Playing “Moneyball for Startups”
There has been a lot of good writing about “moneyball for startups”. Sometimes, “moneyball for startups” is taken to mean a quantitative approach and/or predictive analytics to drive returns in angel investing, venture capital, or both. At other times, it seems to mean making many small bets. There are a lot of good approaches out there, and I believe that if there is a formula for investing in great, or potentially great, startups, it is generally well understood. It often goes something like this: Brilliant or experienced team + 10x better product + Traction/product-market fit + Contrarian investment thesis + Strong distribution/Go-to-market strategy = A successful investment. This is, of course, absolutely right. It probably would be unwise to invest in a startup with any one or more of those conditions missing. However, this doesn’t really do justice to the core concept behind Moneyball, which I believe still has a great deal to teach both entrepreneurs and investors (some entrepreneurs and investors in any event, as I’m sure many out there understand the ideas I will present better than I do).
Before we dive in to what “Moneyball for Startups” could mean, we should step back and examine what Moneyball really is, not in a specific case, but as a subset of a broader class of phenomena. As you may know, the term Moneyball comes from Oakland A’s General Manager Billy Beane, and the deeply contrarian, data-driven approach to building his team, implemented by Paul DePodesta and him in 2002. Beane and DePodesta understood that the market for talent in baseball was largely inefficient, and based on wrong signals. That is, talent was analyzed, bought and sold on a handful of metrics that were not truly indicative of the intrinsic capacities of the players, and therefore had nothing to do with winning. For example, a players’ physical appearance (their looks!) was taken to be important in picking players. Beane and DePodesta saw that this is wrong, and took advantage of that inefficiency to their advantage. In the absence of this approach, wins would tend to accrue to teams with the largest budgets, such as the New York Yankees.
Beane’s approach worked because there were a handful of key conditions present which, when they came together, produced the opportunity set he and DePodesta were able to exploit. These conditions manifest themselves in a bunch of different areas/markets, including:
1. Baseball/Sports (“Moneyball”)
2. Investing in publicly-traded securities
3. Investing in startups/private companies
4. Marketing, including startup marketing, sometimes called “growth hacking”
5. Talent markets
What are the necessary preconditions for the kinds of opportunities exploited by Beane to present themselves? What needs to happen first before we can really play Moneyball, in the realm of startups or otherwise?
I believe we can identify the following features of a system ripe for Moneyball:
1. An auction needs to be taking place, with a lot of potential buyers
2. The buyers need to be bidding on some kind of unknown future event, such as how often a baseball player will get on base or how much money a company will make
3. The bidders need to be able to see or know the other bidders’ bids, or at least their general level of interest (high, medium, low; greed vs fear)
4. The bidders need to have some set of criteria for analyzing the future probability of the unknown future event actually occurring
5. The majority of bidders (or the largest ones) need to be using irrational criteria and/or simply following the herd when making bids
When all of these conditions are present, then we can really begin to play Moneyball. Moneyball is not simply the process of using data to make decisions, but understanding when and how other participants in an auction are failing to use and understand data. What is required is what famed investor Howard Marks calls “second level thinking” — the ability to understand not only what is true, but what others believe to be true but is not. Many readers will rightly be reminded of Peter Thiel’s question in Zero to One — What is true that most people believe is false? Or its inverse, What is false that most people believe is true? Good startups, as Thiel correctly teaches us, are opportunities that most people think are a bad idea, but are in fact a good idea and are therefore priced excessively low vis-à-vis their payoff.
Beane and DePodesta successfully applied this formula in baseball, where they collected a team of players perceived to be misfits and has-beens, which were in fact dramatically undervalued players. Warren Buffett has been successfully applying this approach to buying undervalued, out-of-favor, but fundamentally sound companies for more than 50 years.
What does all of this have to do with startups? Let’s examine how we can apply this framework to two areas of major concern to the startup ecosystem: marketing and investing.
As with any of these systems, startup marketing involves laying out cash now to get a return in the future. It could be short or long term, easy to attribute (as with Facebook or Google Ads) or hard to attribute (as with general brand awareness campaigns in the form of TV commercials). In any case, you want to look for opportunities to acquire users for little to no cost (free is the best price). To get your first several hundred or thousand users, focusing on strategies involving inbound, unpaid marketing may work best. At this early stage, any dollars that you spend will be concentrated on a small coterie of users, thereby raising cost of user acquisition. Over time, as your user base and marketing budget grow, you can spread those dollars out across more users, keeping your CAC well below your CLV. Don’t think of marketing dollars spent as some static campaign — instead, do what Buffett does[i]. Maintain a portfolio of deeply undervalued marketing assets, and “sell” them (in other words, stop spending time or money on a given channel) when they are no longer undervalued. As long as you pump resources into a particular channel, CAC must be below CLV — the most attractive channels will be free or close to free and lead to massive acquisition of customers with a high lifetime value, such as virality. When CAC = CLV, this marketing channel can be said to be exhausted, and therefore extinguished. If CAC > CLV, then you’re overpaying and destroying value.
The hard part about startup investing is that there are no cash flows to analyze. Investing in publicly traded securities is hard enough, because even with a track record of revenue, earnings, and cash flow, coming up with a conservative and accurate picture of a company’s future earnings is profoundly difficult. For startups, this challenge is multiplied many times over. While classes of investors have emerged with risk appetite across the venture spectrum, from seed/angel to late stage, understanding the present value of future cash flows of companies with little to no track record, often operating in new markets, is more art than science. (All investing is, really.)
As a result, investors rely on some version of the formula I mentioned earlier: Brilliant or experienced team + 10x better product + Traction/product-market fit + Contrarian investment thesis + Strong distribution/Go-to-market strategy = A successful investment. The problem is that I believe many, if not most, investors understand this, at least cognitively. That means that startups with all of these features will lead crowded funding rounds, bringing with it risk to investors that seek to participate at ever higher prices. Add to that the strong fear of missing out endemic to the vast majority of investor behavior, and you’ve got yourself a recipe for overpriced funding rounds.
For a VC or Angel to really succeed, it has to do something fundamentally different, and be able to articulate how it is different from, and better than, the crowd. Indeed, most VCs don’t add value (see also Fred Wilson’s piece on this subject). Being “different and better” is truly challenging!
Today, there is one investment firm of which I am aware[ii] that has overtly taken an approach of avoiding the mainstream approach to venture investing: 500 Startups. In his post by the same name (Moneyball for Startups), Founder Dave McClure highlights how he believes the VC/Angel world is missing real opportunity. By focusing too much on traction rather than a clear go-to-market strategy, and over-emphasizing companies in Silicon Valley and New York (or the US in general), most investors miss what McClure sees as the massive opportunity in pre-product/market fit companies and/or those outside the United States. 500 Startups has taken a contrarian approach to startup investing, and I wouldn’t be surprised if it pays off long-term.[iii]
How is a startup investor to be contrarian, and truly play “Moneyball for Startups”? One could look to other geographies or investment criteria, as 500 Startups has done. I also have another idea: an Angel or VC could choose to only look at companies rejected by top tier Angels and VCs. If a company has been turned down by everyone on Sand Hill Road, but it is still able to do well on a number of items in the formula (EG, an obsessive group of founders with deep understanding of a domain, a customer base that is hard to reach, and a flawed product), it could still be a good investment. Collect a portfolio of these unloved ugly ducklings, and you might end up with a (black) swan.
The problem many Angels and VCs face is that returns tend to accrue to top tier firms — VC works only at scale. That means for the majority of VC firms out there, they’re forced to be the industry equivalent of the Oakland A’s, with Sand Hill Road firms acting as the metaphorical New York Yankees and other well-funded teams. That means that this approach, or a general strategy of going for ugly duckling startups, has the potential to work, if your criteria for picking investments is both good and different. The price smaller, pre-scale angel and VC investors will pay is having everyone think they’re nuts, because they’re deliberately avoiding FOMO-driven investments. As with any value strategy, you have to be willing to look crazy for a time, as Billy Beane most certainly did.
The best VCs have been able to do this — they are deeply contrarian, and able to spot and move to the next curve, which always seems fringe and crazy at first, and then obvious and a sure thing later. Fear first, greed next. Even one of the earliest VCs, Arthur Rock, had to deal with people thinking he was nuts.
On raising money for his first investment, Rock said in 2002, “[T]here were a number of companies who were trying to get into more technology and, they had all expressed interest to us that they would like to get into more technology, would we bring them anything in technology we saw. But when we told them that the deal was that they would set up, lend us money and set up a separate company and back it, they said no, they couldn’t do that because they didn’t, they thought it would upset their organizations. So at just about the time we were willing, we were about ready to give up.”[iv] Fortunately they got the money from Sherman Fairchild, and formed Fairchild Semiconductor. Rock and Fairchild’s contrarian investment gave birth to an industry. Later on, when Rock was raising money for what would go on to become Intel, he was able to raise all the money needed even before sending out the 1.5 page business plan to what would become his investors later on. Fear first, greed next.[v]
Rather than going with the crowd, and thereby avoiding the great returns that come from contrarian investments and getting in on crowded, overvalued rounds, Rock showed his truly contrarian colors again by avoiding the internet boom of the late 1990s. “[T]hose prices were ridiculous and I, I didn’t quite understand it. I mean I, sure, I went along in a couple of deals with, you know, friends of mine. But I never, never started a company in that business. The problem was greed.”[vi] As Buffett says, be greedy when others are fearful, and fearful when others are greedy. Easy to understand, but very, very hard to do.
A history of venture capital is beyond the scope of this article, but my guess is that if we looked at all the really successful VCs, at least early on, we’d find a kernel of this contrarian activity. From Arthur Rock to Kleiner Perkins to Y Combinator to Andreesen Horowitz, the best VCs are able to do something really different, and then leverage that success to become an investor of choice for entrepreneurs, and have the best investments come to them. At that point, they’re the Yankees, although they likely started out as the Oakland A’s.
The Big Picture
The big picture here is that in marketing, startup investing, baseball, talent acquisition and so on, you are looking for misunderstood, and therefore mispriced, undervalued opportunities. Moneyball isn’t just about using data, it is about using data to find other people’s mistakes, and then behaving differently. As such, it is as much about having the character to behave differently as it is the data and analytical approach to think differently. You may be evaluating a great company with a near magical product, or a perfect marketing channel, but at a price that is simply too high to warrant investment. It is certainly possible to have a market in which most players are either (1) making data-driven decisions and/or (2) following the bidders making data-driven decisions, thereby driving returns out of that market altogether. The struggle is to relentlessly look for new and different opportunities, look where others won’t or can’t, use data and metrics the crowd ignores, place your bets, and then look foolish for a while.
This is far from the last word on “moneyball for startups” — I’d love to hear what others think, where I may have gone wrong, and how this concept can be built out. I personally find nothing more fulfilling than spotting great opportunities others have missed, be it in the world of startups or as a value investor. I have a long way to go on my journey, and I’d love to learn from your thoughts, feedback, and experience.
[i] Or did, when he ran BPL and had higher portfolio turnover.
[ii] I should add that I don’t have a deep familiarity with the full gamut of startup investors out there.
[v] The round that ended up financing the birth of Intel obviously produced great returns, but it wasn’t exactly an open auction, either. Rock called around and raised the money from associates essentially in no time. Had a massive auction been taking place to fund Robert Noyce and Gordon Moore, it’s not unreasonable to guess that someone would have overpaid, particularly in a frothy capital market environment. For example, Rock raised $2.5 million for Intel, which then raised $6.8 million in its IPO a few years later — assuming those original shareholders sold out, it would have been nearly a 3x. If Rock’s auction had been more open, it (once again) is not unreasonable to assume that price could have been driven so high that making decent returns over a similar time period would have been challenging, or at least less than the 3x achieved. Further still, overpaying would have been the wrong decision for those making it looking forward, whereas in hindsight we clearly know that almost any amount of private money would have undervalued the company, given a sufficiently long term holding period. Hindsight is 20–20, but we have to make investment decisions looking to the future, not the past, so managing the risk of overpaying makes sense.
[vi] http://silicongenesis.stanford.edu/transcripts/rock.htm Emphasis added.
I am grateful to Adam Parrington, Aznaur Midov, and Paul Singh for their thoughtful feedback on this article. Mistakes and opinions are solely my own.