1: \begin{abstract}

2: In quantitative finance, we often model asset prices as

3: semimartingales, with drift, diffusion and jump components. The jump

4: activity index measures the strength of the jumps at high

5: frequencies, and is of interest both in model selection and fitting,

6: and in volatility estimation. In this paper, we give a novel

7: estimate of the jump activity, together with corresponding

8: confidence intervals. Our estimate improves upon previous work,

9: achieving near-optimal rates of convergence, and good finite-sample

10: performance in Monte-Carlo experiments.

11: \end{abstract}