Calmar and Sortino are both risk-adjusted metrics, but they answer different questions.
If you treat them as interchangeable, you can approve fragile strategies.

What each metric measures

Sortino ratio focuses on downside volatility: return divided by downside deviation.
Calmar ratio focuses on drawdown pain: annualized return divided by max drawdown.

Sortino asks: "How efficiently do you generate return relative to bad volatility?"
Calmar asks: "How much return do you earn per unit of worst capital damage?"

Why traders mix them up

Both are "higher is better" and both punish risk. But they punish different risk shapes.

A strategy with smooth daily losses can look acceptable on Sortino while still having one deep drawdown event that crushes Calmar.
A strategy with shallow but frequent losses may hurt Sortino more than Calmar.

When Sortino is more useful

Use Sortino as a first pass when:

your strategy has many observations (enough bar-level history),
you care about consistency of downside fluctuations,
you compare similar strategies in the same market regime.

Sortino is weaker when tail events are rare in-sample but severe out-of-sample.

When Calmar is more useful

Use Calmar as a deployment gate when:

drawdown tolerance is hard-capped,
your capital allocator cares about worst pain more than smoothness,
you need a capital-preservation check before going live.

Calmar is sensitive to one extreme period. That is useful for protection, but noisy on short samples.

Practical workflow for strategy validation

Do not pick one metric in isolation. Use this sequence:

Check sample quality and trade count.
Review Sortino for downside efficiency.
Review Calmar for drawdown burden.
Validate stability with walk-forward results.
Stress test costs and slippage.

If Sortino is strong but Calmar is weak, the strategy may hide occasional deep risk.
If Calmar is strong but Sortino is weak, the strategy may be capital-efficient but operationally noisy.

Typical failure patterns

Overfit hyperparameters produce nice Sortino in-sample and collapse in OOS.
Under-modeled costs inflate return and therefore both metrics.
Regime dependency can keep Sortino high until a volatility regime shift cuts Calmar sharply.