While search engines apply a single optimised search strategy to any user query, selective search, like selective query expansion, aims to apply an adapted search strategy to each query. Search phases include query expansion, search-weighting model, and document ranking. A search strategy is defined by the combination of components and their hyperparameters in these phases. The number of possible search strategies is huge. In this paper, we describe a risk-sensitive approach to optimise the set of search strategies that should be included in a selective search approach. It solves the problem of which and how many search strategies to include in the system. We found that using 20 search strategies is an appropriate trade-off between effectiveness and system complexity. Significant effectiveness improvement is about 23% when compared to L2R documents and about 10% when compared to other selective approaches. This paper is an extended abstract of our paper at CIKM 2021 1 .