The problem with investing based on pattern recognition

A famous story in artificial intelligence is how the US military developed algorithms to determine whether an image had a tank in it. They used a standard machine learning method: feed the computer a “training set” of photos, some of which had tanks in them and some of which didn’t, and let algorithms identify which features in the photos correlated to tanks being shown.

This method worked for a while but then mysteriously stopped working. Since the features the computer identified were embedded in complicated mathematical equations, no one could figure out what it was really doing and therefore why it stopped working. Eventually someone realized that in the training set, all of the images with tanks were taken on a cloudy day, and all the images without tanks were taken on a sunny day. The algorithms had fixated on the most obvious pattern – the color of the sky. When the algorithm was tested on new photos where the weather varied, it was completely flummoxed.

It is commonly said that good startup investors develop “pattern recognition” that allows them to identify great entrepreneurs and companies. If you look at the hugely successful startups of the last decade, the founders have many similarities that are easy to observe. When they started, many were male, young, unmarried, computer programmers, dropouts of elite universities, etc. As a result, a lot of investors look for founders with these characteristics. But without an understanding of the deeper reasons these founders succeeded, these observable characteristics could just as well be the color of the sky and not the tanks.

At the level of individual investors, pattern recognition can lead to bad investments and missed opportunities. In the context of markets, it can cause companies and sectors with the “right patterns” to be overvalued, and ones with the “wrong patterns” to be undervalued. In the broader cultural context, it can cause large groups of talented entrepreneurs to be denied access to capital.

The classic scientific method provides a better model for investing. Scientists observe data, notice patterns, develop hypotheses, and then test those hypotheses. Pattern recognition is only a step along the way to developing hypotheses about the underlying cause.

Perhaps dropping out of college shows a strong level of commitment. Knowing computer science was probably a necessary condition for starting a tech company in the past, but no longer is. Being young could mean you are inexperienced enough to pursue bold ideas that more experienced people would consider crazy. I am just speculating – I don’t know why these characteristics are common among past successful founders. But the mere repetition of patterns shouldn’t be satisfactory to anyone who wants to understand and predict the success of startups.

Later-stage rounds and “setting the bar too high”

I recently had a number of conversations with CEOs of later-stage startups (generating significant revenue) that went something like this. They want to raise more money, and VCs are offering them money at a high valuation. The CEO is worried that taking money at that valuation will “set the bar too high” and make it difficult to sell the company – if the time comes when he/she thinks it makes sense to sell – at a price that isn’t a significant multiple of that valuation.

These CEOs are worrying too much. VCs know what they are doing and almost always invest with a financial instrument – preferred shares – that protects them even when the valuation is very high. Preferred shares behave like a stock on the upside and a bond on the downside.  The only way investors actually lose money is if the company is sold for less than the amount of money raised (which is generally significantly lower than the valuation).

Here is what the payout function looks like for common stock (for example, what you get when you buy stocks in public markets):

And here is what the payout function looks like for preferred shares:


So, to take a concrete example, Dropbox reportedly raised their last financing at a $4B valuation*. If you think of this as a public market valuation of common stock, you might think this means the VCs are betting $4B is the “fair value” of the company, and will lose money if Dropbox’s exit price ends up being less than $4B.  But in reality, assuming the standard preferred structure, the last round investors’ payout is as follows :

Scenario 1: Dropbox exits for greater than $4B ==> investors get a positive return (specifically, exit price divided by $4B)

Scenario 2: Dropbox exits for between $257M (total money raised) and $4B ==> investors get their money back (possibly more if there is a preferred dividend)

Scenario 3: Dropbox exits for less than $257M ==> investors lose money

If reports are true that Dropbox is profitable and generating >$100M in revenue, then scenario 3 – the money losing scenario – is extremely unlikely.

Will investors be thrilled with scenario 2?  No, but they are pros who understand the risks they are taking.

Going back to the entrepreneur’s perspective, in what sense is a high valuation “setting the bar high”?  In the preferred share payout model, there are two “bars”:  money raised and valuation.  I don’t see any reason why entrepreneurs shouldn’t be as aggressive as possible on valuation, especially if they are confident they won’t end up in scenario 3.

An important point to keep in mind is that, in order to maintain flexibility, entrepreneurs shouldn’t give new investors the ability to block an exit or new financings. Investors can get this block in one of two ways – explicit blocking rights (under the “control provisions” section of a VC term sheet) or by controlling the board of directors. These are negotiable terms and startups with momentum should be very careful about giving them away.


* Note that I have no connection to Dropbox so am just assuming standard deal structure and basing numbers on public reports. I am making various simplifying assumptions such as not distinguishing between pre-money and post-money valuation.


The downside of accelerated investment decisions

There has been a lot of talk about how early-stage valuations have risen dramatically over the past few years. Financially, this is probably good for founders and bad for investors. But a side effect of this frothy market is that financings are occurring much faster. It is very common for investors to get introduced to founders with the proviso that a term sheet will be signed in the next few days. As a result, founders and investors are spending very little time getting to know each other before entering into long-term business contracts.

This is bad news for everyone. Most significantly, founders often give up significant control to people they won’t get along with or even might end up hating. Having bad investors might not matter if the company executes flawlessly and the financing market stays frothy. But most companies have difficult episodes, and the financing market will eventually return to normal. Sadly, founders with bad investors will likely face punishing down rounds, key employees being indiscriminately fired, and elaborate financial shenanigans engineered to dilute founders and seed investors.

“It’s only when the tide goes out that you know who’s been swimming naked.” Warren Buffet likes to say this about investors, but it applies to founders as well. Taking on a new major investor should be treated with the same gravitas as taking on a new cofounder. You can’t do it in less time than it takes to really get to know someone, which is usually weeks or months. Quick financings might seem attractive but are actually fraught with risks.

Allocation investing and the social premium

The rational way to invest in something – a startup, public company, venture capital firm, real estate project, etc. – is to base your decision on an assessment of its fundamental value. The most common way to do this is to try to predict the asset’s future profits. In reality, many of the largest pools of capital in the world – pensions, endowments, and mutual funds – think in terms of “allocations.” This means they start with a model for how to distribute their funds across a set of dimensions, including asset classes, industries, and geographies. This allocation mentality is based partly on prevalent academic theories (the “Capital Asset Pricing Model” or “CAPM”) and partly on the success of certain famous money managers (the “Yale Model“).

Allocation investing has a number of perverse effects on financial markets. For example, in the 80s and 90s venture capital was deemed to be a successful, independent asset class. As a result, many funds decided to allocate some portion of their capital to VC. These pools of capital were so large that they caused the VC industry to grow orders of magnitude larger – many say larger than it should be. In turn, this led to many bad venture investments that drove down returns in the industry (these problems were further exacerbated by the fee structure of VC that encouraged funds to get large and rapidly “put money to work”).

Another perverse effect caused by allocation investing happens in public stock markets when investors decide to allocate a portion of their funds to specific sectors. I recently heard some money managers saying they wanted to allocate portions of their funds to “social media”. Combining this “allocated” demand with a constrained supply (due to the small float of many of these IPOs) can lead to prices that are disconnected from fundamental values. In this scenario, supply will try to match demand, which means mediocre social media companies will go public and non-social media companies will reposition themselves as social media companies or acquire social media companies. They will be chasing the “social premium.”

We saw this happen in the 90s with the rush of companies to reposition themselves as internet companies. In that case, many non-professional investors ended up owning shares in crappy companies when the music stopped. The primary difference now is that the flagship companies like LinkedIn and Facebook have excellent fundamentals. Hopefully this time the market will be discerning and value investing will win out over allocation investing.

Notes on raising seed financing

Last night I taught a class via Skillshare (disclosure: Founder Collective is an investor) about how to raise a seed round.  After a long day I wasn’t particularly looking forward to it, but it turned out to be a lot of fun and I stayed well past the scheduled end time.  I think it worked well because the audience was full of people actually starting companies, and they came well prepared (they were all avid readers of tech blogs and had seemed to have done a lot of research).

I sketched some notes for the class which I’m posting below. I’ve written ad nauseum on this blog (see contents page) about venture financing so hadn’t planned to blog more on the topic.  But since I wrote up these notes already, here they are.


1. Best thing is to either never need to raise money or to raise money after you have a product, users, or customers.  Also helps a lot if you’ve started a successful business before or came from a senior position at a successful company.

2. Assuming that’s not the case, it is very difficult to raise money, even when people (e.g. press) are saying it’s easy and “everyone is getting funded.”

3. Fundraising is an extremely momentum-based process.  Hardest part is getting “anchor” investors.  These are people or institutions who commit significant capital (>$100K) and are respected in the tech community or in the specific industry you are going after (e.g. successful fashion people investing in a fashion-related startup).

4. Investors like to wait (“flip another card over”) while you want to hurry. Lots of investors like to wait until other investors they respect commit. Hence a sort of Catch-22. As Paul Graham says:

By far the biggest influence on investors’ opinions of a startup is the opinion of other investors. There are very, very few who simply decide for themselves. Any startup founder can tell you the most common question they hear from investors is not about the founders or the product, but “who else is investing?”

5. Network like crazy:

  • Make sure you have good Google results (this is your first impression in tech). Have a good bio page (on your blog, linkedin and and blog/tweet to get Google juice.
  • Get involved in your local tech community.  Join meetups. Help organize events.  Become a hub in the local tech social graph.
  • Meet every entrepreneur and investor you can.  Entrepreneurs tend to be more accessible & sympathetic and can often make warm intros to investors.
  • Avoid anyone who asks you to pay for intros (even indirectly like committing to a law firm in exchange for intros).
  • Don’t be afraid to (politely) overreach and get rejected.

6. Get smart on the industry:

  • Read TechCrunch, Business Insider, GigaOm, Techmeme, HackerNews, Fred Wilson’s blog, Mark Suster’s blog, etc (and go back and read the archives).  Follow investor/startup people on Twitter (Sulia has some good lists to get you started here and here).
  • Research every investor and entrepreneur extensively before you meet them. Entrepreneurs love it when you’ve used their product and give them constructive feedback.  It’s like bringing a new parent a kid’s toy. Investors like it when you are smart about their portfolio and interests.

6. How much to raise?  Enough to hit an accretive milestone plus some buffer. (more)

7. What terms should you look for?  Here are ideal terms.  You need to understand all these terms and also the difference between convertible notes and equity.  More generally, it’s a good idea to spend a few days getting smart about startup-related law – this is a good book to start with.

8. Types of capital:  strategic angels (industry experts), non-strategic angels (not industry experts, not tech investors), tech angels, seed funds, VCs.

  • VCs can be less valuation sensitive and have deep pockets but are sometimes buying options so come with some risks (more).
  • Industry experts can be really nice complements to tech investors (especially in b2b companies).  (more)
  • Non-strategic angels (rich people with no relevant expertise) might not help as much but might be more patient and ok with “lifestyle businesses.”
  • Tech angels and seed funds tend to be most valuation sensitive but can sometimes make up for it by helping in later financing rounds.

9. Pitching:

  • Have a short slide deck, not a business plan. (more)
  • Pitch yourself first, idea second. (more)
  • Pitch the upside, not the mean (more)
  • Size markets using narratives, not numbers (more)

10.  Cofounders: they are good if for no other reason than moral support. Find ones that complement you. Decide on responsibilities, equity split etc early and document it.  (Legal documents don’t hurt friendships – they preserve them).

11. Incubators like YC and Techstars can be great.  99% of the people I know who participated in them say it was worth it.

12. To investors, the sexiest word in the English language is “oversubscribed.”  Sometimes it makes tactical sense to start out raising a smaller round than you actually want end up with.