# Are the returns in this regression signed returns?

In this paper about combining multiple alphas are the returns signed returns? if not wouldn't they be mean zero?

Also, it mentions "realized alpha returns" - does that just mean "realized" past alpha return forecasts or what the real actual return was? And, for the "expected alpha return" that is the independent variable in this regression, isn't the "expected alpha return" just the mean real return (i.e. zero)? Since the alpha is the $$\hat{y}$$ in the regression where it is fit to the return..