Sortino before Sharpe: why we report the downside first

Every allocator has seen the deck: a smooth equity curve, a Sharpe ratio quoted to two decimal places, and a footnote-sized maximum drawdown. The implication is that risk has been summarized. It hasn't. It has been averaged — and averaging is where the information dies.

What Sharpe actually rewards

The Sharpe ratio divides excess return by the standard deviation of all returns — upside and downside alike. That symmetry has a perverse consequence: a strategy that occasionally jumps upward is penalized exactly as if it had crashed. Two strategies with identical Sharpe ratios can have radically different loss behavior, and the ratio will never tell you which one you own.

For strategies with asymmetric payoff profiles — which describes most derivatives strategies, and certainly ours — the distortion is not a technicality. Option-selling programs look magnificent on Sharpe right up until the tail event that was always in the distribution. Convex strategies look mediocre on Sharpe precisely because their best months are "volatile."

If your strategy is designed around the shape of the payoff, evaluating it with a shape-blind statistic is self-sabotage.

The downside ledger

We lead every internal and external evaluation with what we call the downside ledger — a set of statistics that only move when something bad happens:

Sortino ratio. Excess return divided by downside deviation. Volatility you enjoyed is not risk; volatility you suffered is.
Maximum drawdown, and time under water. Depth tells you the worst moment; recovery time tells you how long conviction had to survive. Allocators redeem during the second one.
Calmar ratio. Return per unit of maximum drawdown — the price of the record's worst stretch.
Loss-tail statistics. The average of the worst 5% of days, the skew of the loss distribution, the longest losing streak. These describe the experience of holding the strategy, not the summary of it.
Profit factor and exposure. Gross gains against gross losses, and how much capital was actually at risk to produce them.

None of these numbers is exotic. What matters is the ordering: they come first, together, before any headline return is discussed. A strategy that cannot survive this page does not get a second page.

Why together matters

Any single downside statistic can be gamed, usually by accident. A strategy can post a pristine maximum drawdown because it has not traded through a hostile regime yet. A high Sortino can coexist with a slow bleed that never triggers a drawdown threshold but never makes money either. Reporting the ledger as a unit makes the failure modes visible: depth, duration, frequency, and tail shape have to be consistent with each other, and with the regime history the strategy actually traded through.

This is also why we distrust any presentation that leads with a single ratio — including a Sortino. The point is not to crown a better statistic. The point is that risk is a distribution, and a distribution cannot be summarized by one number without deciding, in advance, which questions you would rather not be asked.

The standard we hold ourselves to

When we show a forward record, the downside ledger comes attached — computed from the same auditable records as the returns themselves, over every regime the strategy has traded, with no "representative period" selection. If a number looks worse in a specific volatility or liquidity state, that decomposition is part of the record too.

A track record is a claim. The downside ledger is the cross-examination — and we would rather run it on ourselves, continuously, than have someone else run it once, later, with money on the line.