A question on the mason's gain and maximum stable gain

Dear all,
I am currently designing a LNA working at high frequencies. and I have some problem on understanding the Mason's gain and the maximum stable gain

I know that Mason's gain is the maximum power gain that is defined in the condition that the transistor is first unilateralized. And the MSG is the maximum gain that could be achieved when it is stabilized. And there should be a kink over the MSG, at the point of stability factor K=1.

I simulated these two curves in spectre. as shown in the attachment. I could see the MSG is below the Mason's gain, and there is the kink of MSG. But I could understand that the MSG overshoots the Mason's gain around the kink.

I think it is violating the Mason's theory. Am I understanding this in a wrong way? or there is some problems with the Spectre accuracy?
If this is normal, Can I make the LNA working in that region, so that a better gain could be achieved.....

Thanks for your help.

jimiblues

I have updated the question, have a look, thank you guys