J-Curve and Commitment-Pacing Model

Set the scenario

The defaults reproduce the published venture base case described under Method and sources. Change any input and every result below recomputes in your browser, with nothing sent anywhere. Each result reads per 100 committed, so the page uses no currency.

Fund assumptions

Contribution rate, year 1 (% of commitment)

Share of the commitment called in year 1. Allowed: 1 to 100 percent.

Contribution rate, year 2 (% of unfunded commitment)

Share of the commitment still uncalled that is called in year 2. Allowed: 1 to 100 percent.

Contribution rate, year 3 onward (% of unfunded commitment)

Share of the commitment still uncalled that is called in each later year. Allowed: 1 to 100 percent.

Bow (dimensionless)

A higher bow pushes distributions later in the fund life. Allowed: 0.5 to 6.

Annual growth (%)

A single growth assumption; the model returns it as its internal rate of return. Allowed: -30 to 50 percent.

Yield floor (%)

The lowest share of net asset value distributed in any year before the final year. Allowed: 0 to 30 percent.

Fund life (whole years)

The model runs for this many annual steps; everything is distributed in the last one. Allowed: 5 to 20.

Reset to published case

Editing needs JavaScript. The results below show the published case.

Results for a single fund

At these inputs, cumulative net cash flow reaches its low of −79.3 per 100 committed in year 4, first returns to zero or above in year 8, and total distributions are 1.88 times the amount called.

The first chart plots cumulative net cash flow, which is the J. Contributions pull the line below zero in the early years, and distributions later lift it back above. The line plots cash and not reported value. The model charges no fees and holds no investment at cost, so it does not draw the reported-value J-curve in which TVPI falls below one.

Year-by-year figures

The table holds the same numbers as both charts. DPI is cumulative distributions divided by cumulative contributions, RVPI is net asset value divided by cumulative contributions, and TVPI is their sum. Each contribution is a capital call against the commitment.

Pace commitments across years

A program that commits to the same fund every year does not hold the exposure of one fund. Early vintages are still calling capital while older ones distribute, so annual program cash flow and cumulative program cash flow turn positive in different years. The two inputs below set how many annual vintages the program runs and how much net asset value it aims to hold once fully invested. The table commits 100 in each year, and the sentence beneath it converts the target into an annual commitment.

Program assumptions

Annual vintages (whole years)

How many consecutive years the same fund is committed to. Allowed: 1 to 25.

Target net asset value (index)

The exposure the program aims to hold once it is fully invested, as an index with no currency. Allowed: 1 to 10000.

When the same fund is committed every year, annual program cash flow first turns positive in year 8, but cumulative program cash flow does not return to zero until year 13. Once fully invested, each 1 of annual commitment supports a net asset value of 6.78, so holding 100 of net asset value implies a commitment of 14.76 each year.

Program figures by year

How the model works

The model steps through the life of one fund in annual increments and assumes every flow occurs at year end. A new commitment starts with no net asset value. Four relationships determine each year, where CC is the commitment, RC the contribution rate, Y the yield floor, L the fund life, B the bow and G the growth rate.

1. Contribution. C(t) = RC(t) × (CC − PIC(t)), where PIC(t) is the sum of the contributions made before year t. Each year the fund calls a set share of what is still uncalled.
2. Rate of distribution. RD(t) = max(Y, (t / L)^B), and RD(t) = 1 once t reaches L. A higher bow pushes distributions later in the life of the fund.
3. Distribution. D(t) = RD(t) × NAV(t−1) × (1 + G). The fund distributes a share of last year’s net asset value after that value has grown.
4. Net asset value. NAV(t) = NAV(t−1) × (1 + G) + C(t) − D(t). A contribution made in year t earns no growth that year and cannot be distributed before year t + 1.

In the last year of the fund life the rate of distribution is one, so the whole grown net asset value is paid out. Contributions made in the final years leave a small residual, which the table reports instead of hiding. The model returns one average path. It says nothing about the spread of outcomes between funds. For the investment philosophy behind the way LeverVenture thinks about the asset class, read the growth equity thesis.

These limits follow from the design and apply to every result on this page.

• The model is deterministic. It is a planning aid and not a forecast.
• The growth input is the model’s internal rate of return by construction. It is an assumption you state and not a result the model finds.
• Growth is a single assumption. The model does not separate fees or carried interest from it.
• TVPI is at or above one in every year whenever growth is at or above zero, because net asset value compounds from the first contribution. The cash J described above is the cumulative net line.
• Steps are annual, and the model has no quarterly timing.
• The pacing view assumes a constant annual commitment and no growth in the program’s target.

Method and sources

The method is the one Dean Takahashi and Seth Alexander published as “Illiquid Alternative Asset Fund Modeling” in The Journal of Portfolio Management 28(2), 2002, pages 90 to 100 (DOI 10.3905/jpm.2002.319836). The equations and default values on this page follow published descriptions of that model, which are listed below with the date each was last reviewed, and not the paper itself.

The ILPA Model LPA term sheet sets a fund term of ten years with two one-year extensions, which is the context for the 12-year default life. Wellington Management names fees and the time until liquidity or write-up as the two main determinants of the reported J-curve, and neither is modeled here.

Defaults and where each comes from

Sources

1. Dean Takahashi and Seth Alexander, "Illiquid Alternative Asset Fund Modeling," The Journal of Portfolio Management 28(2), 2002, pp. 90–100. The page applies this method. The equations and default values below come from the published descriptions that follow, not from the paper itself. Last reviewed October 16, 2026.
2. arXiv 2108.02853, "Supervised Neural Networks for Illiquid Alternative Asset Cash Flow Forecasting," section 2. Prints the six inputs and the contribution, distribution and net asset value equations. Last reviewed October 16, 2026.
3. SEI Novus, "Balancing Liquidity Constraints in a Private Investment Program," October 2022. States the distribution equation and shows the same three-step contribution pattern. Last reviewed October 16, 2026.
4. Altss, "Commitment Pacing," glossary entry. A commercial glossary and a secondary source. It states the full default set and attributes it to the published Yale base case for venture capital. Last reviewed October 16, 2026.
5. Institutional Limited Partners Association, Model Limited Partnership Agreement term sheet, October 2019. A fund term of ten years with two one-year extensions, which is the context for the 12-year default life. Last reviewed October 16, 2026.
6. Institutional Limited Partners Association, 2019 MCON, "Commitment & Cashflow Modeling: Planning for Uncertainty". Shows a bow of 2.1 fitted to historical buyout data, which sits inside the 0.5 to 6.0 range the page accepts. Last reviewed October 16, 2026.
7. Wellington Management, "Understanding private equity performance". Names fees and the time until liquidity or write-up as the two main determinants of the J-curve. Last reviewed October 16, 2026.

Frequently Asked Questions

Why does the model’s TVPI never fall below one?

Net asset value compounds from the first contribution, so with growth at or above zero the model’s total value always covers what was paid in. The cash J-curve is the cumulative net line in the first chart. Reported values dip in practice because of fees and because investments are held at cost, and this model includes neither.

Is the growth rate a forecast?

No. It is an input, and by construction it equals the internal rate of return the model reports. The default of 13 percent is the figure used in the published illustrative case. It is not a projection, a target or the return of any fund, and you should replace it with whatever assumption you want to test.

How does the pacing view size an annual commitment?

It adds the single-fund net asset value column to find the exposure that each unit of annual commitment supports once the program is fully invested, then divides the target net asset value by that figure. The result assumes the same commitment every year, the same fund behavior, and no growth in the target itself.