Every number we publish has an edge it cannot see past.
That isn't a flaw we're confessing - it's the nature of measurement itself. A metric is a lens: sharp on the one thing it's pointed at, blind to everything beside it. NAV tells you what the book is worth and nothing about how it got there. An attribution tells you where the profit came from and nothing about whether it was luck. Every instrument has a lit circle and a dark surround - and the dark part never shows up on the dashboard, which is exactly what makes it dangerous.
The belief this piece is here to break: that a rich dashboard means the risk is covered. It's backwards. A dashboard is a set of decisions about what to look at, which is the same as a set of decisions about what to ignore. The more complete it looks, the easier it is to forget the dark surround entirely. So here is the uncomfortable inventory - what each of our main numbers sees, and the part almost no one publishes: what it doesn't.
What NAV per unit can't see
NAV per unit is the most honest single number we keep, and it is still a photograph - a value at an instant. A photograph can't show motion.
So it cannot see the path. A unit worth 1.10 looks identical whether it climbed there in a calm line or whipsawed through a 30% pit and clawed back. Same number, completely different experience - and the path is the part you actually have to live through. NAV also can't see the tail: it reports what the book is worth, never what it might be worth after a shock that hasn't arrived. And it can't see liquidity at size - the mark is a fair midpoint, not the price you'd get unwinding everything at once. NAV answers "what is it worth now," honestly. It is silent on "how rough was the ride" and "what could go wrong next," and that silence is not the same as a calm answer.
What attribution can't see
An attribution tells you where realised profit came from - which engine, which leg, which book. Useful, and still half-blind.
It cannot reliably separate luck from skill. A good number can be a lucky one, and over a short record the two are genuinely indistinguishable - which is the whole reason a clean track record proves so little. It also can't see the dog that didn't bark: a loss the hedge quietly prevented never appears as a positive line, because nothing happened, and "nothing happened" doesn't get a row in a P&L. Some of the most valuable work a defensive book does is precisely the kind that leaves no trace in an attribution.
What Liquidity Weather can't see
Liquidity Weather reads current market conditions as a single number from real inputs. It is a thermometer, not a crystal ball.
It cannot see tomorrow. It's a reading of now, not a forecast, and we've said plainly that we don't predict markets. It is blind to the regime it has never seen - any gauge built from history is calibrated on the past and can be surprised by a genuinely new state. And it can be outrun: a shock that arrives faster than the gauge updates is, for a moment, invisible to it. The weather tells you what the sky looks like right now. It does not promise what lands in an hour.
The blind spot they all share
Read the three together and the common edge is unmistakable: every one of them measures the past or the present. None of them measures the future.
NAV, attribution, the weather, drawdown, volatility - all of them describe what already happened or what is true this instant. But risk does not live in the past tense. Risk is a property of paths not yet taken - of the shock that hasn't hit, the correlation that hasn't snapped, the tail that hasn't arrived. The thing that can actually end you is, almost by definition, the thing no number on the page can show, because it hasn't happened yet. This is the seam where measurement ends and judgment begins - and it's the reason a wiggle you can measure (volatility) is so often mistaken for the danger you can't (ruin). That mistake deserves its own piece, and it has one coming.
Why we publish the blind spots
A measurement system that hides its limits is an invitation to over-trust it. The dial that looks authoritative is the one most likely to be believed past the point it can support. Drawing the edges is what makes the lit part safe to use.
So the honest dashboard isn't the one with the most dials. It's the one that tells you where the dials stop - which numbers are photographs, which are thermometers, and which question none of them can answer. We'd rather hand you a smaller set of numbers you know the boundaries of than a richer set you'd over-read.
What you can check
In the spirit of how you can verify us: of any number we ever show you, ask the one question this whole piece is built on - what does this one not see? We should always have an answer. The day a metric is handed to you as complete, with no stated blind spot, is the day to trust it least. A number that admits its edges is doing its job. A number that claims to see everything has stopped being a measurement and started being a sales pitch.
A number tells you what it was built to tell you. Its silence about everything else is not the same as everything else being fine.
The one line to keep
The most honest thing a measurement can do is show you where it stops seeing.
Every dashboard has a dark surround. The only question is whether it's drawn on the map - or left off it.