(3)It is important that an administrator is able to construct a benchmark methodology that is resilient to different market conditions and that allows for the calculation of the benchmark in the widest set of possible market circumstances. The benchmark methodology used should therefore rely on adequate and appropriate historical values of the benchmark. For the same reason, and in particular to validate the calculation of the benchmark and to evaluate the performance of the methodology ex post , the benchmark methodology should be back-tested against available transaction data. Such back-testing should be an ex post exercise, which either uses additional available data that were not used for the calculation of the benchmark or other sources of transaction data, or reconstructs the historical values of the benchmark. To ensure that the benchmark methodology is capable of validation, it is important that back-testing is performed both at each annual review of the benchmark methodology and, depending on the type of the benchmark, either after each material change of that methodology, on an ongoing basis, or at the first provision of the benchmark. The findings from the back-testing should be reflected in the methodology.