Heterosis, i.e. the increased performance of the progeny relative to its parents, is a phenomenon that is systematically exploited in plant breeding with allogamous species to produce high-yielding, resilient varieties. In the case of selfing species, however, the use of heterosis in commercial breeding is severely limited. In addition to the challenges of producing hybrid seed cost-effectively, our limited knowledge of heterosis hinders hybrid breeding in selfing species. To overcome this bottleneck, we have developed a quantitative genetic framework to study the genetic components of heterosis in wheat. In my scientific talk, I will present new results in which we have developed advanced statistical methods that leverage Big Data in wheat for heterosis research. The results show the importance of the interaction between genes, statistical epistasis, for heterosis.