In reality, I am trying to create a control showing price depths of a stock which needs to accept a large number of messages and summarizing them for display purposes. Will I get a better result if I create it in MFC and use that control in my .Net Winform application than writing the whole thing in .NET?
The short answer is yes, BUT... The extra time and hassle it will take to write a control in C++ rather than using .Net will not be worth it. If used correctly, .Net is highly performant and C++ if/when used poorly will be a slug.
I would write the whole thing in .NET, profile if it's too slow, then if it is, port only the processing code to C++ and use P/Invoke to get the summarized results for a .NET control. Trying to interface an MFC control with .NET can take a long time and be prone to error, whereas if you write the display logic in .NET and use C++ for any very performance dependent code you'll have a much easier time.
Have to take into account the overhead for switching to unmanaged code as well. Make it in c# first. IF it is too slow and you can't optimize the c# then look at going to c++.