How to use a t-test to compare a binary classification to pure random data?

So I am new to this and I am not quite sure how to do this. I could be over thinking it.

I have a binary classification where all the data is labelled as Win or Lose. THere is an equal amount of both. I am using libSVM on weka with 10-fold cross validation and I am getting results of around 55%. Since that is just slightly higher than a coin flip I want to ensure that it is actually 55% and not just luck of the data so I need to conduct a t-test I believe.

But what I am conducting a t-test against? The success of each of the folds and what? Just 50%s?

Or is there another way to approach this problem that I am overthinking?

edit Update

Can I just use the student's t-test with one vector of the results of each of the ten folds compared to a vector of all 0.5 and see if that results in a p < 0.05?