I have data on returns for a certain type of stock (we shall call it class A) for several markets, market return data and the risk free rates over the period. I have been asked to assess whether there is an anomaly for the class A stocks using 'CAPM as the underlying pricing model'.

So far I have performed the following CAPM regression for each market:

rm-rf = a + b(rm-rf)

I have two questions:

1. I found the p-value for 'a' to be significant for one of the markets, does this mean that CAPM is not suitable for it?

2. To find this 'anomaly' do i compare rf+β(rm-rf) with the observed class A returns? Or is there a hypothesis test I should run?