Spare parts are particularly challenging to forecast due to their lumpiness and representing a significant part of companies’ expenditures, so even small improvement in new approaches can considerably reduce these items’ total inventory. This paper aims to present a new hybrid forecasting method for spare part inventory management using heuristics and bootstrapping approaches to improve spare parts forecasting in normal use phase in spare parts inventory management. Our study presents an innovative methodology to model autocorrelation in demand to represent data distribution in bootstrapping in alternative to transition probabilities. The results were evaluated and validated through a case study on real data from a large iron ore corporation in Brazil, focusing on demand patterns and their impact on overall costs compared to leading-edge techniques. The mineral sector was selected due to its significant contribution to the emerging Brazilian economy and the lack of research in this field. The results revealed significant improvement in the forecasting total cost reduction up to 40% over leading-edge techniques for erratic and lumpy demand. Results suggest that relaxing autocorrelation in bootstrapping samples could lead to better deal with higher variability in demand sizes in spare parts management compared to parametric methods, as we recommend that this method should be particularly considered when dealing with spare parts with lower intermittence compared to other bootstrapping approaches. The method can be applied in any sector without restrictions, and provides managers with a systematic tool to analyze the trade-off between holding and breakage costs of spare items as well as demand parameters for the mining sector.