If we ask you to judge us by one number, you are entitled to the recipe for it.
NAV per unit is that number. It is the price of what you own and the only honest scoreboard we keep. Most of the industry treats the method behind it as a black box - "proprietary methodology," a phrase that means don't ask. We think the opposite, because the most important number a fund publishes should be the easiest one to explain, not the hardest.
This is the belief this piece is here to break: that NAV is a number the manager decides.
It isn't. NAV is a number the method decides - and the method is the same whether the month was good or bad, whether you're watching or not.
Where a NAV can be faked
NAV per unit is the most trusted number a fund publishes and the least interrogated. People will argue about a single fee line for an hour and then accept a unit price at a glance. That asymmetry is exactly what makes it the easiest number to bend.
And there aren't many places to bend it. A NAV can only be massaged in three spots: the price you assign to each position, the moment you choose to compute it, and the entries you choose to count. Move any one of them quietly and the number flatters you - without a single outright lie being told.
So the only honest response is to expose the whole machine, then tie our own hands at each of the three. Here is the machine.
The whole formula
There is no second formula for the brochure. This is it:
Read it left to right. Take the pool's cash. Add up every open position - each one's quantity q times its current mark m. That total is what the pool is worth. Divide by U, the units outstanding. The result is what one unit is worth right now.
The equation is the easy part. The trust is in the three places it could be gamed - the marks, the timing, the inclusions - so we'll take them one at a time.
How each position is marked
The whole formula lives or dies on the marks m. A mark is just the price we say a position is worth right now - and it is the single easiest place in finance to lie without appearing to.
So we tie our hands. Every open position is marked at a real, observable price - never a model's opinion of one. Spot holdings are marked at live market price. Options are marked off live spot and live implied volatility, the same inputs the market itself uses to quote them. And the marking source and timestamp are recorded with the mark - so any single point in the NAV history can be traced back to where its number came from and the second it was taken.
There is an honest limit here, and we'd rather state it than have you find it: an option's mark and the cash you could actually realise by unwinding it at size are not the same thing. A mark is a fair midpoint; a forced exit pays a spread. That gap is exactly why hedge value can move against the book in calm markets, and it's why redeemable value is fixed at a monthly window rather than promised continuously - the mark is honest, but liquidity isn't free.
How often - and reconciled against what
A correct formula computed once a quarter is still a stale number. Ours refreshes on a fixed cadence, not on convenience and never on demand. Marks update on a regular tick; NAV per unit is recomputed from those marks; and the whole thing is written to a history that only ever appends.
One distinction sits underneath all of this, and it is the one that matters most:
We do not write NAV into the ledger. NAV is derived from the ledger. The ledger is the source; NAV is the consequence.
That single fact is what makes the number hard to fake. You can't reach in and type a flattering NAV, because nobody types NAV at all - it falls out of what the ledger already says we hold. To move the output you'd have to move the source, and the source only ever appends.
And it has to add up. It's one thing to compute a number; it's another to prove it. So an automatic invariant runs every hour and asserts that the pool's treasury - the actual ledger of what we hold - equals the NAV we publish, within a tight tolerance. If the two ever disagree beyond that tolerance, the number is treated as suspect until the discrepancy is explained - not quietly smoothed over. The published number is not the one we like. It's the one that survives the reconciliation.
What counts - and what is kept out by construction
Half of an honest NAV is what you exclude. Four rules govern the numerator, and none of them bend for a flattering month.
Realised profit stays in the pool. When the book earns, the gain isn't paid out as a dividend that quietly resets the scoreboard - it lifts NAV per unit. One unit simply becomes worth more. That's what makes the unit price a true running record of performance rather than a snapshot between payouts.
**Deposits and redemptions move cash and units together.** When you deposit, cash enters the numerator and new units enter the denominator in the same proportion as the current price - so the ratio doesn't budge. Your arrival can't dilute an existing holder, and an existing holder's gains can't be handed to you for free. This is the entire reason balance and NAV per unit are different animals: balance reacts to your cash flows, NAV per unit reacts only to performance.
A worked example, in full:
Deposit \$10,000 at that price and you receive about 9,804 new units; cash becomes \$112,000, units about 109,804, and ν is still \$1.02. The pool got bigger. The unit price didn't twitch. That is the design working.
Phantom and synthetic entries are excluded - on purpose. Internal accounting sometimes needs placeholder or synthetic legs to balance a model. Those are not profit and they are not loss, so they have no business in a NAV that claims to be real. They are filtered out by construction.
We learned this one the hard way.
A single synthetic bootstrap leg, left sitting in the numerator, will inflate NAV per unit until you find it and take it out. So we find it, and we take it out - and the exclusion rule exists so the next one never gets counted in the first place.
Fees are taken at the line, not hidden in the mark. The number you read is after the haircut, not before it. A NAV computed gross-of-fees is a menu price; you live on the net one - the same gap that separates quoted yield from kept alpha.
The limits
Every honest number has an edge where it stops being perfect, and pretending otherwise is its own kind of lie. Ours has three:
- Liquidity. Options can't be unwound continuously at size, so redeemable value is fixed at the monthly window. The mark is fair; the exit isn't frictionless.
- Basis. SOL-denominated capital is hedged with the deepest options market available, which isn't always the same asset - a known, stated weakness, not a buried one.
- Audit. This is internal tooling with hourly self-reconciliation and an append-only history. It is not yet an external audit, and we will not call it one until it is.
We write these down because a number you can verify is worth more than a number you're asked to admire - and verifiability is the only thing that separates a real record from a beautiful fiction.
The day a number becomes hard to explain is the day to distrust it. This one is easy to explain on purpose.
The one line to keep
Strip everything else away and one principle remains: a fair NAV is one the manager cannot influence. Not one you're asked to trust - one built so that trust isn't required. The formula, the marks, the ledger, the hourly reconciliation - every piece exists to take the number out of our hands and leave it in the ledger's.
The fairest number in finance is not the one you trust. It's the one nobody is allowed to edit.