VC fund metrics: MOIC vs IRR, TVPI vs DPI, and how the six numbers fit together

Ask a general partner how the fund is doing and you will get a multiple. Ask the pension fund that backed it and you will get an IRR. Ask the endowment’s chief investment officer in a bad year and you will hear one word: distributions. They are all describing the same fund, and they are all right. Fund metrics are not competing truths; they are different camera angles on a single stream of cash going in and cash coming out.

This guide explains how the six numbers that dominate venture and private equity reporting fit together, why investors have shifted their weight from one to another over four decades, which of them matter to whom, and where each one quietly misleads. Each metric also has its own guide, with a calculator and dated benchmarks: MOIC, IRR, TVPI, DPI, RVPI and the J-curve.

The three questions behind every fund metric

Strip away the acronyms and every fund metric answers one of three questions.

How much? For every dollar that went in, how many dollars of value exist now? That is the job of the multiples, MOIC for a deal or a portfolio and TVPI for a fund.

How fast? Doubling your money in three years is a different achievement from doubling it in twelve. That is the job of IRR, which turns the timing of every cash flow into one annual rate.

How real? Some of that value is cash already in investors’ bank accounts; the rest is a valuation that may or may not survive contact with a buyer. DPI measures the first part and RVPI the second.

Oxford’s Ludovic Phalippou put the first two in one sentence in January 2026: “In private equity, there are two basic dimensions of performance: how fast money comes back (IRR) and how much money comes back (the multiple)” (Phalippou, 2026). The third dimension, real against paper, is the one that has dominated LP conversations since 2022.

The J-curve is the odd one out: it is not a number but a shape, the path the other metrics trace over a fund’s life. It explains why a perfectly good fund looks like a failure in its first three years.

The six numbers in one minute

MOIC, multiple on invested capital, is realized proceeds plus unrealized value divided by invested capital. It is the deal-maker’s number: invest $1M, end up with $3M, and you have a 3.0x. It says nothing about how long that took.

TVPI, total value to paid-in capital, is the same idea at fund level: distributions plus net asset value, divided by what the limited partners have paid in. It is usually reported net of fees and carry, which is why a fund’s TVPI sits below the gross MOIC of its portfolio.

DPI, distributions to paid-in capital, counts only cash returned. The GIPS standards give it its most important threshold in a single sentence: “Once the DPI is greater than one, the fund has broken even” (GIPS, 2005).

RVPI, residual value to paid-in capital, is the unrealized remainder. GIPS describes its life cycle precisely: “the RVPI will increase to a peak and then decrease as the fund matures and eventually liquidates to a residual market value of zero.”

IRR, the internal rate of return, is the annual discount rate that makes the present value of all the fund’s cash flows equal zero. Turn $1M into $2.5M in five years and the IRR is 20.1%; take ten years and it falls to 9.6%.

The J-curve is the shape of returns over time: negative while the fund calls capital, pays fees and holds new investments at cost, then rising as companies grow and are sold.

One identity ties four of them together, and it is worth memorising: TVPI = DPI + RVPI. The 2020 GIPS standards require firms to present all three, plus the paid-in capital multiple, for any pooled fund with committed capital. Every headline multiple is cash plus paper. The only question is the mix.

1. 01

   Capital goes in

   How deep is the early dip, and how long does it last?

   • J-curve
2. 02

   Value builds on paper

   What is still held, and at what mark?

   • RVPI
3. 03

   Cash comes back

   How much have investors actually received?

   • DPI
4. 04

   The whole picture

   How much value, cash and paper, per dollar paid in?

   • TVPI
   • MOIC
5. 05

   The clock

   How fast did the money work?

   • IRR

MOIC vs IRR: how much against how fast

The oldest argument in fund metrics is between the multiple and the rate. It matters because the two can tell opposite stories about the same investment.

The arithmetic linking them is simple for a single cheque and a single exit: IRR = MOIC^(1 ÷ years) − 1. A 2.5x in five years is a 20.1% IRR (our arithmetic), and 2.5x over a five-year hold is the classic buyout target in Bain’s words: “a target 2.5x multiple on invested capital (MOIC) over a five-year holding period” (Bain, 2026). Hold the same 2.5x for ten years and the IRR falls to 9.6%. Sell a winner after eighteen months for 1.4x and the IRR is about 25% (our arithmetic), while the fund has made little money.

That is why practitioners use both, and why surveys show both in heavy use. In a study of 79 private equity firms, MOIC was used to judge 94.8% of deals and gross IRR 92.7% (Gompers, Kaplan & Mukharlyamov, 2016). Venture investors lean further towards multiples: in a survey of 885 VCs, 63% evaluated deals with cash-on-cash multiples, 42% with IRR and 22% with net present value (Gompers et al., 2020). The VCs who did use IRR wanted 31% on average from a new deal.

The case against IRR has been made forcefully and repeatedly. Phalippou again: “IRR is not a rate of return. It embeds a very strong and often absurd assumption: that all interim cash flows can be reinvested at the same IRR for decades.” In his 2024 essay “The Tyranny of IRR” he set a 62% annual return against a multiple of 2.1 per pound invested on the same track record (Phalippou, 2024). By June 2026 he told PitchBook that “for evaluating private equity funds, IRR is largely obsolete” (PitchBook, Jun 2026).

The most practical objection is that IRR can be engineered. A subscription credit line lets a fund invest with borrowed money and call capital from LPs months later; the money is at work for less time on paper, so the IRR rises. In ILPA’s own example, delaying capital calls with a credit line lifts IRR from 6.62% to 7.98% while TVPI falls from 1.45x to 1.35x (ILPA, 2017). Across a large sample, Albertus and Denes measured credit lines raising IRR-based performance by 6.1 percentage points while multiples slightly declined (Albertus & Denes, 2019).

The multiple has its own blind spot, and it is the mirror image. “The major drawbacks of MOIC are that it does not account for investment horizon or level of investment risk”, as Brown, Lundblad and Volckmann put it in 2025 (Brown et al., 2025). A 3x over fifteen years is not obviously better than a 2x over four.

The SEC captured why the pair belongs together when it explained its 2023 private fund rule: “a high internal rate of return but a low multiple of invested capital likely means that the investment was not held long enough to generate substantial returns for the fund” (SEC, 2023). The rule would have required both in quarterly statements; the Fifth Circuit vacated it in June 2024, writing that “no part of it can stand” (Fifth Circuit, 2024). The industry adopted the substance anyway: ILPA’s 2025 template standardizes “IRRs and TVPI/MOIC, with designated breakouts for reporting the relevant gross and net figures with and without the impact of fund-level subscription facilities” (ILPA, 2025).

So the honest answer to the MOIC-or-IRR question is both, side by side, with the credit line stripped out. Harris, Jenkinson, Kaplan and Stucke said as much in their 2023 paper: “The IRR and MOIC are the standard performance measures used by PE practitioners” (Harris et al., 2023).

TVPI vs DPI vs RVPI: paper against cash

If MOIC against IRR is the oldest argument, TVPI against DPI is the loudest one today. It is the difference between what a fund says it is worth and what it has paid.

Look at a mature fund. The median 2017-vintage US venture fund on Carta reported a net TVPI of 1.64x in the first quarter of 2026, of which only 0.31x had been paid out in cash. Roughly four-fifths of its value was still on paper after about nine years (our arithmetic), and fewer than 20% of 2017 and 2018 funds had reached a DPI of 1x (Carta, Q1 2026). Carta’s own summary: “Unrealized valuations of VC-owned assets may be trending up. But realized gains—the deals that actually put cash in investors’ pockets—are still relatively few and far between.”

Newer vintages are further behind. The 2021 vintage’s average DPI of 0.05x is “the lowest five-year DPI multiple this century”, PitchBook reported in August 2026, and “Since 2022, net cash flow to LPs has been negative $202 billion, even as market value and AUM have continued to increase” (PitchBook, Aug 2026). AngelList’s mid-year refresh said it plainly: “Paper marks are recovering faster than cash is coming back to LPs” (AngelList, 2026).

The paper part is only as good as the marks behind it, and venture marks are soft. Stanford’s Ilya Strebulaev: “A venture portfolio is often marked at the last round, which is a number produced by a negotiation that may be eighteen months stale and is not equal to fair value in the first place” (Strebulaev, 2026). Allocate co-founder Samir Kaji went further in July 2026: “We’re somewhere in the late innings of this phase, and a lot of today’s marks will turn out to have been fiction” (Kaji, 2026).

The secondary market puts a price on that doubt. Venture LP stakes traded at 79% of NAV in the first half of 2026, against about 91% for buyout (Jefferies, Jul 2026). An LP who sells turns RVPI into DPI at a discount, which is one reason secondaries have become a tool for managing DPI.

Yet TVPI is not useless; it is early. PitchBook’s Kyle Stanford found that at a fund’s midpoint, TVPI is “a better indicator” of where a vintage will end up than year-five DPI (PitchBook, Aug 2026). VenCap’s David Clark found the same from the other side: across 71 funds from the 2005–08 vintages, year-five DPI correlated just 0.22 with later DPI (Clark, 2024). Late cash is normal. Sapphire Partners’ Beezer Clarkson: “We leveraged our database of funds that have returned capital, and on average funds get to 1x DPI by year 8” (Clarkson, 2024).

The practical reading: TVPI tells you where a fund is probably heading; DPI tells you how much of the journey is finished; RVPI tells you how much is still at risk. A young fund should be judged mostly on the first, a mature fund mostly on the second, and any fund past its tenth year with a large RVPI deserves hard questions about the marks.

The J-curve: why every fund looks bad first

Every closed-end fund starts in a hole. Management fees are charged on committed capital from day one; a 2% annual fee over ten years consumes 20% of the fund’s committed capital (Metrick & Yasuda, 2010). Meanwhile new investments sit at cost and the early write-downs arrive before the early winners. Hamilton Lane says the negative period “generally spans three to four years following the fund’s inception” (Hamilton Lane).

On cash alone the hole is deeper and longer. In Ljungqvist and Richardson’s study of fund cash flows, it took “almost until year 8” for the average and median fund IRR to turn positive, and “venture funds take about a year longer to break even” (Ljungqvist & Richardson, 2003). Capital Dynamics described the mechanics in 2009: “The more quickly fund managers invest capital, the steeper the J-Curve. The longer it takes to generate distributions, the longer (and usually deeper) the trough of the J-Curve” (Capital Dynamics, 2009).

The J-curve is also where IRR is at its most misleading. Early IRRs are computed on tiny, recent cash flows, so a credit line or a single early mark-up can swing them wildly; Albertus and Denes found subscription lines inflating young funds’ IRRs by 9.7 points on average. ILPA’s guidance notes that a subscription line “shortens the J-curve”. That is a feature for a GP raising its next fund, and a reason for LPs to ask for every IRR with and without the line.

How investors’ weight has shifted

If you lined up LP reports from 1990, 2005, 2021 and 2026, you would see the same six numbers. What changes is the order in which people read them.

For most of private equity’s history, IRR led. Venture Economics collected it from GPs and LPs every quarter from 1980 onwards, and the first large studies of fund returns were built on that data (Kaplan & Schoar, 2005). Multiples became formal reporting requirements only when the GIPS private equity provisions took effect in 2005–06.

The first serious challenge came from inside the industry. In July 2006, Oaktree’s Howard Marks titled a memo “You Can’t Eat IRR” and wrote: “A high internal rate of return does not in and of itself put money in one’s pocket” (Oaktree, 2006). Academics followed. Kaplan and Schoar’s 2005 paper introduced the public market equivalent, which Kaplan later described as a “market-adjusted multiple” (Kaplan, 2024); Harris, Jenkinson and Kaplan argued in 2014 that “multiples of invested capital should be preferred to IRRs as summary measures of private equity performance” (Harris et al., 2014).

The boom of 2019 to 2021 pulled attention back to paper. Three years in, the median 2019-vintage US VC fund on Carta showed a 19.4% net IRR; the 2021 vintage, invested at the peak, showed −1.5% at the same age (Carta, 2024). Then rates rose, exits stalled and the conversation flipped. Bain’s July 2023 midyear report put it in a phrase that stuck: “For cash-strapped LPs, DPI (distributed to paid-in capital) is becoming the new IRR (internal rate of return)” (Bain, 2023). By February 2024 an investor’s T-shirt reading “DPI is the new IRR” made the news as payouts at major private equity firms fell 49% in two years (Bloomberg, 2024). Mercury’s Aziz Gilani summed up the mood: “But from my perspective, DPI is the metric that rules them all” (Carta, Oct 2024).

Where has it settled? McKinsey’s 2026 survey of LPs says IRR has not been dethroned: “IRR remains the leading focus”, while “DPI is now considered “critical” or “most critical” by 54 percent of LPs, tied with MOIC as the second-most-important performance metric” (McKinsey, 2026).

The shift is less a revolution than a rebalancing. IRR is still the headline, but it no longer travels alone.

1. 1980–2001

   The IRR era begins

   • IRR
   • TVPI
   • DPI

   Venture Economics collects IRR, TVPI and DPI from GPs and LPs every quarter, the data behind the first large studies of fund returns. (Kaplan & Schoar, 2005)

2. 2005–06

   Multiples become required reporting

   • TVPI
   • DPI
   • RVPI

   The GIPS private equity provisions require the investment multiple (TVPI), the realization multiple (DPI) and RVPI for each year presented, next to a since-inception IRR. (GIPS, 2005)

3. 2006

   The first famous warning about IRR

   • DPI

   Howard Marks: "A high internal rate of return does not in and of itself put money in one's pocket." (Oaktree, 2006)

4. 2014–2019

   Researchers push multiples

   • MOIC
   • TVPI

   Harris, Jenkinson and Kaplan argue that "multiples of invested capital should be preferred to IRRs as summary measures of private equity performance". (Harris et al., 2014)

5. 2019–2021

   Paper returns boom

   • IRR
   • TVPI

   Three years in, the median 2019-vintage US VC fund on Carta showed a 19.4% net IRR. The 2021 vintage, invested at the peak, showed −1.5% at the same age. (Carta, 2024)

6. 2022–2024

   Cash becomes king

   • DPI

   Bain, July 2023: "For cash-strapped LPs, DPI (distributed to paid-in capital) is becoming the new IRR (internal rate of return)." (Bain, 2023)

7. 2025–2026

   Both, reported both ways

   • IRR
   • DPI
   • MOIC
   • TVPI

   McKinsey's 2026 LP survey: "IRR remains the leading focus", with DPI and MOIC tied second; ILPA's new template asks for every figure with and without subscription lines. (McKinsey, 2026)

Who reads which number

LPs in venture funds live with long J-curves and soft marks, so they read TVPI for direction and DPI for proof. The bar they talk about is 3x net TVPI, which Carta calls “commonly viewed as the threshold for true success among mature VC funds”. Few clear it: the top quartile of 2017-vintage funds starts at 2.20x. Venture returns are also extremely concentrated. Looking at US venture-funded companies that exited over the previous decade, Correlation Ventures found that “less than 4% of the capital invested into venture-funded companies exiting over the last decade generated a 10X or greater multiple”, and that “nearly half of financings lost money for investors” (Correlation Ventures, 2023). That is why venture LPs care about whether a manager has “fund returners”: VenCap’s David Clark says “what the data tells us is that fund returners are more predictive of performance than unicorns” (Balentic, 2025).

LPs in buyout funds back cash-generating companies, often bought with debt, so IRR and MOIC carry more weight. Private equity investors have typically targeted a 22% IRR, most between 20% and 25%, and their funds break even sooner: Ljungqvist and Richardson found that “venture funds take about a year longer to break even”. Their current headache is the backlog: Bain counts “32,000 unsold companies worth $3.8 trillion” (Bain, 2026), which is RVPI waiting to become DPI.

GPs raising a fund lead with whatever their track record shows best, which is exactly why LPs ask for all of it. Carta’s 2017-vintage figures show how far apart the numbers sit: a 9.2% median net IRR and a 15.5% top-quartile line, against VCs’ 31% required return on a new deal. Fees, carry and the deals that fail make up the gap (our comparison).

Founders and boards rarely see fund metrics, but they feel them. A fund under pressure to show DPI wants exits and secondary sales; a fund raising its next vehicle wants mark-ups. Understanding where your investor’s fund sits on its J-curve explains a lot about its behaviour in your boardroom.

Who leans on which number

Metric	VC fund LPs	Buyout LPs	GPs raising a fund	Founders and boards
IRR	Core	Core	Core	Rarely used
TVPI	Core	Useful	Core	Rarely used
DPI	Core	Core	Useful	Rarely used
MOIC	Useful	Core	Core	Useful
RVPI	Useful	Useful	Rarely used	Rarely used
J-curve	Useful	Useful	Useful	Rarely used

Beyond venture: buyouts, secondaries and every closed-end fund

These metrics were not invented for venture capital and are not confined to it. The GIPS standards require TVPI, DPI, RVPI and the PIC multiple for any pooled fund with committed capital, and the same arithmetic runs through buyout, growth and other closed-end private funds.

What changes across strategies is the weighting. In buyouts, the 2.5x-in-five-years rule anchors expectations and IRR targets cluster between 20% and 25%. Venture’s power law makes multiples and DPI the more honest measures. Continuation vehicles, where a GP sells assets from an old fund to a new one it also manages, now write the GP’s pay in both: in Morgan Lewis’s 2026 study, “79% of CVs include a tiered carry with 60% adopting both internal rate of return (IRR) and multiple on invested capital (MoIC) return thresholds” (Morgan Lewis, 2026). In PitchBook’s example, “a 2x return might net the manager a 20% profit share, with a 3x return bringing it 30%” (PitchBook, Aug 2026).

Early-stage venture is even inventing variants. Homebrew’s Hunter Walk described a column some managers now add to LP reports: “The increasingly standard way to communicate this is by adding a column to your financial reporting that’s essentially ‘SAFE Adjusted TVPI’ alongside your more standard TVPI calculations” (Walk, 2026).

Where each metric misleads

IRR flatters short holds, early exits and anything financed with a credit line, and it assumes interim cash can be reinvested at the same rate. Ask for it with and without the subscription line, and read it next to a multiple.

MOIC ignores time and risk. A 3x that took fifteen years can be worse than a 2x that took four. Check whether it is gross or net; the difference is fees and carry.

TVPI and RVPI are only as good as the marks. Venture holdings are often marked at the last round, and as Carta noted of marking venture holdings, “There is no widely accepted industry standard” (Carta, Nov 2024). LP Chris Douvos called TVPI “the private equity equivalent” of a misleading early score as long ago as 2013 (Douvos, 2013).

DPI is hard to fake but easy to borrow. ILPA’s 2024 guidance on NAV facilities warns that distributions funded with fund-level loans flatter DPI (ILPA, 2024). And early DPI is a weak predictor of final DPI.

The J-curve can be bent cosmetically by credit lines without changing what investors eventually receive.

A reading order for a fund report

When a quarterly report lands, read it in this order.

1. Vintage and age. A five-year-old venture fund and a ten-year-old one are judged by different numbers.
2. DPI. How much cash has come back, and is it above 1x?
3. TVPI, split into DPI and RVPI. How much of the headline is still paper, and how old are the marks?
4. Net IRR, with and without the credit line. Does the rate survive once borrowing is stripped out?
5. Gross MOIC by holding. Is the fund carried by one or two winners, and are they realized?
6. The benchmark. Compare with funds of the same vintage and strategy, and with a public market equivalent if the GP provides one.

Fireside: what the numbers can’t tell you

Fund metrics are summaries of cash flows, and every summary throws something away. IRR throws away scale, multiples throw away time, and TVPI hides the difference between a check in the post and a number in a spreadsheet. The professionals who read these reports best tend to do the same thing: they look at two or three metrics at once and treat any number that travels alone with suspicion.

The current moment is unusual. As of mid-2026, paper value across venture is high, cash returned is low, and net cash flow to venture LPs has been negative $202 billion since 2022. That puts enormous weight on the RVPI line of every report, which is exactly the line built on the softest inputs. Steven Kaplan’s notes on net IRR list its basic limits plainly: “Net IRR » Absolute (not relative) - does not control for the market. » Is sensitive to sequencing of investments” (Kaplan, 2024). None of these metrics, on its own, tells you whether the fund beat what the same money would have earned in public markets.

There is a temptation to crown a winner: “DPI is the new IRR”, or Phalippou’s claim that IRR is “largely obsolete”. We would put it differently. Each number answers a different question, and the questions that matter change with the cycle. In a boom, LPs need help discounting paper; in a drought, they need help valuing patience. The right habit is the one regulators, ILPA and the best LPs have converged on: report everything, gross and net, with and without leverage, by vintage, and let the reader see how the numbers disagree.

David Zhou’s 2026 framing is the one we keep coming back to: “We live in a world where DPI (distributions to paid-in capital) is a harsh judge” (Zhou, 2026). Harsh judges are useful. So are patient ones. A good fund report gives you both, and lets you decide how much weight each deserves for the fund in front of you.