An argument about a track record is usually an argument about what it is being compared to. The returns themselves are arithmetic and can be checked. What they should be measured against is a choice, and whoever makes that choice has already decided most of the answer.
A benchmark is not a formality at the bottom of a factsheet. It is a claim: this is what would have happened to your money if we had not been involved. Get it wrong and the entire exercise measures something nobody cares about, however carefully the arithmetic was done.
A benchmark of zero, applied to money that would otherwise have been invested.
If the alternative to your fund is holding an index, beating zero is not an achievement, it is arithmetic with a long enough horizon. Any positive drift in the underlying market shows up as skill. The manager is being credited for the passage of time.
The mechanism is worth stating precisely, because it is what makes the comparison useless rather than merely generous: the benchmark has to contain the drift the strategy is exposed to, or the drift lands in the result and is read as skill. Zero contains no drift at all.
The question to ask is therefore not "did you make money" but "did the money do better with you than it would have done without you", and those are different questions whenever the alternative was not literally a bank account.
The opposite trick is subtler and shows up in more respectable places.
A strategy that keeps most of its capital idle and compares itself to a fully invested index is being flattered by the comparison. It is being measured on capital it never put at risk. Return on the money actually deployed and return on the money entrusted are different numbers, and the gap between them is a decision the manager made, not a result they produced.
Neither number is wrong on its own. Publishing only the convenient one is. State the average deployment alongside the return, or the return cannot be interpreted.
A benchmark has to be denominated in whatever the investor counts their wealth in, and this is where comparisons quietly fall apart.
The same track record can look excellent in one currency and unremarkable in another without a single trade changing. Nothing about the trading differs between the two views. Only the unit of account does, and the unit of account is chosen by whoever writes the report.
The rule that follows: state the currency of the return series in the same breath as the return. If the two lenses disagree, publish both. The disagreement is information about the structure of what you are doing, not an inconvenience to be smoothed away by choosing the flattering one.
There is a situation in which a benchmark of zero is not a dodge but the only honest choice, and it is worth understanding because the principle generalises.
If an account is denominated in the asset itself, so that positions, profits and losses are all counted in that asset rather than in currency, then "hold the asset" is identically zero by construction. There is no drift to be credited with, because the drift is the unit.
The investor in that situation already owns the asset. Their real alternative is not a bank account and not an index: it is doing nothing with what they already hold. A zero benchmark answers their question literally, and a dollar benchmark would answer a question they never asked.
The general principle underneath: the benchmark is the investor's actual alternative, not the convention in your category. Sometimes that is an index, sometimes a deposit rate, sometimes the asset itself, and occasionally it is another manager they would otherwise have hired.
The Sharpe ratio is defined on excess return, so subtracting a risk-free rate is not a convention somebody chose: it is in the formula. Which means the assumption travels with the ratio whether or not it fits.
It fits when the alternative genuinely was to hold cash: the capital was uncommitted and could have earned that rate for no risk. It does not fit when the capital was always going to be invested in something, because then the rate being subtracted is not the alternative. It is an assumption inherited from the formula rather than from the situation.
The failure mode is mechanical rather than dishonest. Someone reaches for a ratio, the ratio has a risk-free rate in it, the rate goes in without the question being asked. While rates are near zero the subtraction is negligible and the mismatch costs nothing. When they are not, the same strategy changes its apparent quality by the size of the rate, which has nothing to do with the strategy.
Sometimes the honest answer is that no published index resembles what you do. That happens, and the response is not to pick the nearest one and hope.
Two things work better. Construct the naive version of your own strategy, the one anybody could run without your research: same asset, same direction, no filters, no timing. Beating your own naive version is a real claim and a hard one. Or report against the specific alternative the investor in front of you actually has, and name it, which turns a vague comparison into a concrete one.
Either is more useful than a benchmark chosen because it is the convention in a category you happen to have been filed under.
When an argument about benchmarks goes in circles, one question ends it:
If this manager had never existed, what would this money have been doing?
That is the benchmark. Not the index your category quotes, not zero by default, not the risk-free rate because the formula has a slot for it. Whatever the money would actually have been doing.
Sometimes the answer is embarrassing for the manager, and a manager who is comfortable when it is embarrassing has told you something worth knowing.