What Early-Stage Investors Can Learn From Quants & Poker Pros

Rafe Furst
12 min readJan 12, 2016


The Big Idea: Optimal strategy as an early-stage investor is massive diversification behind lead angels & VCs.

Back in the day, poker and equities trading were both games dominated by gregarious showmen with brass balls who followed their gut, honed by many years in the trenches. These days, both fields are dominated by geeks with data who follow optimal game theory, honed by many (computer-simulated) lifetimes in the trenches.

Having been a successful professional in the trenches in poker, artificial intelligence, entrepreneurship and startup investing, I can tell you with 93% certainty that the same thing is happening now in venture capital. That’s right, the industry responsible for disrupting every other industry is now succumbing to its own slow motion meta-arbitrage game.

Okay, the 93% part is made up. But lets look at the parallels between VC, public stock markets and poker to decode the writing on the wall. First we need to understand the classes of players in these professional ecosystems.

Enthusiast: Anyone who does not devote full-time to the profession. There are some very skillful people who make a lot of money as an enthusiast, so this is not a value judgment, simply a description.

Old School Pro: Anyone who uses “back in the day” to mean “better than now.”

New School Pro: Anyone who uses new technology and new thought processes. They recognize that the fundamentals have changed in some important way, and that the “old rules” are giving way to new winning strategies. Some Old School Pros make the transition to being New School Pros, and some don’t.

World Class: Those who have become so good (or so market-dominant) that a different set of rules apply to them than the other players. Or more accurately, they can beat both Old Schoolers and New Schoolers at their own games because they have mastered both realms, and they know when and why to ignore the conventional wisdom, and they know when the math doesn’t apply. They can also play “above the rim” because they’ve gained a reputation as having world-class skill, and because of their reputation, people treat them differently. For instance, people play worse against world class poker pros because they want to topple the champ (and are willing to take unwarranted risks and pay a premium to do so). World class VCs can get into any deal they want, at the terms they want, whereas other investors may not even know about some deals until the deal is closed.

I’m going to ignore both enthusiasts and world class players, and instead focus on the other Old School vs New School. Enthusiasts will be the last to adapt, since this is not their primary avocation; if the market went away entirely, they would move on to something else without too much grief. World class players (Andreesen Horowitz, Sequoia, et al) will figure out the new rules before others and adapt, plus their reputation allows them to play a different strategy than everyone else. I’m also going to ignore tier 2 VCs that invest both early and late. These VCs are only investing early to have optionality on their late stage investments.

This article is for any angel investor who considers themselves a professional, and any VC focused strictly on early-stage deals.

One of the things that happens when software eats an industry is that styles, conventional wisdom and dogma give way to optimal strategies. Optimal strategies are ones that cannot be beaten, no matter how smart or experienced the other players are (or how big their balls are). Anyone who does not employ an optimal strategy will either lose — in the case of a zero-sum game like poker — or will never make the transition to world class in a positive-sum game like venture capital. This is critically important for VCs because you might return a phenomenal 30% IRR to your LPs, but if quant VCs are returning 35% and can show their LPs the mathematics behind their strategy, you won’t be able to raise your next fund.

Some old schoolers will argue (because they always do just before they are disrupted) that it won’t happen to their industry. “There is too much [insert a skill humans are good at] involved here for a quantitative approach to work.” Back in the day, when computer scientists began playing poker and sharing notes online, the professional poker world hardly noticed. Then a few nerds, like Chris “Jesus” Ferguson, began to win world championships. Now poker is all but a solved game. There is still plenty of action around, but you’d be a fool to play against a computer these days. Or even against an 18 year old who has been multi-tabling online for 3 years nonstop and thus has played more hands than Amarillo Slim did in his entire lifetime.

Unlike old school poker pros, there will always be room for traditional VCs, specifically because VC is a positive-sum game. But they won’t make the kind of returns they make currently by investing other people’s money. The people who will make real returns will be investing their own capital, and will look more like this

than this

Some disclaimers. Much of my thinking has been shaped Right Side Capital, a quant VC firm starte by friends of mine. Right Side has yielded a lot of data and analysis that is not generally known or well-understood amongst VCs. And if it is well-understood, it’s often ignored. I am also co-founder of Crowdfunder, where I have direct access to more data on early-stage deals and investment activity than most traditional VCs. That said, I have not done any rigorous mathematical modeling myself, nor am I concerned with proving any of the claims made below beyond the references provided and the rhetorical arguments I am making. If you have evidence that I’m wrong about anything here, please share that in the comments. Finally, neither Right Side nor Crowdfunder are responsible for anything I say here. I am speaking my own opinions, not as a representative of either company.

#Moneyball for #Seed #Investors

  1. You can’t pick winners (nor can anyone) [ref]
  2. The market yields 31% IRR [ref]
  3. To have a 90% chance of tripling your money invest in 400 startups [ref]
  4. Returns are dominated by sub $100M acquisitions (not unicorn IPOs)
  5. Most VCs lose money for their LP investors [ref]
  6. Optimal strategy is to invest early and often (at the same terms as any professional investor)

Let’s examine each of these results and what it means for professionals as the quants move in.

Result #1: Nobody can pick winners

With hyper-efficiency no old school value investor can out-pick the public markets anymore. Mutual funds are dominated by index funds (see here for the empirical evidence), and day traders are dinosaurs. Even the quants (new school) have lost their edge, because the nature of arbitrage is to consume itself.

Venture capital is anything but efficient, so the analogy is not a complete one. But those who hold on to strategies that rely on human pattern matching — once enough data is available to either quants or the crowd — are asking to be dominated.

In the limit, angels and VCs will have to settle for market returns, just as public market value investors do. And that’s good news because…

Result #2: The market returns 31% annually

To put in perspective just how phenomenal 31% is, consider that public equities average 9% historically. Obviously, the past is no guarantee of future, but seed stage returns have been remarkably consistent for the past 35 years, especially as compared with VC as a whole and as compared to the public markets.

As greater liquidity and market efficiency comes to VC, one might expect the private market to become more volatile, and for returns to drop. I predict a liquidity premium of about 5% to be arbitraged away (using historical prime rates for liquidity).

Still, if you were able to invest $50K in a market yielding just 20%, by reinvesting returns over 30 years, you’d end up with $12 Million. As a professional money manager, if you could get 20% gross IRR with low volatility, you’d deserve a pretty fat salary.

Result #3: Market returns are about quantity not quality

In the long run, this is bad for old school angels and VCs, but good for new school. Sourcing, diligence and advising will always be valuable (even critical), but these skills are becoming commoditized, and what was once a differentiator will become a race to the bottom on fees and carry.

On the backs of thousands of old school angels and VCs doing this hard work will sit a layer of quants who simply follow them into deals. Old school will still be doing a dozen deals a year, but they will have the same IRR (with a lot more risk) as the new school VCs doing hundreds if not thousands of deals per year each.

Result #4: IPOs don’t matter anymore

IPOs are all-important for old school VCs because their model is to choose seed stage companies very carefully and invest more as the next unicorn (hopefully) is birthed from their portfolio. For new school VCs unicorns and IPOs are great, but the returns they generate are dwarfed by acquisitions that get done at under $100M.

The JOBS Act now allows anyone to invest at the seed stage (not just accredited investors). It also allows private companies to run public fundraising campaigns. Combine these with the advent of venture exchanges, and I doubt most of the old school VC funds will last unless they adapt to the new reality and thus enter then new school.

Result #5: Most angels & LPs lose money

It may seem paradoxical that the average startup investment triples or so in value, while most LPs lose money investing in the VCs who lead those deals. The paradox is explained by the difference between mean outcome and modal outcome. And the fact that traditional VCs don’t diversify enough to capture the market average.

Angels don’t have to pay the fund load that LPs do, but they are even bigger losers (individually speaking) than LPs because they are so concentrated in deals that will likely go to zero. There’s a reason why proper diversification won a Nobel Prize.

Old school VCs are not being irrational in a context where LPs will pay them to source, diligence and diversify for them. However, this means that VCs must “gamble with LP money” (as Dave McClure says), to try to land in the top quartile and attract LPs. Which is like the poker pro taking a shot at a big game with lots of fish, but without the proper bankroll.

This is great for new school angels and VCs because it means that there is a diverse ecosystem of extremely skillful and experienced VCs off of whom they can draft. The old school argument against this likely future will be that there is a limit to the number of good deals, and those who have access will monopolize them. But the data suggests otherwise, and unless the old school firms raise and deploy ten times more capital than they do currently, they can’t properly diversify into their existing deal flow, let alone into the market as a whole.

The macroeconomic argument that there are plenty of untapped deals goes like this. Half the GDP and 75% of new jobs are created by startups and small businesses with unmet capital needs. The entire angel+VC market is under $100 Billion, whereas the NYSE and NASDAQ alone have $20 Trillion in long-term investment dollars chasing 9% returns. You don’t have to be a quant to realize that venture capital right now is like the taxi industry before Uber.

Result #6: Optimal strategy is to massively diversify behind a lead

In the days before poker on TV there was a wealthy enthusiast who wanted to play in the World Series of Poker Main Event. He wanted the best shot at winning the tournament given his limited skill level. So he asked his friend, who was a professional with a mathematical approach to the game, to come up with a strategy that would be simple enough to execute on without a lot of skill or practice. The pro said, if your first two cards look like blah blah blah then go all-in; otherwise fold.

While not the optimal strategy for no limit holdem, it’s surprisingly close, and has been used effectively for years by rank amateurs to outperform world class pros in major tournaments. (Word of caution: don’t employ this strategy if you want to win money in the long run; your best strategy is never to play against players who are more skillful than you).

My point in mentioning this colorful example is that while the skill difference between an amateur and a top professional may be vast and require 30 years to acquire, the structure of the market/game sometimes leads to optimal strategies which are simple and accessible to anyone. This is becoming the case in early-stage investing.

The keys are:

  1. Coattail off of other people’s sourcing, diligence and deal terms
  2. Massively diversify (> 100 deals per year)

For angel investors, #2 will be the harder of these to achieve; the minimum direct investment for most deals makes the diversification criteria cost-prohibitive for all but the wealthiest individuals. For early-stage VCs #1 is more difficult due the fact that this “dumb” strategy is not something they are used to selling to LPs; plus it’s harder to differentiate themselves as fund managers.

For any would-be moneyballers, the biggest hurdle to overcome will be reducing the overhead of getting deals done (and administering them) at scale. But technology has a way of solving for that. This is why I started Crowdfunder.

Whether or not spray-n-pray behind lead investors is truly optimal I can’t say for certain. There may be VCs that have figured out how to out-pick the market, but the time-horizon for fund results is too long to do a proper analysis.

Paul Graham has said:

“I know of no reputable investor who invests based on data. I once heard of someone who planned to, but I forget who it was; probably nothing came of it…

We are the far opposite end of the spectrum from an analytical approach. We decide based on gut feel after a 10 minute convo. It may seem ironic that we who have the most data make the least use of data.”

The actual irony is that Paul Graham, being the founder of Y Combinator, is one of the most diversified early stage investors in the world. He doesn’t know it, but he’s playing proper moneyball by making gut decisions. After all, the data that he refuses analyze shows that while he can’t do better than the market by picking this way, he can’t do any worse either!

Here’s a way you can earn money from Paul Graham’s gut: go to every YC demo day and make an offer to each graduating startup to invest in them at some reasonable increase in valuation above what YC got. Accelerators like YC tend to earn much better than the market, for several reasons. First, they are in earlier than everyone else and the data shows that the earlier you invest, the higher the IRR. Second, accelerators typically get in a fantastic terms relative to cash-only investors because they give more than cash. Finally, they probably do add extra value with the acceleration (or at least the imprimatur of acceleration). All this adds up to an IRR of somewhere between 40% to 60%. So even if you paid a 50% premium on YC’s terms, you’d likely still do very well.

I’m sure there will be those who argue that a strategy like this will lead to adverse selection: the founders who would accept such an offer are the ones who won’t land funding otherwise. But this is falling into the same logic trap that would have had you pass on AirBNB.

Finally, consider this: there are 300,000 investors each year trying to pick the next AirBNB at the seed stage (see here). Your chances of being one of them are not very good, but if you do get lucky, you will get a 2000x return. Let’s say you invested $25K, you would end up with $50 Million.

However, if you invested that same $25K in a diversified asset class earning 31% and didn’t touch it for 30 years, you’d end up with $85 Million and you could spend your time on better pursuits than trying to predict the unpredictable.

This is one of the reasons why I helped start Crowdfunder… so I can invest not just in the next big thing, but in all the next big things.