It has been indicated by Litterman and Scheinkman (1991) that the term structure of interest rates is reliably modeled by an affine three factor model attained by principal component analysis. Such a model is inconsistent with no-arbitrage. In this paper Reisman and Zohar derive an explicit formula for the theoretical rate of profit from such arbitrage, use historical data of US Treasuries to estimate its parameters, and find that excess returns predicted by their formula are surprisingly large. Moreover, the returns that are obtained in practice are in excellent agreement with the predicted ones. Their result can be applied to improve hedging and to beat a benchmark portfolio