High Watermarks of Market Risks.

We present several estimates of measures of risk amongst the most well-known, using both high and low frequency data. The aim of the article is to show which lower frequency measures can be an acceptable substitute to the high precision measures, when transaction data is unavailable for a long history. We also study the distribution of the volatility, focusing more precisely on the slopee of the tail of the various risk measure distributions, in order to define the high watermarks of market risks. Based on estimates of the tail index of a Generalized Extreme Value density backed-out from the high frequency CAC 40 series in the period 1997-2006, using both Maximum Likelihood and L-moment Methods, we, finally find no evidence for the need of a specification with heavier tails than in the case of the traditional log-normal hypothesis.