This paper proposes a method for asset allocation based on partitional clustering. This method is different from the approaches already proposed in the literature, which essentially use either an optimization-based approach or a hierarchical clustering algorithm to allocate resources in assets. After finding the clusters, the method uniformly allocates the resources over the clusters and then within the clusters, thus guaranteeing that all assets are allocated. The method was tested using data from the Brazilian Stock Exchange (B3) and the assets eligible to enter the allocation were those that were part of the Ibovespa Index at the time of portfolio rebalancing. The results were compared with the Ibovespa index for different metrics, such as volatility, return, sharpe ratio, turnover and drawdown. The proposed approach illustrates the potential of machine learning techniques in portfolio allocation.