The GP's IRR is the fund's, not yours
Fund-level return is computed from fund-level cashflows. Yours depends on when your calls were funded and what you actually received, and a late close or a secondary purchase moves it materially.
Performance
IRR, TVPI and DPI computed from your cashflows, not copied from the report
Your three sources disagree, and nobody has made an arithmetic error. They counted different cashflows, as of different dates. This computes yours from records you can open.
In use at a small number of family offices and investment managers
Why the numbers disagree
Spreadsheet IRRs are quietly fragile
XIRR over a hand-maintained cashflow column is only as good as the column. One missing fee call, one distribution booked to the wrong date, and the number is wrong in a way nothing flags.
Nobody agrees what counts as paid-in
Fees drawn outside commitment, recallable distributions, transaction costs on a secondary. Each source treats them differently, and each choice moves TVPI and DPI without anyone writing the choice down.
What you get instead
Metrics derived, never stored
IRR, TVPI, DPI and RVPI are computed from the cashflow records each time. Correct a call date and every metric that depends on it moves with it, so the numbers cannot drift away from their inputs.
The cashflows are visible behind the figure
Open any metric and see the dated cashflows that produced it, each linked to the statement it came from. When your number differs from the GP's, you can point at the reason instead of arguing about it.
Stated conventions
Whether fees outside commitment count as paid-in, and how recallable distributions are treated, are explicit settings rather than assumptions buried in a formula, so the same policy applies across every position.
Position, vintage and portfolio level
The same metrics per position, grouped by vintage year or manager, and consolidated across every entity in your reporting currency.
Where the data sits
Hosted in the EU
Your cashflows and the documents behind them stay on EU infrastructure, under a DPA that names every subprocessor.
Every input is inspectable
There is no black box between a statement and a metric. Each cashflow shows the document and page it was read from, and every correction is recorded next to the original.
Export in full
Cashflows, positions and computed metrics export whenever you want them, in a form your own model can read.
Questions we get about performance
Which IRR do you compute?
A since-inception, money-weighted IRR over your dated cashflows including the current residual value, which is the one the private markets convention means. Interim and annualised cuts are available, and each says which it is.
Why does your IRR differ from the GP's?
Almost always because of which cashflows were counted and as of when. Your funding dates, fees drawn outside commitment, a late close, or a secondary purchase price. The cashflow list behind the figure is there so you can find the difference in a minute rather than a morning.
Can we compare funds fairly?
Within your own portfolio, yes: group by vintage year, because a 2019 fund against a 2022 fund is not a comparison. Third-party benchmark data is not in the product today.
What is TVPI if a fund is still calling capital?
Distributions plus residual value over paid-in, which early in a fund's life is mostly residual value and therefore mostly the manager's estimate. RVPI is shown next to it for exactly that reason.
Can we override a figure we know is wrong?
You can correct the underlying cashflow, and the correction is recorded beside the original with who made it and when. You cannot type over a computed metric, because a metric that can be typed over is not evidence of anything.
Compute your own numbers
If your three sources disagree, the cashflows will say why. Tell us about your portfolio and we will show you on it.