This paper takes a new look at the relation between volume and realized volatility. In contrast to prior studies, we decompose realized volatility into two major components: a continuously varying component and a discontinuous jump component. Our results confirm that the number of trades is the dominant factor shaping the volume-volatility relation, whatever the volatility component considered. However, we also show that the decomposition of realized volatility bears on the volume-volatility relation. Trade variables are positively related to the continuous component only. The well-documented positive volume-volatility relation does not hold for jumps.
- Bi-power variation