On Due Diligence and Decision Making: Venture Capital

December 8, 2015December 7, 2015 / Nathan Guo / Leave a comment

This week, I am continuing a series of posts on due diligence and decision making across different asset classes. Here’s the link to see the first post in the series on poker as an asset class: Link

Due diligence varies substantially across asset classes, and I believe that different investors are better or worse suited for investment in different assets based on how characteristics of asset class decisions match an individual’s personal preferences. I will continue to use the overall framework developed in the last post in order to analyze the decision characteristics of venture capital (VC) against that framework. I’ll end with analysis of the due diligence performed in VC, and discuss the typical investor traits necessary to be successful in VC.

Framework

Based on my experience, there are 7 decisions characteristics that I believe inform how these decisions are made and subsequently whether you are suited for due diligence of an asset class:

Frequency of Decisions: How often are decisions made?
Distribution of Outcomes: How good or bad can the outcome be? Do outcomes follow a normal distribution or a power law distribution?
Length of Decision Impact: How long lived is the decision? (e.g., venture capital firms might hold a company for a decade, while high frequency traders might hold an equity for a fraction of a second)
Decision Uniqueness: How similar is this decision to other decisions?
Stakeholders: Who benefits or loses from this decision, and how are their motives aligned?
Accessibility to the Public: How easy is it for someone to access this asset class?
Ownership of Outcomes: Who is taking the risk?

Decision Characteristics of Venture Capital

My analysis will center on early-stage venture capital, which is the only segment I’ve had experience in. The later stage you go in VC, the more the asset class looks to private equity, and eventually public equity investing.

Frequency of Decisions: Sourcing decisions happen daily for VCs, whereby they say no to spending time on startups. Actual investment decisions are relatively few, with most partners leading an average of 1-2 investments per year.

Distribution of Outcomes: Venture capital investments exhibit power law behavior, as pointed out by many prominent VCs such as Peter Thiel. I am not privy to VC returns, although some such as the Kauffman Foundation have published copious data on returns to help triangulate. Below is my approximation of how outcomes look in VC as compared to poker for an individual professional poker player or a partner at a VC firm.

There are several conclusions and assumptions embedded in this drawing:

Assumption: The Y axis is not drawn to scale. Professional poker players have hundreds of thousands of hands to draw from, while VCs might reach 100 investments over a very long career. If plotted on the same Y axis, the VC distribution would be barely visible, since there are so few actual investments
Assumption: The X axis is logarithmic
Assumption: The distribution I’ve drawn is a somewhat idealized view of VC that follows a J-curve. Some critics of venture capital including the previously linked Kauffman report have shown that there are a variety of variables that can have strong impacts on return distributions, including things like vintage year, fund quartile ranking, and others. It’s likely that many VC funds out there do not exhibit the positive power law behavior distribution illustrated above, and have instead a somewhat negative normal distribution.
Assumption: These are only the company outcomes, not the VC partner’s actual outcome distribution
Conclusion: While poker exhibits negative power law distributions, VC exhibits much more of a normal distribution on the downside, and a power law distribution on the upside. VCs have relatively frequent, large, negative outcomes due to the fact that so many startups fail
Conclusion: VCs (or at least ideal VCs) have a longer positive tail of outcomes than poker players do, with several orders of magnitude larger positive outcomes ($1B+ vs. $1M+)

Length of Decision Impact: Active VC investors may live with their investment decisions for a very, very long time. Good VCs with a board seat are engaged with their portfolio companies on a day-to-day basis beyond attending meetings once a quarter. The Kauffman Foundation reported many funds have not fully unwound positions more than a decade into the future. Thus, the impact of an investment decision is substantially longer than say poker or sales and trading.

Decision Uniqueness: Decision Uniqueness is extremely high for VC, with each investment having different people, products, and industries. While some VC firms try to specialize to some extent, the decisions still look very different from one deal to the next. It’s incumbent upon VCs to do their own homework and be familiar with a portfolio company’s industry if they hope to have even a chance of adding value.

Stakeholders: VC investments have a very complex set of stakeholders. VCs have both a fiduciary responsibility to the investors (also known as limited partners or LPs), and a fiduciary responsibility to the shareholders of the companies (including founders, employees, and other investors) whose boards they sit on. These stakeholders tend to be more aligned when things are going well, but things can get more complicated if things are going poorly.

Accessibility to the Public: Most VC deals are extremely proprietary, and limited to a small handful of investors. Prior to the adoption of Regulation A+, investing for individuals was limited to accredited investors. Although Regulation A+ has expanded accessibility to venture capital, it’s likely that the best deals will still be proprietary to a handful of VC firms, and mostly inaccessible to the public.

Ownership of Outcomes: Traditional VCs invest someone else’s money, typically money from their Limited Partners (LPs). Although management fees can help reduce VC reliance on carried interest as a salary, Jason Lemkin and others have written about how the capital contributions that General Partners (GPs) are required to put in to help align incentives with the LPs can nullify the salary you get through management fees. This means that VCs have relatively large ownership of outcomes, since they are giving up a substantial amount of immediate cash flow for hopefully large, future carried interest cash flows.

Due Diligence and Investor Traits in Venture Capital

Although venture capital is often compared to poker, there are a number of decision characteristics that dramatically affect how due diligence is performed, and what traits to look for in investors. The relatively small number of investment decisions coupled with the very long 10+ year decisions means that VCs are very “all-in” with individual investments, much more so than a poker player who plays many hundreds of thousands of hands a year. Thus, VCs must be very long-term focused, and willing to be patient and wait for out-sized results far in to the future.

The distribution of outcomes for VC investments also has a substantial impact. Because so many startups ultimately fail, and the fact that 1 or 2 portfolio companies could make or break a fund, VCs really have to swing for the fences with every deal they take. This is in heavy contrast with poker, where professional players are focused on squeezing small edges over a large number of hands. On a day-to-day basis, sourcing tends to be focused on finding reasons to say yes, with a key question being whether or not a startup can make it to the far right end of the power law tail. VCs attempt to filter through hundreds of startups and founders a year to get to 1-2 ultimate investments. As a result, VCs say no to spending more time on startups on a daily basis, often with only a light amount of diligence. While this does mean that they sometimes say no to great deals, this is largely a function of the frequency of decisions and distribution of outcomes.

For those startups that DO make it through the sourcing funnel, VCs perform due diligence in a very different manner than later stage private equity investors. Because the decision uniqueness is so high in VC, there are usually few if any comparable companies or products. This makes it nearly impossible to accurately forecast the future trajectory of the company. This stands in stark contrast to later stage private equity deals, which generally involve companies with tens or hundreds of millions of dollars a year in real revenues and profit. Private equity stage companies have a relative dearth of information available to analyze to assess the potential of an investment. Little data usually exists for many startup markets, and VC diligence tends to be relatively qualitative as a result.

While people of many backgrounds ranging from finance to journalism have succeeded as VCs, I have seen a few common traits among the VCs I’ve met:

Long-term optimism: While some like Founders Fund push the envelope on the pessimistic side, allocation of capital in wait of very long-term, future outcomes requires a high degree of optimism. The way the numbers work out, VC partners are very all-in
People-focused: The challenge of filtering through hundreds of startups and founders a year to invest in 1 or 2, means that VCs speak with tons of people on a daily basis. This task is not onerous given that most of the people you are speaking with are bright, passionate founders, but it’s certainly not for everyone
80/20: Despite only making 1 to 2 investments per year, VCs tend to make very frequent decisions on a daily basis, about where to invest their time. In addition, because the outcomes being chased are so far in the future, it’s near impossible to get 100% of the truth today. Thus, VCs are constantly trying to filter between signal and noise, and they need to be able to get to 80% of the “truth” quickly to decide whether they should continue to invest time in a startup or a founder.

It’s clear that while there is overlap between poker and VC, there are some decision criteria that dramatically change how due diligence is performed, and subsequently who is best suited for each. For more insights specific to VC, check out my previous post, What I Learned About Venture Capital This Summer, which goes more in-depth into who is best suited for VC. Next week, I’ll dive into private equity to attempt to tease out why it looks so different from both VC and poker, despite the fact that VC is technically a subset of private equity.

Please tweet me and let me know if you liked this post or have any comments – I’d love to hear from you!
Tweet to @Nathan_YG

On Due Diligence and Decision Making: Poker

December 1, 2015December 1, 2015 / Nathan Guo / 1 Comment

I’ve spent a substantial amount of my career thus far performing due diligence. While there is a legal definition of due diligence, I tend to use it most in reference to decisions around investment in assets. I’ve always been attracted to the decision making around whether or not to allocate capital, and I’ve seen a variety of decision making processes across different asset classes and decision types.

While I was an undergraduate, I played online poker part-time to pay for school, and continued playing after graduation to provide supplemental income. After graduation and working as a process engineer and project manager for a period of time, I shifted to the business development side and worked on negotiating a joint venture in China, including performing both the financial and technical diligence. Since then, I’ve worked in venture capital and in private equity consulting performing commercial due diligence – primarily assessing product-market fit, market growth and size, and competitive positioning.

Due diligence varies substantially across asset classes, and I believe that different investors are better or worse suited for investment in different assets based on how characteristics of asset class decisions match an individual’s personal preferences. To analyze poker as an asset class, I will first layout my overall framework and analyze the decision characteristics of poker against that framework. I’ll end with analysis of the due diligence I performed in poker, and discuss the typical investor traits necessary to be successful in poker.

Framework

Based on my experience, there are 7 decisions characteristics that I believe inform how these decisions are made and subsequently whether you are suited for due diligence of an asset class:

Frequency of Decisions: How often are decisions made?
Distribution of Outcomes: How good or bad can the outcome be? Do outcomes follow a normal distribution or a power law distribution?
Length of Decision Impact: How long lived is the decision? (e.g., venture capital firms might hold a company for a decade, while high frequency traders might hold an equity for a fraction of a second)
Decision Uniqueness: How similar is this decision to other decisions?
Stakeholders: Who benefits or loses from this decision, and how are their motives aligned?
Accessibility to the Public: How easy is it for someone to access this asset class?
Ownership of Outcomes: Who is taking the risk?

Decision Characteristics of Poker

My analysis will center on Texas Hold’Em since this was the game I played the most, and the game most familiar to most people. Most of the numbers referenced are from my own recollection of approximately 750k hands played in my early 20s. Graphs and figures come from a more recent set of ~150k hands played part-time during my MBA.

Frequency of Decisions: At 8 tables for hour, ~650 hands/hr * ~1.5 decisions/hand = ~1K decisions/hr or more than 1 decision every 4 seconds

Distribution of Outcomes: Poker has extremely high variance, with outcomes exhibiting power law characteristics rather than normal distributions. Below is a histogram showing the distribution of outcomes for individual hands I experienced over about a 150k hand sample size:

Poker exhibits a few small outcomes very frequently and many extreme outcomes very infrequently. It also exhibits both positive AND negative power law behavior. Due to the extremity of outcomes, it’s difficult to tell from this picture alone whether these are outcomes of a winning player. Below is what the outcomes in the above chart look like over time, and not as a histogram.

Length of Decision Impact: The effects of a decision in poker could be felt for a very long time if you are playing riskily and for example putting your entire bankroll on the table. For most poker professionals, the impact is close to 0 seconds assuming they are following appropriate bankroll management and not putting a substantial portion of their bankroll on the line. This contrasts strongly with asset classes like private equity and venture capital, where the impact of decisions are extremely long, and you must live with an investment decision for 5+ years.

Decision Uniqueness: Based on the ability to make a decision every 4 seconds, the vast majority of decisions are repetitive. For a 6-max tight aggressive player, approximately 75-80% of hands are folded after the hands are dealt, regardless of any prior action. The complexity of decision uniqueness in poker is that:

There are more possible arrangements of a poker deck than there are molecules in the known universe
Decisions that occur late in a hand (e.g., calling all-in on the river) tend to involve substantially more chips than early-hand decisions

Thus, there is low decision uniqueness for a large percentage of decisions (such as deciding whether to fold a hand before the flop). However, there is extremely high decision uniqueness for a small percentage of decisions, that also tend to be for very high stakes. This bifurcation of decision uniqueness isn’t completely unique to poker – a reasonable comparison might be a venture capital firm specializing in online marketplaces. This firm sees high decision uniqueness in individual deals with different founders, products, etc., but low decision uniqueness across the body of the investments they look at, which are all online marketplaces.

Stakeholders: Poker is a negative sum game, where the house takes a rake and you play against other players. To break even, you must be better than average in the field to win money from other players. No matter what, someone is losing. As a result, poker tends to be a very individual game. The closest players come to having a team dynamic are through networks of poker player in forums, Skype groups, and real life relationships built to analyze hands, mutually improve performance, and diffuse emotional stress from the variance of the game.

Accessibility to the Public: Live poker is relatively accessible for Americans over the age of 18 near a casino. Online poker has been inaccessible in the U.S. since the UIGEA was passed in 2006. This substantially limited the ability of many poker pros to continue their careers and caused many to leave the country or seek other employment.

Ownership of Outcomes: Some professional players sell pieces of their play (essentially equity in the outcome of the session) to help lower variance/risk. Outside of these scenarios, poker players are generally playing from their own bankroll similar to a proprietary trader.

Due Diligence and Investor Traits in Poker

Due diligence in poker varies substantially by player – many fish never really think critically about their decision making, while poker pros might pore over a hand for days or weeks. The standard for most poker pros is to balance their time between both playing the game, and studying the game. This means that a very large fraction of hands end up having little to no analysis performed.

The combination of the fact that outcomes are power law and decision uniqueness is both low for starting hands and high for hand outcomes informs much of how the game is analyzed. The study of poker itself tends to be bifurcated, with some time dedicated to in-depth analysis of individual hands with large outcomes, and other time spent on analyzing overall game theory that affects the long tail of low decision uniqueness hands. Because poker has extreme outcomes and high decision uniqueness, it’s frequently difficult to identify concrete right and wrong answers in detailed hand analysis. As a result, a common mantra heard among poker pros is to not be results oriented. Substantial effort is made to achieve a focus on the decision making process itself, such that results are not provided upfront in the problem statement, in much the same way that business school cases tend to hold back the result until after the case.

Edges in online poker are small and getting smaller, so winning players are pushing to make 1% more decisions correctly. I recall a thread that appeared once on Two Plus Two, the online forum where the best poker players congregated a decade ago, in which many players posted results of a Myers Briggs personality test. Easily 90% of the players tested as some variation of INTJ or INTP. At the end of the day, online poker is a game that rewards the same skills that proprietary and high frequency trading firms select for: high analytical logic, comfort with ambiguity, and high risk tolerance.

Although there are certainly shared decision characteristics between poker and more risky asset classes such as venture capital, they are different enough that diligence and investor traits are substantially different. A major difference in decision criteria that emerges immediately: while poker, trading, and venture capital all exhibit power law returns, venture capital does not exhibit negative power law returns while poker and trading exhibit both positive and negative power law returns.

These nuances and more next week, when I analyze due diligence and investor traits in venture capital.