Expectiles are a coherent and elicitable alternative to commonly used market risk measures, but practical backtesting tools have lagged behind. This study proposes new backtests that separate unconditional coverage from independence in expectile forecasts. By connecting a Wald-testing framework to Box–Pierce-style autocorrelation testing under a location-scale setting, the authors develop tests with improved finite-sample behavior in simulations and demonstrate their use on S&P 500 return data.

In a pragmatic cluster non-randomized trial, Alexandra Kulinkina and colleagues investigate the effectiveness of a clinical decision support algorithm for guiding antibiotic prescribing in pediatric outpatient care in Rwanda.