For each period compute the percentage below the running high-water-mark, square it, average across all periods, take the square root. Squaring makes deep drawdowns dominate; averaging over time makes long ones count.
It captures something no single-point measure does. Two strategies both had a 25% worst drawdown; one recovered in two months, the other stayed underwater for three years. Maximum drawdown says they are equal, the ulcer index says they are nothing alike - and anyone who held the second one agrees.
Return divided by ulcer index gives the ulcer performance index, sometimes called the martin ratio, which is arguably the most psychologically honest risk-adjusted measure available. Its drawback is unfamiliarity: there are no widely shared benchmarks for what a good value looks like, so it is best used to compare your own strategies against each other. See underwater-curve.
Related: underwater-curve, max-drawdown, drawdown-duration, calmar-ratio