the momentum returns will change when a chosen strategy is constructed/implemented compared to an alternative strategy. This lack of precision concerning how to construct/implement momentum strategy potentially leads to confounding results. This current thesis contributes to an understanding of the structure impact effect by computing momentum returns for changing portfolio structures and observing the magnitude of the impact on returns on a specially crafted database of 52,593 stocks from 54 countries over the time period 1973-2007. The two important empirical extensions are the industrial momentum and the 52-week high momentum models. Both the industrial momentum and 52-week high momentum strategies claim that their returns are superior to the traditional momentum return and possess superior explanatory power. Empirical evidence, to date, has not been available to attest to whether the results hold true when applied in different markets. This gap in knowledge motivated this research investigation of multiple countries addressing how different methods of calculating returns, different approaches to momentum strategy, different portfolio weighting process impact upon the robustness of results. This research also addresses the question of whether momentum returns can be increased through the use of optimisation algorithms. Traditionally, little attention has been paid to the portfolio weighting with either an equal-weighted or value-weighted approach to allocating funds to the Winner and Loser portfolios. This study proposes an alternative way of allocating money to the Winner and Loser portfolios with the goal of generating increased returns. Eight different algorithms are applied to the share returns to determine whether one method is clearly superior to others in maximising the momentum returns for the synthesised portfolios over a period of time. This is the first study of its type where optimising programmes are applied to momentum returns and portfolio selection and covers several countries. The results indicate that momentum returns are robust on a global scale and the returns are by and large statistically significant under different portfolio construction approaches. Both the industrial and 52-week momentum strategies remain positive and statistically significant but do not generate the same magnitude of returns as the conventional momentum return model. The optimisation of momentum return shows promising results as a number of optimisation techniques do enhance momentum returns. As the potential to increase returns becomes known, traders will quickly react and there is likely to be a range of new investment products arising.