1: \begin{abstract}

2: Discretely sampled variance and volatility swaps trade actively in OTC

3: markets. To price these swaps, the continuously sampled approximation is

4: often used to simplify the computations. The purpose of this paper is to

5: study the conditions under which this approximation is valid. Our first set

6: of theorems characterize the conditions under which the discretely sampled

7: swap values are finite, given the values of the continuous approximations

8: exist. Surprisingly, for some otherwise reasonable price processes, the

9: discretely sampled swap prices do not exist, thereby invalidating the

10: approximation. Examples are provided. Assuming further that both swap values

11: exist, we study sufficient conditions under which the discretely sampled

12: values converge to their continuous counterparts. Because of its popularity

13: in the literature, we apply our theorems to the 3/2 stochastic volatility

14: model. Although we can show finiteness of all swap values, we can prove

15: convergence of the approximation only for some parameter values.

16: 

17: KEY\ WORDS: variance swaps, volatility swaps, NFLVR, semimartingales

18: \end{abstract}

19: