I said independent variables must not be perfectly correlated, error term should be normally distributed, Homoscedasticity. While that doesn't include all, he kept interrupting me and said all of those are wrong.