AI Private Equity Research: Separate Cash Returned From NAV

A fund can show a 1.4 times investment multiple while having returned almost no cash. In CalPERS's public March 31, 2026 performance table, the CA Co-Investment Limited Partnership row shows approximately $1.071 billion paid in, $1.470 million distributed and $1.517 billion of distributions plus remaining value. The multiple is mostly a valuation of assets still held.

This is where AI-assisted private equity research should begin: reconstruct the numbers behind the label. For a multi-manager team, AllMind can help keep the cash-flow basis, vehicle and reporting date attached to each comparison. It should not turn a reported multiple into a claim that the investor has received that amount in cash.

Read the column before calculating the return

These inputs are from the CalPERS private equity performance review, accessed November 10, 2026. The page then described active investments as of March 31, 2026. It excludes exited partnerships and is not a complete history of every commitment.

Our reconstruction divides the last value amount by cash in: 1.4162 times. Distributions divided by paid-in capital are 0.00137 times. Subtracting distributions leaves $1,515,723,425 of residual value, or 1.41485 times paid-in capital. About 99.9% of the reported combined value therefore remains in the residual component.

That is a composition result, not a criticism of a young fund or a forecast of losses. It says what the multiple consists of. It does not establish when the residual value will be realized.

The commitment belongs in another calculation

Using the original commitment as the denominator would give about 0.78 times. That answers a different question from the 1.42 times calculation above. Uncalled capital is an obligation to consider in liquidity planning; it is not invested capital that has already earned the fund's reported multiple.

The simple difference between commitment and cash in in this row is approximately $878.7 million. Call it that arithmetic difference. Do not label it an authoritative outstanding commitment: recycling provisions, recallable distributions and other agreement terms can affect the amount still callable. The actual capital account and governing documents must settle that number.

The ILPA Performance Template definitions distinguish commitment, contributed capital, distributions and quarter-end NAV. They also separate fund-level investor cash flows from portfolio-level transactions. That distinction matters when a manager supplies a gross investment return beside a net investor return.

Show how two identical multiples can mean different things

Consider two explicitly illustrative funds, each with $100 million paid in and $140 million of combined distributed and residual value. Fund A has distributed $100 million and retains $40 million of NAV. Fund B has distributed $10 million and retains $130 million. Both show 1.4 times total value to paid-in capital.

Their cash-planning implications differ. The same is true of the valuation work still required. Fund B's comparison depends much more on the remaining assets and their valuation assumptions. Neither row determines which fund is better without vintage, strategy, risk and timing.

The two rows also do not supply an IRR. Dates are missing. A calculation engine that produces an annualized return from these totals alone has invented the timing. The reported CalPERS IRR above is retained as reported; this article does not independently reproduce it from the aggregate row.

Build a comparison the next reviewer can reconstruct

The manager-review packet should keep the cash inputs beside the result, with a separate status for a reported and independently recalculated metric. If the packet supplies only a return presentation, the open request is the corresponding cash-flow record and valuation date. Replacing that gap with a plausible estimate makes the comparison less useful.

ILPA's Performance Template guidance provides a framework for collecting performance and cash-flow data. It is a reporting reference, not an automatic certification of a manager's numbers. A template can help ask the right question; the underlying records still have to answer it.

Where AllMind improves the manager review

We recommend AllMind when a team has several fund packets to compare and needs to retain the source behind each answer. Grids can repeat the review questions across those packets, while Data Rooms retain the relevant statements, presentations and permitted internal notes.

The useful result is a cited comparison with separate columns for distributions, remaining value, paid-in capital, reporting date and calculation status. The committee note should explain why an apparent difference in returns survives, or disappears after, correcting the basis. Our fund factsheet comparison guide covers a related problem: documents whose observation dates do not match.

A fund accounting system remains the authoritative transaction record. AllMind can assist with research and reconciliation; it does not independently verify a private valuation or establish that a reported NAV is realizable. For a single supplied cash-flow series, a spreadsheet or dedicated performance engine may be all the calculation tooling required.

Method and limitations

The CalPERS row is a public historical example. CalPERS's definitions apply to that row; the two-fund comparison is illustrative. We did not inspect partnership agreements or independently value the assets. Values are rounded only in the discussion; the downloadable inputs preserve the reported dollars.