Raw return is uninformative on its own. A 60% year at 5x leverage and a 12% year unlevered can be the same skill expressed at different sizes - and the levered version is one bad week from being a minus 60% year.
The family of measures differs only in what goes in the denominator: total volatility for the sharpe-ratio, downside volatility for the sortino-ratio, maximum drawdown for the calmar-ratio, depth and duration of drawdown for the ulcer-index. Each encodes a different definition of what counts as risk, and a strategy can look excellent under one and poor under another. Compute at least two.
The practical use is allocation. Between two strategies, the higher risk-adjusted return can be levered up to match the other's volatility and will still win; raw return comparisons cannot tell you that. What none of them capture is the tail, so a high ratio from selling options is not the same as a high ratio from trend following. See return-skew.
Related: sharpe-ratio, sortino-ratio, calmar-ratio, return-skew