This thesis consists of three empirical studies in market-based accounting research. In particular, this thesis focuses on earnings forecasts and the market valuation of profit-making and loss-making firms separately. Overall, these studies contribute to the understanding of forecasting earnings and the properties of the resulting estimates of both profit and loss persistence classifications in understanding the valuation of profit-making and loss-making firms. In the first study, we investigate the possibility of building better cross-sectional models to forecast earnings for profit-making and loss-making firms. We first examine the accuracy of the Hou et al. (2012) (HDZ) model when generating one year-ahead earnings forecasts for profit and loss-making firms separately. We then develop an extended cross-sectional earnings forecasting model that contains all the financial statement items that are reported to be useful for forecasting earnings and the valuation of the firms in prior studies. Our findings suggest that it is better to develop a cross-sectional earnings forecasting model for profit and loss-making firms separately. Further, our expanded models outperform the HDZ models in terms of forecast accuracy for profit-making and loss-making firms generally. In the second study, we examine the ability of one year-ahead earnings forecasts to capture the future prospects of profit-making firms. We use the extended cross-sectional earnings forecasting model developed in the first study to compute the earnings forecasts. We then introduce a classification scheme that assigns profit-making firms into two categories based on whether firms are expected to report a profit (persistent) or a loss (transitory) in the next year. Building on a simple earnings and book value valuation model, we find that our one year-ahead earnings forecasts have an incremental value over and above current earnings and book value in the valuation of profit-making firms. Furthermore, the relative valuation importance of our one year-ahead earnings forecasts, current earnings, and book value depends on profit persistence as defined by our earnings forecasts. In the third study, we examine the ability of one year-ahead earnings forecasts to capture the future prospects of loss-making firms. We use the extended cross-sectional earnings forecasting model developed in the first study to compute the earnings forecasts. We then classify loss-making firms into persistent (negative earnings forecasts) and transitory (positive earnings forecasts) groups based upon the sign of the forecasted earnings. Using the Darrough and Ye (2007) valuation model as our baseline model and the sample of loss-making firms, we demonstrate that the earnings forecasts have incremental information content over and above other important value drivers that are used to indirectly capture a firmâs future prospects. In addition, investors price our earnings forecasts conditional upon loss persistence. Further, investors price both current earnings and book value conditional upon loss persistence.