Let us have the standard single-period binomial pricing model, and denote the up and down states of the underlying by $S_u$,$S_d$ respectively. Let us say we have a call option on the underlying with strike $K$ such that: $$ K < S_d < S_u $$ I.e., an option that will surely be in the money. This (apparently, according to these notes: http://galton.uchicago.edu/~lalley/Courses/390/Lecture1.pdf) violates the no-arbitrage assumptions of the model.
Question: What is the arbitrage opportunity here? It seems that if the option is expensive enough there will be no arbitrage.