Exactly what are "sgima3 lines"?
Please read this link before attaching your data with the paper clip icon to allow people to answer you.
Otherwise all we can say is to subtract the fit from your actual, and compute the residuals or something like that.
Note: if your data does not include uncertainties (i.e., no noise with respect to a line), meaning it fits a line perfectly, then there will be no error in the slope and intercept you get from polyfit()