A strategy returning 15% a year with a 30% worst drawdown has a Calmar of 0.5. One returning 8% with a 10% drawdown has 0.8 and is the easier thing to keep trading through, which is a real advantage rather than a statistical one.
Its weakness is that max-drawdown is a single observation, the worst thing that happened once. It has enormous sampling error, it grows mechanically with the length of the track record, and it says nothing about the rest of the distribution.
Use it alongside distribution-based measures, not instead of them. A better version of the same question is the median of the worst drawdowns across a block-bootstrap, which gives a range rather than one unrepeatable number.
Related: max-drawdown, sortino-ratio, sharpe-ratio, block-bootstrap