TVPI, or total value to paid-in capital, is the multiple a venture fund has returned to its LPs plus the value it still holds, divided by the capital those LPs have actually paid in. It's the headline number on almost every fund report, and the one most likely to be misread, because half is cash and half is opinion.
The anchoring arithmetic is short. TVPI equals DPI plus RVPI. DPI is money already sent back. RVPI is what's still in the portfolio, at whatever value the manager says it's worth. Most explainers define the terms and stop. The useful part is the gap between the half you can bank and the half you have to believe.
TVPI, DPI, RVPI, MOIC and IRR, defined precisely
Five numbers do almost all the work in fund reporting, and mixing them up is the fastest way to lose an LP's confidence. Start with the denominator three of them share. Paid-in capital is what LPs have wired, not what they committed. On a $10MM fund that has called 20%, paid-in is $2MM, so multiples run against a small base and one early markup swings them hard.
The three multiples
DPI, distributions to paid-in, is cumulative cash and stock distributed to LPs over paid-in capital. It's realized, and verifiable against bank records.
RVPI, residual value to paid-in, is the fund's net asset value over paid-in capital. It's unrealized, the manager's estimate of what the remaining positions are worth.
TVPI is the sum. If a fund reports 1.55x TVPI with 0.25x DPI, RVPI is 1.30x and 84% of the reported value hasn't happened yet.
MOIC and IRR measure something else
MOIC, multiple on invested capital, is usually computed at the deal or portfolio level: total value of the investments over the dollars put into companies. It excludes fees and fund expenses, so it's almost always higher than TVPI on the same fund. A manager quoting MOIC when an LP asked for TVPI is quoting a bigger number for a different question.
IRR is the annualized rate that sets the fund's cash flows to zero, so it's sensitive to timing in a way no multiple is. Two funds can both return exactly 2.0x and report wildly different IRRs. Return 2.0x over ten years and the IRR is about 7.2%. Over seven years, about 10.4%. Over five, about 14.9%. Same multiple, three stories, only the calendar changed.
TVPI is a claim. DPI is a fact.
DPI is settled. The wire either went out or it didn't, and an LP can check it against their own account. No interpretation is available.
RVPI is not settled. It's a valuation, produced under a policy, applied to illiquid positions with no market price. The structural issue, stated plainly and without accusing anyone: the manager who sets the marks is the manager judged on them. The person valuing the portfolio is the person whose next fund depends on that estimate looking good.
That isn't a claim about dishonesty. Most managers apply their policy carefully and most valuations are defensible. It's a claim about incentives. When a hundred judgment calls get made by someone with a consistent directional interest, the aggregate leans, one reasonable decision at a time. Whether to mark up on a small internal round. Whether a SAFE at a higher cap counts as a priced event. Whether a company that hasn't raised in 30 months is still worth its last round.
So an LP seeing 2.5x TVPI and 0.0x DPI isn't looking at a 2.5x fund. They're looking at a manager's opinion that it might one day be one, and that opinion converts to cash only through exits the manager doesn't control.
The consequence for an emerging manager is counterintuitive. Leading with a high unrealized TVPI reads as naive, because every experienced LP discounts unrealized marks and the size of that discount depends on how you present them. "TVPI is 1.9x, DPI is zero, here's our valuation policy and the three positions carrying most of that value" earns more credit than "we're at 1.9x." The second asks to be taken at its word. The first shows the work. That's the instinct behind LP due diligence, where the manager who volunteers the weak spot gets believed on everything else.
The J-curve: what TVPI and DPI look like at years 1, 3, 5 and 10
Fund metrics move in a predictable shape. Knowing it is how you read a number in context instead of in isolation.
Year 1
TVPI is below 1.0x, usually around 0.8x to 0.95x, and it should be. Fees have been drawn, a few positions sit at cost, nothing has marked up. DPI is zero. A Fund I at the bottom of the J-curve isn't underperforming. It's new. A manager reporting above 1.0x in year one is either lucky or marked something up on thin evidence.
Year 3
The first real markups arrive as two or three companies raise priced rounds led by outside investors. TVPI crosses 1.0x and can climb fast, because the paid-in base is still small. DPI is almost always still zero. This is maximum divergence between what the fund looks like and what it has proven, and it's when managers raise Fund II, which is why LPs discount year-three TVPI hardest.
Year 5
Capital is fully called or close to it. Write-offs have shown up, because failures resolve faster than successes. A first exit may have produced a small DPI. TVPI now runs against the full denominator, so it moves slowly. This is the earliest point at which it carries real information.
Year 10
DPI converges toward TVPI as positions liquidate, and the two meet at the end of the fund's life. Whatever gap remains never converted. A fund that reported 2.4x TVPI at year seven and finishes at 1.9x DPI didn't lie. It marked optimistically and the market disagreed. That happens constantly, which is why DPI carries the weight.
A worked example: one fund's TVPI and DPI across ten years
Take a $10MM Fund I, inside the band where roughly 90% of emerging manager commitments go, funds under $15MM. At our published $159K average LP check, that's about 63 LPs. The fee is 2% of committed capital for ten years, $200K annually and $2.0MM total, leaving $8.0MM investable. The strategy is 25 initial checks of $200K with $3.0MM for follow-ons.
Year 1
Called capital is $2.0MM, or 20%. Of that, $200K went to fees and $1.6MM into eight companies at $200K each, with $200K in cash. Everything sits at cost, so NAV is $1.8MM. TVPI is 0.90x, DPI 0.00x, RVPI 0.90x. The fund is worth less than the money in it, which is correct.
Year 3
Called capital is $6.0MM, fees to date $600K, and $5.2MM is at work: 20 initial checks at $200K plus $1.2MM of follow-ons into four companies at $300K each. Cash is $200K.
Those four raised priced up rounds and carry a combined mark of $4.0MM against a $2.0MM cost basis. Two companies are dead, writing off $400K. The other 14 sit at cost, $2.8MM. NAV is $7.0MM. TVPI is 1.17x, DPI 0.00x, RVPI 1.17x. Every dollar of that 1.17x is unrealized, and four positions carry it.
Year 5
Capital is fully called at $10.0MM. Total invested across 25 companies is $8.0MM, fees paid are $1.0MM, and $1.0MM of cash is reserved for remaining fees. One company was acquired, returning $2.5MM on a $500K cost basis. Eight are written off, costing $1.8MM. Sixteen positions remain with a $5.7MM cost basis, marked at $12.0MM.
DPI is 0.25x. NAV is $13.0MM, so RVPI is 1.30x and TVPI is 1.55x. For the first time the fund has a fact in it, and the fact is 0.25x.
Year 7
Cumulative distributions reach $6.2MM after a secondary sale, and NAV is $14.0MM. DPI is 0.62x, RVPI 1.40x, TVPI 2.02x. This is the number a manager puts on a Fund III deck, and the one an LP discounts hardest.
Year 10
Cumulative distributions total $17.0MM. Two positions remain, carried at $3.0MM. DPI is 1.70x, RVPI 0.30x, TVPI 2.00x.
Now look back at the year-seven marks. That $14.0MM of residual value produced $10.8MM in distributions plus $3.0MM still held, so $13.8MM against a $14.0MM estimate. Close to right. Run the alternative: had those marks been 25% optimistic, the honest carrying value was $11.2MM and year-seven TVPI would have read 1.74x rather than 2.02x. DPI was 0.62x in both worlds.
The arc: TVPI runs 0.90x, 1.17x, 1.55x, 2.02x, 2.00x. DPI runs 0.00x, 0.00x, 0.25x, 0.62x, 1.70x. Seven years of divergence, then convergence. That shape is the fund, and producing a good version of it is most of what portfolio construction is trying to do.
Gross, net, fee load and the ways TVPI gets flattered
Gross versus net
Two funds can hold identical assets and report different numbers. Run the fund above three ways. Deal-level gross MOIC is $20.0MM of total value against $8.0MM invested in companies, or 2.50x. Net of fees, it's $20.0MM against $10.0MM paid in, or 2.00x TVPI. Net of fees and a 20% carry over return of capital, LPs receive $18.0MM against $10.0MM, or 1.80x.
Same portfolio. 2.50x, 2.00x, 1.80x. A manager quoting the first without saying which it is isn't lying, but it sits 0.70x above what the LP takes home. State the basis every time.
What the fee load actually costs
On a $10MM fund charging 2% flat for ten years, $2.0MM never reaches a company, so investable capital is 80 cents on the committed dollar. Step the fee down to 1.5% after the five-year investment period and fees fall to $1.75MM, leaving 82.5 cents. Flip it: to deliver $20.0MM of total value the flat-fee fund needs 2.50x on invested capital, the step-down fund 2.42x. That's 0.08x of required gross performance, free, from a term you negotiate once. How these compound is covered in our piece on venture fund economics.
Why IRR flatters early distributions
IRR rewards speed. A fund returning $1.0MM in year two off $3.0MM called can post a striking IRR while DPI sits at 0.33x and the fund has proven very little. Small early wins move IRR far more than they move any multiple.
Subscription credit lines push the same button. If a fund draws on a facility to make an investment and calls LP capital nine months later, that capital was outstanding for nine fewer months, so IRR rises. TVPI, DPI and MOIC don't move, because multiples don't care when the money arrived. Subscription lines are a legitimate way to smooth capital calls. They're also the easiest way to make an IRR look better without earning it, which is why experienced LPs ask whether IRR runs from the facility draw or the LP call date.
Valuation policy changes the number without changing the assets
Return to year five. Sixteen positions, $5.7MM cost basis. Marked to the most recent round they carry at $12.0MM, NAV is $13.0MM, TVPI is 1.55x. Under a policy that holds at cost until a third party leads a priced round and caps markups at the last institutional price, the same positions might carry at $8.5MM, NAV is $9.5MM, and TVPI is 1.20x.
Identical companies, identical cash, a 0.35x difference in reported TVPI, produced entirely by policy. Comparing two funds' TVPI without knowing both policies is close to meaningless, and stating yours in every report is worth more than the extra 0.35x would have been.
What to report to LPs, and how often
Consistency beats flattery. Reporting the same metrics the same way every quarter for ten years buys what a good number can't.
The quarterly package
Report committed capital, called capital, paid-in capital, cumulative distributions and current NAV, then TVPI, DPI, RVPI and net IRR, each labeled net of fees and stated as of a specific date. Add a schedule of investments listing every position at cost and at fair value, with the date and nature of the last valuation event. Restate the valuation policy every time, rather than filing it once. Explain any material mark change in writing. If one position moved 40% of your NAV, say which one and why. LPs find out anyway, and whoever says it first gets believed later.
Cadence
Quarterly unaudited reports within 45 to 60 days of quarter end, plus an annual audited financial statement, is the standard emerging managers should hold themselves to. Over a ten-year fund that's 40 quarterly reports and 10 audits. By the time you raise Fund III you have a decade of consistent reporting a diligence process can read, which is a different asset from a good number.
Most managers underestimate the operating load. Capital account statements, allocations, waterfall math and audit support are real work, and it's what slips first when a solo GP is also sourcing. 61% of VC Lab managers are solo GPs, so that's the common case. Running it on infrastructure like Decile Hub, with administration through Decile Partners, is how more than 1,000 firms keep it from slipping.
What benchmarks are and aren't for
Benchmarks are useful for one thing: knowing whether your vintage year is generally kind or generally brutal, so you can have a calibrated conversation instead of an anxious one. They're much less useful for grading a small fund.
Three reasons. Vintage, stage, geography and strategy all move outcomes, and a sub-$25MM specialist fund investing pre-seed has few true comparables in any vintage. That profile is now the norm: generalist funds fell from 22% of new funds in 2020 to 5% in Q1 2026 while specialists rose from 78% to 95%, and 85% of our managers invest at pre-seed or seed. Second, quartiles for young vintages are built substantially from unrealized marks, so comparing your unrealized TVPI to a benchmark's compares one set of claims to another. Third, funds that report aren't a random sample of funds that exist. Use benchmarks for calibration, not as a grade, and don't quote one unless you can source it precisely.
Frequently asked questions
What is a good TVPI for a venture fund?
There's no single answer, and a number quoted without the fund's age, DPI and valuation policy is close to useless. A 1.5x TVPI at year two and a 1.5x TVPI at year nine describe different funds. Ask three questions instead: how much of that TVPI is DPI, how old is the fund, and what policy produced the marks. A 1.6x TVPI with 1.1x DPI at year eight is stronger than a 2.4x TVPI with zero DPI at year four.
What's the difference between TVPI and DPI?
TVPI equals DPI plus RVPI. DPI is money actually distributed to LPs over paid-in capital, and it's verifiable. RVPI is the manager's valuation of what's still held over paid-in capital, and it's an estimate. TVPI is the realized part plus the unrealized part. When a fund reports 1.55x TVPI and 0.25x DPI, roughly 84% of the reported value is unrealized and rests on marks no buyer has tested.
Is TVPI the same as MOIC?
No, though they get used interchangeably and shouldn't be. TVPI is measured against paid-in capital, which includes management fees and fund expenses. MOIC is typically measured against capital invested in companies, which excludes those costs. On a $10MM fund with $2.0MM of lifetime fees, $20MM of total value is a 2.50x MOIC and a 2.00x TVPI at the same moment. Ask which denominator a quoted multiple uses.
Is TVPI reported gross or net of fees and carry?
It should be net of fees, since paid-in capital includes the fees LPs funded, but it's frequently reported gross of carry. Those are different numbers. A fund at 2.00x TVPI net of fees, carrying 20% over return of capital, delivers about 1.80x to LPs. State the basis on every report. An LP who finds out later that your headline was gross will re-examine everything else.
Can TVPI go down?
Yes, routinely. TVPI falls when positions get written down or written off, when a down round resets a carrying value, or when fees are drawn against a portfolio that hasn't appreciated. It also falls mechanically as capital is called, since the denominator grows before the new investments mark up. A TVPI that only ever rises across a decade is unusual enough that a careful LP will ask about the policy behind it.
Report the number you can defend
TVPI is the right headline metric and a bad standalone one. It's the sum of something true and something claimed, and the skill is being clear about which is which. Show DPI next to it. State your valuation policy. Say whether the number is net of carry.
The market rewards this now. February through May 2026 ranked among the top five fundraising months in four years, each running 1.2x to 2.2x the same month in 2025, across 1,000+ PACTs, 1,000+ LPAs and 900+ funds, detailed in our emerging manager performance data and generalist versus specialist analysis. But you're competing for a $159K average check against managers whose unrealized numbers look like yours, where checks in the $150K to $250K band convert to signed LPAs at 1.2x to 2.4x the rate of other bands, per our first fund fundraising data. Your TVPI won't differentiate you. Knowing what it's made of will. And a $10MM fund needs only $20MM of total value to return 2.0x, which one company can produce, part of why small funds win big and why the power law works differently here.
If you're on Fund II or beyond and want to sharpen how you report, benchmark and defend performance, Emerging Institute is built for established managers working through exactly this. First-time managers should start with VC Lab, the free accelerator that runs from fund formation to first close. LPs who want to read these numbers the way an allocator does can look at LP Institute. The data sits in our VC research hub.
Apply to the next cohort, and start reporting a number you can stand behind in year ten.