Conventional measures of risk in earnings based on historical standard deviation require long time series data and are inadequate when the distribution of earnings deviates from normality. We introduce a methodology based on current fundamentals and quantile regression to forecast risk reflected...