A track record that starts at onboarding is not a track record

Most portfolio tools start counting on the day they are connected. For a fund that has been trading on-chain for a year or two, that date is an accident of procurement, not an event in the fund's life. Everything before it is a gap. The usual ways of closing it (a spreadsheet, a few month-end balances, an estimate) produce a series that looks continuous and isn't.

The problem is usually described as a shorter history. It is worse than that. A record that starts at onboarding, or is patched together before it, reports a different return, loses the cost basis of anything bought before the start date, and contains net asset value (NAV) figures that nobody can test against the chain. Each one shows up eventually: in an LP letter, on a due diligence call, or in an audit sample.

Missing cash flows change the return

Take a hypothetical fund that launches on 1 Jan 2025 with $10.0M. It returns +8.0% in the first quarter. On 1 Apr 2025 a second LP subscribes $4.0M in USDC, which is deployed the same day. The fund then returns −5.0% in the second quarter and +4.0% in the third. A tracker is connected on 1 Oct 2025.

To give allocators a figure since inception, the team rebuilds the first nine months from quarter-end wallet balances. Three versions of that rebuild are common, and they disagree:

Only the first row is the fund's return. The second puts the $4.0M in the wrong sub-period: the second quarter's $740,000 loss was incurred on $14.8M of capital, but the calculation divides it by $10.8M, so the quarter looks worse than it was. The third treats a subscription as performance. Quarter-end balances cannot tell the two apart, because a wallet that grows by $4.0M looks the same whether the money was earned or deposited.

None of these errors shortens the record. Each produces a different one. A time-weighted return (TWR) is only as good as its list of external cash flows and the dates attached to them, and that list cannot be inferred from balances. It has to be read from the transactions. Which return to put in front of an LP is a separate question, covered in TWR, Modified Dietz and XIRR.

Cost basis does not survive the onboarding date

A tracker that starts on 1 Oct 2025 sees the positions the fund held that day, at that day's prices. It has no record of when they were bought or at what price, so it takes the onboarding price as the cost.

Say the fund bought 500,000 governance tokens at $4.00 in early 2025. They are worth $9.00 when the tracker is connected, and the fund sells them at $7.00 in November. The tracker books a realized loss of $1.0M. The fund realized a gain of $1.5M.

Measured from 1 Oct, the position lost $1.0M under either view. What differs is everything that depends on the entry: the realized figure, the position's lifetime PnL, its holding period, and the attribution of PnL to the strategy that bought it. For perpetual futures the same applies to fills and funding. The position on the onboarding date says nothing about the entry price or the funding paid and received to hold it.

Gaps in daily NAV leave nothing to test

An auditor testing NAV picks a date and asks for the figure to be recomputed: which positions, which balances at which block, which prices from which source. If the series before onboarding is a set of month-end estimates, there is nothing to recompute. The answer to "how was the 31 May NAV derived" is a spreadsheet, and the auditor is left relying on the manager's own figures.

Gaps also change the risk statistics. Volatility, Sharpe, Sortino and maximum drawdown are all computed from the return series, and a monthly series cannot see inside the month. A fund that suffers a −18.0% drawdown in the middle of a month and recovers fully by month-end shows no drawdown at all in monthly data. On-chain books are marked continuously, by markets and by liquidation engines. A record that samples them once a month understates the risk the LPs carried.

What allocators expect

Allocators' due diligence questionnaires (DDQs) ask for monthly net returns since inception, the method used to calculate them, who computes NAV and how positions are priced, and an explanation of any break in the record or change in methodology. A record that begins at onboarding answers the first question with a shorter period and invites the last one: what happened before, and why is it missing? A record that is complete but switches from spreadsheet estimates to tracker output partway through invites it too, because the methodology changed on a date that has nothing to do with the strategy.

For managers who claim compliance with the GIPS standards the bar is explicit: at least five years of compliant performance, or performance since inception if the history is shorter. Since inception means since inception.

Reconstruction is not backfill

Backfill estimates. It starts from what is known (a few balances, the manager's records, today's positions) and fills the space between them with interpolation, assumptions or a second methodology. The result has the shape of a track record, but its figures can't be traced to anything.

Reconstruction replays. Every transaction from the fund's first on-chain interaction is read from the chain at the block where it happened and processed by the same engine, with the same pricing and the same rules, that computes today's NAV. Subscriptions are recorded at the block where they arrived. Positions carry their real entry prices. Every day has a valuation. There is no seam at the onboarding date, because the engine has no reason to know there was one.

Reconstructing by hand is possible and slow. The steps are the same with or without a tool:

1. Assemble the complete set of addresses since inception, including retired wallets, chains the fund no longer uses, and exchange or perp accounts.
2. Classify every transfer in and out as an external flow, an internal transfer between the fund's own addresses, income or an expense. Bridging between chains is internal. LP subscriptions are flows. Reward claims are income.
3. Value every position daily at a fixed time, such as 00:00 UTC, with a documented price source for each asset.
4. Reconcile the result to independent points: the administrator's capital statements, bank records of fiat subscriptions, LP capital accounts.
5. Record every correction as a separate entry with an author, a date and a reason, rather than overwriting the figure.

The second step is where most rebuilds go wrong, and it is the one balances cannot do for you. Reconstruction establishes what the book was worth. Whether it could have been turned into cash at that value is a separate question, covered in Marked at NAV is not liquid at NAV.

How Syncrone approaches it

Syncrone replays a fund's history from its first on-chain interaction through the same engine as today's data: entry dates, cost basis and daily valuations, with the block height, transaction hash and price source attached to every event. Raw NAV is preserved next to any adjusted figure, corrections are stored as records with an author, a timestamp and a reason, and finalized reports are locked. The record an allocator sees starts where the fund started.