If you are really interested in the downside risk Steiner's version is the best. Then you have more insight about the tail. This is very important when looking at regimes.
I think this version is more practical and quite better than Steiner's. Also, is faster since it does not computes the whole drawdown vector (which I don't need in particular) and its more understandable. Advantage vs. MATLAB's routine: data can be <= 0, and still computes (i.e. MaxDD = 120%, for instance).
Alexc, I don't have the Financial Toolbox, so I can't help you wih MATLAB's maxdrawdown routine. My Version simply computes the maximum relative drawdown of a given time series.
The input for your function is a vector of cummulative returns, not a period return vector. You should point that out in your description of the arguments.
Personally, I prefer my version because it allos not only computation of maximum drawdown, but computation of the full drawdown vector to for example plot time-under-water charts and calculation of the average of the n-th largest drawdowns (needed in the calc of certain risk-adjusted performance measures).
Comment only