Dave McClure has been a Silicon Valley entrepreneur and investor for over 25 years. He has invested in hundreds of startups around the world, including 10+ IPOs and 40+ unicorns (Credit Karma, Twilio, SendGrid, Lyft, The RealReal, Talkdesk, Grab, Intercom, Canva, Udemy, Lucid, GitLab, Reddit, Stripe, Bukalapak).
Prior to launching PVC in 2019, he was the founding partner of 500 Startups, a global VC firm with $1B AUM that has invested in over 2,500 companies and 5,000 founders across 75 countries. Dave created 20 VC funds under the 500 brand and invested in 20 other VC funds around the world.
Dave began his investing career at Founders Fund where he made seed-stage investments in 40 companies, resulting in 4 unicorns and 3 IPOs. He led the Credit Karma seed round in 2009 (acq INTU, over 400X return). His $3M portfolio returned more than $200M (~65X) in under 10 years.
Before he became an investor, Dave was Director of Marketing at PayPal from 2001-2004. He was also the founder/CEO of Aslan Computing, acquired by Servinet in 1998. Dave graduated from the Johns Hopkins University (BS, Engineering / Applied Mathematics).
[00:00] Intro [03:37] How did Narnia inspire the start of Dave’s entrepreneurship? [08:32] On the brink of bankruptcy [11:42] The lesson Dave took away from his first acquisition [13:19] What did Dave do that no one else did as a marketing director? [16:06] What do most people fail to appreciate about secondaries? [22:31] The 3 bucket method for secondaries [28:46] How much do fund returners matter for secondaries? [33:01] When do LPs typically think about selling fund secondaries? [42:04] What are two questions that Dave asks to see if a portfolio is good for a secondary? [46:10] Why is it complicated if a GP wants to buy an LP’s stake? [55:03] When do most funds return 1X? 2-3X? [57:13] Underwriting VC vs PE secondaries [1:01:49] How do institutional LPs react to VC secondaries? [1:07:01] The founding story of Practical VC [1:15:36] Closing Josh Kopelman in Fund I [1:18:47] How often does the PayPal Mafia get together? [1:23:49] What’s the most expensive lessons Dave learned over the years? [1:27:38] Thank you to Alchemist Accelerator for sponsoring! [1:28:29] If you enjoyed the episode, would deeply appreciate you sharing with one other friend!
Soooooooo… (I know, what a great word to start a blogpost) I started this essay, with some familiarity on one subject. Little did I know I was going to learn about an entirely different industry, and be endlessly fascinated about that.
The analogy that kicked off this essay is that re-upping on a portfolio company is very much like re-signing a current player on a sports team. That was it. Simple as it was supposed to sound. The goal of any analogy was to frame a new or nuanced concept, in this case, the science of re-upping, under an umbrella of knowledge we were already familiar with.
But, I soon learned of the complexity behind re-upping players’ contracts, as one might assume. And while I will claim no authority over the knowledge and calculations that go into contracts in the sports arena, I want to thank Brian Anderson and everyone else who’s got more miles on their odometer in the world of professional sports for lending me their brains. Thank you!
As well as Arkady Kulik, Dave McClure, and all the LPs and GPs for their patience and willingness to go through all the revisions of this blogpost!
While this was a team effort here, many of this blogpost’s contributors chose to stay off the record.
The year was 1997.
Nomar Garciaparra was an instantaneous star, after batting an amazing .306/.342/.534. For the uninitiated, those are phenomenal stats. On top of batting 30 home runs and 11 triples – the latter of which was a cut above the rest of the league, it won him Rookie of the Year. And those numbers only trended upwards in the years after, especially in 1999 and 2000. Garciaparra became the hope for so many fans to end the curse of the Bambino – a curse that started when the Red Sox traded the legendary Babe Ruth to the Yankees in 1918.
Then 2001 hit. A wrist injury. An injured Achilles tendon. And the fact he needed to miss “significant time” earned him a prime spot to be traded. Garciaparra was still a phenomenal hitter when he was on, but there was one other variable that led to the Garciaparra trade. To Theo Epstein, above all else, that was his “fatal flaw.”
Someone that endlessly draws my fascination is Theo Epstein. Someone that comes from the world of baseball. A sport that venture draws a lot of inspiration, at least in analogy, like one of my fav sayings, Venture is one of the only types of investments where it’s not about the batting average but about the magnitude of the home runs you hit.
If you don’t follow baseball, Theo Epstein is the youngest general manager in the history of major league baseball at 26. But better known for ending the Curse of the Bambino, an 86-year curse that led the Red Sox down a championship drought that started when the Red Sox traded Babe Ruth to the Yankees. Theo as soon as he became general manager traded Nomar Garciaparra, a 5-time All-star shortstop, to the Cubs, and won key contracts with both third baseman Bill Mueller and pitcher Curt Schilling. All key decisions that led the Red Sox to eventually win the World Series 3 years later.
And when Theo left the Red Sox to join the Chicago Cubs, he also ended another curse – The Curse of the Billy Goat, ending with Theo leading them to a win in the 2016 World Series. You see, in baseball, they measure everything. From fly ball rates to hits per nine innings to pitches per plate appearance. Literally everything on the field.
But what made Theo different was that he looked at things off the field. It’s why he chose to bet on younger players than rely on the current all-stars. It’s why he measures how a teammate can help a team win in the dugout. And, it’s why he traded Nomar, a 5-time All Star, as soon as he joined, because Nomar’s “fatal flaw” was despite his prowess, held deep resentment to his own team, the Sox, when they tried to trade him just the year prior for Alex Rodriguez but failed to.
So, when Danny Meyer, best known for his success with Shake Shack, asked Theo what Danny called a “stupid question”, after the Cubs lost to the Dodgers in the playoffs, and right after Houston was hit by a massive hurricane, “Theo, who are you rooting for? The Dodgers so you can say you lost to the winning team, or Houston (Astros), because you want something good to happen to a city that was recently ravaged by a hurricane.”
Theo said, “Neither. But I’m rooting for the Dodgers because if they win, they’ll do whatever every championship team does and not work on the things they need to work on during the off season. And the good news is that we have to play them 8 times in the next season.”
You see, everyone in VC largely has access to the same data. The same Pitchbook and Crunchbase stat sheet. The same cap table. And the same financials. But as Howard Marks once said in response how you gain a knowledge advantage:
“You have to either:
Somehow do a better job of massaging the current data, which is challenging; or you have to
Be better at making qualitative judgments; or you have to
Be better at figuring out what the future holds.”
For the purpose of this blogpost, we’re going to focus on the first one of the three.
What is value?
To begin, we have to first define a term that’ll be booking its frequent flier miles for the rest of this piece – expected value.
Some defined it as the expectation of future worth. Others, a prediction of future utility. Investopedia defines it as the long-term average value of a variable. Merriam-Webster has the most rudimentary definition:
The sum of the values of a random variable with each value multiplied by its probability of occurrence
On the other hand, venture is an industry where the beta is arguably one of the highest. The risk associated with outperformance is massive as well. And the greatest returns, in following the power law, are unpredictable.
We’re often blessed with hindsight bias, but every early-stage investor in foresight struggles with predicting outlier performance. Any investor that says otherwise is either deluding you or themselves or both. At the same time, that’s what makes modeling exercises so difficult in venture, unlike our friends in hedge funds and private equity. Even the best severely underestimate the outcomes of their best performers. For instance, Bessemer thought the best possible outcome for Shopify was $400M with only a 3% chance of occurring.
Similarly, who would have thought that jumping in a stranger’s car or home, or live streaming gameplay would become as big as they are today. As Strauss Zelnick recently said, “The biggest hits are by their nature, unexpected, which means you can’t organize around them with AI.” Take the word AI out, and the sentence is equally as profound replaced with the word “model.” And it is equally echoed by others. Chris Paik at Pace has made it his mission to “invest in companies that can’t be described in a single sentence.”
But I digress.
Value itself is a huge topic – a juggernaut of a topic – and I, in no illusion, find myself explaining it in a short blogpost, but that of which I plan to spend the next couple of months, if not years, digging deeper into, including a couple more blogposts that are in the blast furnace right now. But for the purpose of this one, I’ll triangulate on one subset of it – future value as a function of probability and market benchmarks.
In other words, doubling down. Or re-upping.
For the world of startups, the best way to explain that is through a formula:
E(v) = (probability of outcome) X (outcome)
E(v) = (graduation rate) X (valuation step up from last round) X (dilution)
For the sake of this blogpost and model, let’s call E(v), appreciation value. So, let’s break down each of the variables.
Graduation rates
What percent of your companies graduate to the next round? I shared general benchmarks in this blogpost, but the truth is it’s a bit more nuanced. Each vertical, each sub-vertical, each vintage – they all look different. Additionally, Sapphire’s Beezer recently said that it’s normal to expect a 20-30% loss ratio in the first five years of your fund. Not all your companies will make it, but that’s the game we play.
On a similar note, institutional LPs often plan to build a multi-fund, multi-decade relationship with their GPs. If they invest in a Fund I, they also expect to be there by Fund III.
Valuation step ups
How much greater is the next round’s valuation in comparison to the one in which you invested? Twice as high? Thrice? By definition, if you double down on the same company, rather than allocate to a net new company, you’re decreasing your TVPI. And as valuations grow, the cost of doubling down may be too much for your portfolio construction model to handle, especially if you’re a smaller sub-$100M fund.
It’s for the same reason that in the world of professional sports, there are salary caps. In fact, most leagues have them. And only the teams who:
Have a real chance at the championship title.
Have a lot in their coffers. This comes down to the composition of the ownership group, and their willingness to pay that tax.
And/or have a city who’s willing to pay the premium.
… can pay the luxury tax. Not to be too much of a homer, but the Golden State Warriors have a phenomenal team and are well-positioned to win again (at least at the time of this blogpost going out). So the Warriors can afford to pay the luxury tax, but smaller teams or teams focused on rebuilding can’t.
The Bulls didn’t re-sign the legendary Michael Jordan because they needed to rebuild. Indianapolis didn’t extend Peyton Manning’s contract ‘cause they didn’t have the team that would support Peyton’s talents. So, they needed to rebuild with a new cast of players.
Similarly, Sequoia and a16z might be able to afford to pay the “luxury tax” when betting on the world’s greatest AI talent and for them to acquire the best generative AI talent. Those who have a real chance to grow to $100M ARR, given adoption rates, retention rates, and customer demand. But as a smaller fund or a fund that has a new cast of GPs (where the old guard retired)… can you?
Dilution
If a star player is prone to injury or can only play 60 minutes of a game (rather than 90 minutes), a team needs to re-evaluate the value of said player, no matter how talented they are. How much of a player’s health, motivation, and/or collaborativeness – harkening back to the anecdote of Nomar Garciaparra at the beginning – will affect their ability to perform in the coming season?
Take, for instance, the durability of a player. If there ‘s a 60% chance of a player getting injured if he/she plays longer than 60 minutes in a game and a 50% of tearing their ACL, while they may your highest scorer this season, they’re not very durable. If that player missed 25% of practices and 30% of games, they just don’t have it in them to see the season through. And you can also benchmark that player against the rest of the team. How’s that compared with the team’s average?
Of course, there’s a parallel here to also say, every decision you make should be relative to industry and portfolio benchmarks.
How great of a percentage are you getting diluted with the next round if you don’t maintain your ownership? This is the true value of your stake in the company as the company grows.
How does one use the appreciation value equation?
E(v) = (graduation rate) X (valuation step up from last round) X (dilution)
If the expected value is greater than one, the company is probably not worth re-upping. And that probably means the company is overhyped, or that that market is seeing extremely deflated loss ratios. In other words, more companies than should be, are graduating to the next stage; when in reality, the market is either a winner-take-all or a few-take-all market. If it is less than or equal to one, then it’s ripe to double down on. In other words, the company may be undervalued.
And to understand the above equation or for it to be actually useful (outside of an abstract concept), you need market data. Specifically, around valuation step ups as a function of industry and vertical.
If you happen to have internal data across decades and hundreds of companies, then it’s worth plugging in your own dataset as well. It’s the closest you can get to the efficient market frontier.
But if you lack a large enough sample size, I’d recommend the below model constructed from data pulled from Carta, Pitchbook, and Preqin and came from the minds of Arkady Kulik and Dave McClure.
The model
The purpose of this model is to help your team filter what portfolio companies are worth diving deeper into and which ones you may not have to (because they didn’t pass the litmus test) BEFORE you evaluate additional growth metrics.
It is also important to note that the data we’ve used is bucketed by industry. And in doing so, assumptions were made in broad strokes. For example, deep tech is broad by design but includes niche-er markets that have their own fair share of pricing nuances in battery or longevity biotech or energy or AI/ML. Or B2B which include subsectors in cybersecurity or infrastructure or PLG growth.
Take for instance…
Energy
Energy sector appreciation values and follow-on recommendations
The energy sector sees a large drop in appreciation value at the seed stage, where all three factors contribute to such an output. Valuation step-up is just 1.71X, graduation rates are less than 50% and dilution is 38% on average.
Second phase where re-upping might be a good idea is Series B. Main drivers as to such a decision are that dilution hovers around 35% and about 50% of companies graduate from Series A to Series B. Mark ups are less significant where we generally see only an increase in valuation at about 2.5X, which sits around the middle of the pack.
Biotech
Biotech sector appreciation values and follow-on recommendations
The biotech sector sees a large drop in appreciation value at the Seed stage. This time, whereas dilution seems to match the pace of the rest of the pack (at an average of 25%), the two other factors shine greater in making a follow-on decision. Valuation step up are rather low, sitting at 1.5X. And less than 50% graduate to the next stage.
In the late 2023 market, one might also consider re-upping at the Series C round. Main driver is the unexpectedly low step-up function of 1.5X, which matches the slow pace of deployment for growth and late stage VCs. On the flip side, a dilution of 17% and graduation rate of 60% are quite the norm at this stage.
In closing
All in all, the same exercise is useful in evaluating two scenarios – either as an LP or as a GP:
Is your entry point a good entry point?
Between two stages, where should you deploy more capital?
For the former, too often, emerging GPs take the stance of the earlier, the better. Almost as if it’s a biblical line. It’s not. Or at least not always, as a blanket statement. The point of the above exercise is also to evaluate, what is the average value of a company if you were to jump in at the pre-seed? Do enough graduate and at a high enough price for it to make sense? While earlier may be true for many industries, it isn’t true for all, and the model above can serve as your litmus test for it. You may be better off entering at a stage with a higher scoring entry point.
For the latter, this is where the discussion of follow on strategies and if you should have reserves come into play. If you’re a seed stage firm, say for biotech, using the above example, by the A, your asset might have appreciated too much for you to double down. In that case, as a fund manager, you may not need to deploy reserves into the current market. Or you may not need as large of a reserve pool as you might suspect. It’s for this reason that many fund managers often underallocate because they overestimate how much in reserves they need.
If you’re curious to play around with the model yourself, ping Arkady at ak@rpv.global, and you can mention you found out about it through here. 😉
Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!
The views expressed on this blogpost are for informational purposes only. None of the views expressed herein constitute legal, investment, business, or tax advice. Any allusions or references to funds or companies are for illustrative purposes only, and should not be relied upon as investment recommendations. Consult a professional investment advisor prior to making any investment decisions.
There’s this line I love in Jerry Colonna’s Reboot, and I’m loosely paraphrasing just because I’m travelling and I don’t have the book in front of me, “The saying is buy low, sell high; not buy lowest, sell highest.”
The reason I bring up that line is that I’ve been hearing a lot of investors talk about timing the market. At least that was the case before this wonderful trip I’ve been taking across the Pacific, as I sip my hojicha atop my hotel in the backdrop of the Kyoto evening metropolis. When’s a good time to sell? What price makes sense on the secondary market? Should I be investing now? When’s a good time to re-up? Is it a good idea to re-up? Should I be generating DPI for my investors now? Or should I hold? When should I start my fund? When should I begin fundraising?
Now, I don’t pose the above questions as if I have all the answers. In fact, I don’t. I try to. But I don’t. Although I’ve heard 50-60% is the discount secondary buyers have been able to get for great companies that became overvalued in the pandemic days. On the flip side, while Dave and I did published a blogpost not too long ago on early DPI, the truth is there are different ways to make money. Ed Zimmerman shared some of his investments’ data recently to illustrate that exact point.
Another obvious truth is that as investors for an alternative asset class — hell for any asset class, our job is to make our LPs money. Ideally, more money than we were given. For other asset classes, it’s measured in percentages. For venture, it’s multiples. And because of that raison d’être, it’s our job to think not only about the upside, but also the downside protection. Hence, why early DPI matters in some of your best outliers. It always matters.
But from what I’m seeing and hearing, it matters more in a bear market, like today. Than the bull we were in yesterday. Why?
Liquidity is a differentiator.
Because of the point 1, giving LPs some liquidity back makes it easier to get to conviction as you raise your next fund.
Point 2 holds the most weight if you’re an emerging manager on Funds I through III, or have sub $100M AUM. Although Funds I and II, you have little to go off of. As such, sticking to your strategy may be more important to some LPs. In other words, consistency.
Also seems to matter more if your LPs are investing off balance sheet. For instance, corporates.
While I was in Tokyo earlier this trip, I caught up with a colleague. We spent the evening chatting about fund managers and current deployment schedules. (In case you’re wondering, no, we didn’t spend the whole time talking the biz.) And we see a lot of folks slowing down their pace of deployment. Could be the case of deal flow contraction, as Chris Neumann recently wrote about. Could be the case of loss of conviction behind initial fund strategy. We’ve also seen examples of VCs stretching their deployment schedule as their fundraises have been extended to 2024. All in all, that means VCs’ bar for “quality” has gone up.
But let me explain in a bit why I put “quality” in quotation marks.
So, timing comes down to two things:
Entry point
Exit point
I’ve seen a plurality of investors consider exit options as a means to *crossing fingers* convince existing LPs to re-up to the next fund. Debatable on how effective it is. As many LPs I’ve chatted with are “graduating” a lot more of their GPs than years prior. In other words, fancy shmancy word for they’re not re-upping on certain existing managers. Some LPs say it’s an AUM problem (but I’ve also seen them make exceptions). Others say it’s strategy drift. But more so say that certain GPs haven’t been a good fiduciary of capital, which ends being a combination of:
High entry points
Faster than promised deployment schedules (i.e. 1-1.5 years instead of 2-4 years)
Investing in a company where the preference stack is greater than the valuation of the company (similar to the first bullet point)
Reactive communication of strategy drift, instead of preemptive and proactive
Logo shopping which led to strategy drift
All that to say, there are a good amount of LPs who, though appreciate the extra liquidity from partial exits, are not re-investing in existing managers. In addition, they’re holding off until on new ones till earliest Q1 next year to build the relationship earlier. Especially those $5M+ checks.
So, quality, for both GPs and LPs, is this new sugar coating of a term to account for time it takes to figure out where they want to put the next dollar. Investors on both sides are waiting to pull the trigger at 90% conviction, instead of the usual 70%. And realistically, for pre-product market fit companies and firms (i.e. pre-seed, seed startups and Funds I-III), 90% usually never comes until it’s too late. Meaning one misses their entry point.
I have no doubt (as well as many if not all my peers) that the greatest companies of the next generation are being built today. But only a small handful will make it out the gauntlet of fire. Even good companies won’t make it, unfortunately.
So, for the one building, the importance of communicating focus and discipline will be more powerful than ever. My buddy Martin also recently tweeted by an unrelenting focus on a niche audience may serve more useful than targeting a seemingly large TAM.
For the one investing, there is no good time. Our job is to buy low, sell high. Not buy lowest, sell highest. Waiting for the right moment will only have you miss the moment. In the surfing analogy, where the market is the wave, the product is the board, the team is the surfer, and you need all three to be a great surfer, you don’t want to be on the shore when the wave hits. It is better to be paddling in the water before the wave hits than on the shore when the wave does hit. Timing is only obvious in hindsight, never in foresight.
There’s also a great Chinese proverb that the best time to plant a tree was 20 years ago, the next best time is today.
So in this flight to quality, consider what quality actually means. Is it a function of you doubting your original thesis? Then re-examine what caused the doubt. Was your thesis founded on first principles? For consumer, which is where I know a little bit more about, is it founded on the basis and habits of the human condition? Is it secular from technological and hype trends?
Is quality waiting on numbers or external validation? That’s fine if you’re a growth or late stage investor. You’re never going to get it if you’re a true pre-seed and seed. If you’re waiting on a large amount of traction, you’re not an early-stage investor. Round-semantics aside.
You built a fund around a 10-15 year vision. Deploy against that. Or… although we don’t see this much these days, return any remaining capital back to your LPs.
Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!
The views expressed on this blogpost are for informational purposes only. None of the views expressed herein constitute legal, investment, business, or tax advice. Any allusions or references to funds or companies are for illustrative purposes only, and should not be relied upon as investment recommendations. Consult a professional investment advisor prior to making any investment decisions.
Some of you reading here are busy, so we’ll keep this top part brief, as an abstract sharing our top three observations of leading fund managers.
Generally speaking, don’t sell your fast growing winners early.
Except when…
Selling on your way up may not be a crazy idea.
You might sell when you want to lock in DPI. Don’t sell more than 20% of your fund’s positions unless you are locking in meaningful DPI for your fund. For instance, at each point in time, something that’s greater than 0.5X, 1X, 2X, or 3X of your fund size.
You might consider selling when you’ve lost conviction. Consider selling a position when you feel the market has over-priced the actual value, or even up to 100% if you’ve lost conviction.
You might consider selling when one is growing slower than your target IRR. If companies are growing slower and even only as fast as your target IRR, consider selling if not at too much of a discount (Note: there may be some political and/or signaling issues to consider here as well. But will save the topic of signaling for another blog post).
Do note that the above are not hard and fast rules. Every decision should be made in context to other moving variables. And that the numbers below are tailored to early-stage funds.
Net TVPI Benchmarks from Years 5-15Net DPI Benchmarks from Years 5-15
Let’s go deeper…
On a cloudless Friday morning, basking in the morning glory of Los Altos, between lattes and croissants, between two nerds (or one of whom might identify as a geek more than a nerd), we pondered one question:
Everyone seems to have a financial model for when and how to invest, but part of being a fiduciary of capital is also knowing when to distribute – when to sell. When RVPI turns into DPI. And we haven’t seen many models for selling yet. At least none have surfaced publicly or privately for us. The best thought piece we’ve seen in the space has been Fred Wilson’s Taking Money “Off the Table”. At USV, they “typically seek to liquidate somewhere between 10% and 30% of our position in these pre-IPO liquidity transactions. Doing so allows us to hold onto the balance while de-risking the entire investment.”
As such, it gave birth to a thought… What if selling was more of a science?
What would that look like?
Between two Daves, it was not the Dave with sneakers and a baseball cap and with the profound disregard to healthy diets, given the fat slab of bacon in his croissan’wich, who had the answer there.
“To start off, in a concentrated portfolio of 30 investments, a fund returner is a 30x investment. For a 50-investment fund, it’s 50x. And while hitting the 0.5x DPI milestone by years 5-8, and a 2x DPI milestone by years 8-12, is the sign of a great fund, you shouldn’t think about selling much of your TVPI for DPI unless or until your TVPI is starting to exceed 2-3x.” Which seems to corroborate quite well with Chamath Palihapitiya’s findings that funds between 2010 and 2020 convert have, on average, converted about 25% of their TVPI to DPI.
“Moreover, usually you shouldn’t be selling more than 20% of the portfolio at one time (unless you’re locking in / have already locked in 3X or more DPI). You should be dollar-cost averaging – ensuring time diversity – on the way out as well. AND usually only if a company that’s UNDER-growing or OVER-valued compared to the rest of your portfolio. Say your portfolio is growing at 30% year-over-year, but an individual asset is growing slower at only 10-20% OR you believe it is overvalued, that’s when you think about taking cash off the table. Sell part (or even all) of your stake, if selling returns a meaningful DPI for the fund, and if you’re not capping too upside in exchange for locking in a floor.”
Meaningful DPI, admittedly, does mean different benchmarks for different kinds of LPs. For some, that may mean 0.25X. For others that may mean north of 0.5X or 1X.
“On the other hand, if a company is outperforming / outgrowing the rest of the portfolio, generally hold on to it and don’t sell more than 10-20% (again, unless you’re locking in meaningful DPI, or perhaps if it’s so large that it has become a concentration risk).”
I will caveat that there is great merit in its counterpart as well. Selling early is by definition capping your upside. If you believe an asset is reaching its terminal value, that’s fine, but do be aware of signaling risk as well. The latter may end up being an unintended, but self-fulfilling prophecy.
So, it begged the question: Under the assumption that funds are 15-year funds, what is meaningful DPI? TVPI? At the 5-year mark? 7.5 years in? 10 years? And 12.5 years?
The truth is the only opportunities to sell come from the best companies in your portfolio. And probably the companies, if anything, you should be holding on to. By selling early, you are capping your downside, but at the same time capping your upside on the entire portfolio. When the opportunity arises to lock in some DPI, it’s worth considering the top 3-5 positions in your fund. For instance, if your #2 company is growing quickly, you may not be capping the upside as much.
Do keep in mind that sometimes it’s hard to fully conceptualize the value of compounding. As one of my favorite LPs reminded me, if an asset is growing 35% year-over-year, the last 20% of the time produces 56% of the return. Or if an asset is growing 25% YoY, if you sell 20% earlier (assuming 12 year time horizons), you’re missing out on 45% of the upside.
As a GP, you need to figure out if you’re IRR or multiple focused. Locking in early DPI means your IRR will look great, but your overall fund multiple may suffer.
As an LP, that also means if the gains are taxable (meaning they don’t qualify for QSBS or are sold before QSBS kick in), you need to pay taxes AND find another asset that’s compounding at a similar or better rate. As Howard Marks puts it, you need to find another investment with “superior risk-adjusted prospective returns.”
And so began the search for not just moolah in da coolah, but how much moolah in da coolah is good moolah in da coolah? And how much is great?
Net TVPI Benchmarks from Years 5-15Net DPI Benchmarks from Years 5-15
Some caveats
Of course, if you’ve been around the block for a minute, you know that no numbers can be held in isolation to others. No facts, no data points alienated from the rest.
Some reasons why early DPI may not hold as much weight:
Early acqui-hires. Usually not a meaningful DPI and a small, small fraction of the fund.
There’s a possibility this may be the case for some 2020-2021 vintages, as a meaningful proportion of their portfolio companies exit small but early.
In other words, DPI is constructed of small, but many exits, rather than a meaningful few exits.
TVPI is less than 2-3x of DPI, only a few years into the fund. In other words, their overall portfolio may not be doing too hot. Obviously, the later the fund is to its term, the more TVPI and DPI are alike.
As a believer in the power law, if on average it takes an outlier 8 years to emerge AND the small percentage of winners in the portfolio drive your return, your DPI will look dramatically different in year 5 versus 10. For pre-seed and seed funds, it’s fair to assume half (or more) companies go to zero within the first 3-5 years. And in 10 years, more than 80% of your portfolio value comes from less than 20% of your companies. Hell, it might even be 90% of your portfolio value comes from 10% of your companies. In other words, the power law.
GPs invested in good quality businesses. Some businesses may not receive markups, but may be profitable already, or growing consistently year-over-year that they don’t need to raise another round any time soon.
Additionally, if you haven’t been in the investing game for long, persistence of track record, duration, and TVPI may matter more in your pitch. If you’ve been around the block, IRR and DPI will matter more.
As the great Charlie Munger once said, “selling for market-timing purposes actually gives an investor two ways to be wrong: the decline may or may not occur, and if it does, you’ll have to figure out when the time is right to go back in.” For private market investors, unless you can buy secondaries, you’ll never have a time to go back in until the public offering. As such, it is a one-way door decision.
Some LPs are going to boast better portfolios, and we do admit there will be a few with portfolios better than the above “benchmarks.” And if so, that’s a reason to be proud. In terms of weighting, as a proponent of the power law, there is a high likelihood that we’ve underestimated the percent of crap and meh investments, and overestimated the percent of great investments in an LP’s portfolio. That said, that does leave room for epic fund investments that are outliers by definition.
We do admit that, really, any attempt to create a reference point for fund data before results speak for themselves is going to be met with disagreement. But we also understand that it is in the discourse, will we find ourselves inching closer to something that will help us sleep better at night.
One more caveat for angels… The truth is as an angel, none of the above really matter all that much. You’re not a fiduciary of anyone else’s capital. And your time horizons most likely look different than a fund’s. It’s all yours. So it’s not about capping your downside, but more so about capping your regret. In other words, a regret minimization framework (aka, “spouse regret/yelling minimization insurance”).
That will be so unique to you that there is no amount of cajoling that we could do here to tell you otherwise. And that your liquidity timelines are only really constrained by your own liquidity demands.. For instance, buying a new home, sending kids to college, or taking care of your parents (or YOU!) in their old age.
But I do think the above is a useful exercise to think through selling if you had a fund. You would probably break it down more from a bottoms up perspective. What is your average check size? Do you plan to have a concentrated portfolio of sub-30 investments? Or more? Do you plan to follow on? How much if so? And that is your fund size.
In closing
Returning above a 3x DPI is tough. Don’t take our words for it. Even looking at the data, only 12.5% of funds return over a 3x DPI. And only 2.5% return three times their capital back on more than 2 separate funds.
In the power law game we play, as Michael Mauboussin once said, “A lesson inherent in any probabilistic exercise: the frequency of correctness does not matter; it is the magnitude of correctness that matters.” Most will return zero, or as Jake Kupperman points out: More than 50%.
But it’s in the outliers that return meaningful DPI, not the rest. Not the acqui-hire nor really that liquidation preference on that small acquisition.
At the end of the day, the goal isn’t for any of the above to be anyone’s Bible, but that it’d start a conversation about how people look at early returns. If there is any new data points that are brought up as a result of this blogpost, I’ll do my best to update this thread post-publication.
Big thank you to Dave McClure for inspiring and collaborating on this piece, and to Eric Woo and all our LP friends who’ve helped with the many revisions, sharing data, edits, language and more. Note: Many of our LP friends chose to stay anonymous but have been super helpful in putting this together.
Footnotes
For the purpose of this piece, we know that “good” and “great”, in fact all of the superlative adjectives, are amorphous goalposts. And those words may mean different things to different people. This blogpost isn’t meant to establish a universal truth, but rather serve as a useful reference point for both LPs, looking for “benchmarking” data, and GPs to know where they stand. For the latter, if your metrics do fall in the “good” to “great” range, they’re definitely worth bragging about.
And so with that long preamble, in the piece above, we defined “good” as top quartile, and “great” as top decile. “Good” as a number on its own, enough for an LP to engage in a conversation with you. And “great” as a number that’ll make LPs running to your doorstep. Or at least to the best of our portfolios, leveraging both publicly reported and polled numbers as well as our own.
Our numbers above are also our best attempt in predicting steady state returns, divorcing ourselves from the bull rush of the last 3-5 vintage years. As such, we understand there are some LPs that prefer to do vintage benchmarking, as opposed to steady state benchmarking. And this blogpost, while it has touched on it, did not focus on the former’s numbers.
EDIT (Aug 18, 2023): Have gotten a few questions about where’s the data coming from. The above numbers in the Net DPI and Net TVPI charts are benchmarks the LPs and I agreed on after looking into our own anecdotal portfolios (some spanning 20+ years of data), as well as referencing Cambridge data. These numbers are not the end-all-be-all, and your mileage as an LP may very much vary depending on your portfolio construction. But rather than be the Bible of DPI/TVPI metrics, the purpose of the above is give rough reference points (in reference to our own portfolios + public data) for those who don’t have any reference points.
Stay up to date with the weekly cup of cognitive adventures inside venture capital and startups, as well as cataloging the history of tomorrow through the bookmarks of yesterday!
The views expressed on this blogpost are for informational purposes only. None of the views expressed herein constitute legal, investment, business, or tax advice. Any allusions or references to funds or companies are for illustrative purposes only, and should not be relied upon as investment recommendations. Consult a professional investment advisor prior to making any investment decisions.
