When Pattern Recognition Becomes Prejudice

Originally a thread on X/Twitter:

Everyone knows the speed at which VC deals are being done has accelerated to a dizzying pace.

While this can be good for some Founders, it’s magnifying a flaw in the VC ecosystem.

A few thoughts on “Bad Pattern Recognition” and how it’s creating have and have-nots:

2/21: 14 years ago I hung up my operating hat to become a Venture Capitalist. Knowing nothing about investing, I sought out seasoned Investors so that I could learn from their experiences. Borrowing a degree sounded like a better strategy than earning one from scratch.

3/21: It shouldn’t come as a surprise that much of the advice was generic and in the “no duh” camp. It started to feel like many Investors’ diligence processes consisted of evaluating startups on a laundry list of “generally true” criteria. Ticking the right boxes = Term Sheet.

4/21: Things I heard a lot:

Focus on how big a business could be if everything goes right. Invest in serial Founders because they succeed more than first time Founders. Large TAM has to be present to generate large outcomes. Answer the question “is the market ready now?”.

5/21: On and on and on the genericized list went. My conclusion at the time was that making good investment decisions collapsed to an exercise in Pattern Recognition and was told this explicitly by many successful Investors. Invest, analyze, learn, repeat.

6/21: Fast forward 14 years and 150+ funded companies firmwide and I have a very different perspective about what it takes to be a great Investor.

One of the main pieces of advice I would give to the “younger me” is to do the necessary work to avoid bad pattern recognition.

7/21: While this sounds obvious, it’s a concept that’s worth internalizing. The key is to know when to trust previous patterns and when to ignore them which is anything but a simple task. Sometimes art needs to override science and intuition needs to override history.

8/21: I’ve backed many amazing companies over the past 14 years and can say with certainty that a few of the best suffered early on from bad pattern recognition. Funding became easier once they had enough evidence that their businesses were working, but this took time.

9/21: I’ve seen bad pattern recognition around “Location”:

Some Investors won’t fund a company outside of a major tech hub based on the pattern recognition that hiring talent as a company scales is much more challenging in all but a few cities.

10/21: I’ve seen bad pattern recognition around “Team”:

Some Investors won’t fund a first time Founder based on the pattern recognition that first time Founders have thin networks, they need help hiring talent and they don’t have experience raising capital.

11/21: I’ve seen bad pattern recognition around “Investor Signal”:

Some Investors won’t fund a startup when there are other startups in the space that have top tier investors on their cap tables based on the pattern recognition that these institutions crown the winners.

12/21: I’ve seen bad pattern recognition around “Competition”:

Some Investors won’t fund a startup when there are successful late-stage startups and incumbents in the space based on the pattern recognition that the opportunity is limited because momentum is real and winners win.

13/21: I’ve seen bad pattern recognition around “Failures”:

Some Investors won’t fund a startup in a space that’s produced marginal outcomes in the past based on the pattern recognition that outcomes are commonly a result of an industry’s structure.

14/21: In today’s environment, bad pattern recognition is increasing in frequency because most VCs have an endless pipeline of deals to work through but limited time to evaluate them. Speed matters and relying on historical “successful norms” is where triage typically starts.

15/21: The net effect is a “two things can be true at the same time” environment where some Founders are finding it easy to raise capital while others are finding it nigh impossible.

Fit the Pattern = Fast Process + Amazing Terms

Don’t Fit the Pattern = Triage Casualty

16/21: The challenge for everyone in the ecosystem is that the industry’s typical triage process does accomplish two important things. It massively reduces the volume of companies to diligence and it shifts the distribution of outcomes favorably.

17/21: The overuse of pattern recognition helps reduce Type 1 error (i.e. – Funding a business that ultimately fails) but it comes at the expense of increasing Type 2 error (i.e. – Not funding a business that would ultimately succeed with capital).

18/21: Today’s environment feels like cruel and unusual punishment for the diamonds in the rough that have great ideas but don’t fit the normative patterns that survive the industry’s typical triage process.

This sucks because every great opportunity deserves a shot.

19/21: There’s a similar situation unfolding in the public markets where 10 stocks have been responsible for ~25% of the current bull market’s return. It doesn’t mean there weren’t other stocks worth investing in. But finding them has been a challenging sorting exercise.

20/21: This leaves alpha on the table for investors who want to chase it. Figuring out how to spot great opportunities that depart from historical “successful norms” is one way that the next generation of investors will generate outsized returns.

21/21: Investors who do the work to understand opportunities holistically will have an edge because they’re playing a different game than everyone else.

Finding great non-consensus opportunities is hard and takes differentiated skills, but it can produce great results.