How to reduce adverse selection
“If we avoid the losers, the winners will take care of themselves” - Howard Marks, Oaktree Capital founder
Howard Marks is known as much for founding Oaktree Capital, one of the largest alternative asset managers in the world, as he is for his “memos” where he shares his investment philosophy. A core theme of his writing is his focus on risk control and the avoidance of losses, rather than positively seeking outsized returns.
By way of contrast, the caricature of the venture capitalist is the swashbuckling risk taker, recklessly seeking out returns. Yet, as Michael Moritz points out in Nate Silver’s On The Edge, successful VCs prioritize reducing risk in investments rather than seeking out wild risk. Moritz advocates for disciplined risk management, guided by rigorous analysis and a deep understanding of market dynamics, thereby protecting the downside and maximizing the potential for success.
We are going to look at how SignalRank manages risk, both by reducing the number of losses and by maximizing our opportunity to see the entire of the top 5% of Series B distributions.
What is adverse selection?
Adverse selection describes where information asymmetry leads to imbalanced and potentially undesirable outcomes. The party with less information unintentionally selects lower quality investments because they lack the information to properly assess risk and/or quality.
This occurs in VC at two levels. One is the fact that founders know more about their businesses than the VCs ever will. Proper diligence & networking can reduce this risk for VCs. The larger adverse selection problem is that VCs with less powerful brands than say Kleiner Perkins or Sequoia may never even see power law companies. This is ecosystem level information asymmetry (which applies as much to LPs looking for stellar GPs as it does to GPs seeking the next unicorns). Upfront’s Mark Suster has the great piece where he asks himself “why did I get so lucky” to see a particular company – is he just a victim of adverse selection?
In truth, VC is the game of getting the best founders to pick you, not the other way around. This is how VCs can overcome adverse selection, leading to advantageous selection. Hence the focus on VC brand building. This magnetic force powers the best VCs in the world, from Sequoia to A16Z to Founders Fund. This leads to self-fulfilling prophecies which we have talked about before here.
SignalRank does not have a brand or its own track record, yet. So our main focus today is on reducing adverse selection. Let’s look at how we achieve this.
Playing The Loser’s Game
At SignalRank, we believe that the Marks memos are a more useful treasure trove for a VC than appears immediately obvious. In particular, his focus on improving returns by avoiding the losers (see Figure 1). He particularly likes to cite a 1975 article by Charles Ellis called “The Loser’s Game” as being seminal in his development as an investor. No apologies for quoting this in full:
Charley’s article described the perceptive analysis of tennis contained in “Extraordinary Tennis for the Ordinary Tennis Player” by Dr. Simon Ramo, the “R” in TRW. Ramo pointed out that professional tennis is a “winner’s game,” in which the match goes to the player who’s able to hit the most winners: fast-paced, well-placed shots that his opponent can’t return. But the tennis the rest of us play is a “loser’s game,” with the match going to the player who hits the fewest losers. The winner just keeps the ball in play until the loser hits it into the net or off the court. In other words, in amateur tennis, points aren’t won; they’re lost. I recognized in Ramo’s loss-avoidance strategy the version of tennis I try to play. Charley took Ramo’s idea a step further, applying it to investments. His views on market efficiency and the high cost of trading led him to conclude that the pursuit of winners is unlikely to pay off. Instead, you should try to avoid hitting losers. I found this view of investing absolutely compelling. I can’t remember saying, “Eureka; that’s the approach for me,” but the developments over the last three decades certainly suggest his article was an important source of my inspiration.
Because of his conviction that markets are efficient, Charley recommended passive investing as the best way to end up the winner – let others try the tough shots and fail. Oaktree’s view is a little different. Although we believe in the existence of inefficient markets as well as efficient ones, we still view the avoidance of losers as a wonderful foundation for investment success. Thus we diversify our portfolios, limit the fundamental risk we’ll take, try to buy things that provide downside protection, and emphasize senior securities. We, too, try to win by not losing.
Figure 1. Howard Marks’ risk/return chart based on different investment styles
But isn’t VC about finding the power law companies that can deliver 100x? Don’t VCs talk about how you can only lose your money 1x? Well, yes.
At SignalRank, we are playing a slightly different game. We are seeking to turn power law distributions into distributions that closer approximate a normal distribution. We are seeking to index venture capital such that our returns at least outperform the mean (not median) of the venture capital asset class. We are offering a smart beta or “alpha capture” index product to venture.
In essence, our model achieves this by significantly reducing the zeroes in our portfolio. Our algorithms are very good at predicting true negatives (Series B companies which do not deliver 5.0x MOIC from Series B) with 87% accuracy. Our algorithms are NOT predicting a company will be the next Databricks / Uber / Looker. By eliminating companies which have a lower probability of success, we create a pool of assets which have a higher probability of becoming the next generational companies. In other words, like Howard Marks, we remove the losers.
Our model optimizes for high precision. This model increases the probability of identifying a company which can deliver 5.0x MOIC from Series B to 30%+, compared to the market average of 10%. But we do miss about 50% of all $1bn+ companies with this model.
If you are playing this “negative” game, you have to ensure that you are seeing the full distribution of remaining potential outcomes. With a power law skewed asset class, this is particularly important. You could otherwise cut off both ends of the distribution to deliver sub par returns. You have to have the opportunity to see the next Stripe.
Our algorithms cut off what we believe are the bottom 95% of the Series Bs distribution. The focus of the rest of this post is on how we seek to ensure that we see the full distribution of the top 5%. Or how do we reduce adverse selection.
How we reduce adverse selection
1. Generous profit share
Our model is to finance the pro rata of seed manager’s who do not have sufficient capital to continue investing in their best companies at Series B.
We offer generous 20% deal by deal profit share with our seed partners. If we invest $1m on behalf of a seed investor, we pay our partner $1.8m for a $10m exit (20% of $9m profit). How can we do this? Why do we do this?
We can do this because we are an index with a corporate structure whereby we reinvest exit proceeds into the next cohort of investments to compound returns for our shareholders. This should deliver outsized returns to our shareholders, with our management team incentivized by the overall net increase in NAV over time. This is different to a traditional 2/20 fund (where if we offered 20% carry to our seed partners, there would be zero for our GP).
We offer this 20% carry because we want to reward our seed partner for the work they have done finding and investing in the company. This ensures that a partner shares with us all of their Series Bs, and especially their best investments. We encourage our partners to share dealflow with their LPs first. We want to be the next call. We then rely on the algorithms to eliminate the bottom 95% of Series Bs and focus returns for our partners and ourselves into the best prospects.
We are also big believers in the power law. We would prefer to own 80% of an asset than can deliver 100x than 90% of an asset that can deliver 3x. We want to have the opportunity to see the power law potential Series Bs.
2. Speed
Speed is a weapon. Entrepreneurs want a round to close yesterday, so they can go back to building. Lead investors also weaponize speed because 1) they don’t want a competitor to trump their term sheet and 2) as long as the company is sufficiently capitalized, they would prefer pesky existing investors did not participate with their pro rata as this impacts ownership targets.
We can move quickly with our approach, as we rely on the our quantitative models for selection. Our diligence focuses on confirmation of facts. We only invest where a world class Series B investor is participating in the round afresh (ie no insider rounds), where world class seed & Series A investors have already participated.
We also think we offer an intellectually honest approach to VC. We do not claim to offer any operational value to our companies. Our seed partners and the Series B leads can deliver that in spades.
Our pace empowers seed managers to defend their pro rata with gusto. We can deliver capital in a week. This ensures that we can see the fast moving rounds.
3. Pro rata access point
In the top 5% of Series Bs, there should not be space for a new Series B following investor to access the company. These should be competitive rounds where everyone has sharp elbows. In fact, if there is space for a new following investor, this is potentially an adverse selection flag. New investors in Series Bs should be asking themselves – why am I so lucky?
Our pro rata access point enables us to support our seed partners in the most competitive and most sought after Series Bs globally.
4. Price takers
We only invest where a world class set of humans sets the price of a round. This is how indices work (hence the moral panic about what happens when the public markets trend towards 100% passive investing – who is then pricing assets?).
One of the big mistakes made by Tiger during ZIRP was pricing rounds. You can’t index and set prices. We liked Tiger’s “indexing” strategy but believe that setting prices as an index distorted the markets and potentially worsened returns. We want, nay need, price setters to be alpha-seeking VCs.
There is however a paradox with valuations. We follow world class alpha-seeking Series B investors. They know Peter Thiel’s adage that the fastest growing companies are the most undervalued (even if they appear overvalued in the short term). Price agnosticism reduces adverse selection further. This is particularly important when it appears that each technology cycle increases the potential size of outcomes by an order of magnitude – OpenAI could be the first of many trillion dollar AI-native private companies. In other words, entry price valuation can become more flexible.
Similarly we are sector & geography agnostic. Our model will dynamically follow the trends of the best VCs in the world.
5. Track record
It is possible for a VC to move from reducing adverse selection to maximizing for positive selection. Building a track record & brand is critical to this path.
We started investing in May 2023. We have already made 20 Series B investments (making us the #2 most active Series B investor globally after A16Z into high quality Series Bs). We have seen two unicorns (Chainguard & Saronic) plus two additional mark-ups. Lots to do but a good start. Cue the irritating “we’re just getting started” phrase.
By developing a track record, we expect to see ever more companies in the top 5% distribution, as we start to be considered a valuable component in the ecosystem. We want to work with more seed partners to finance their pro rata in ever more companies in the top 5% of Series Bs. In fact, in due course, we would like to see that the ecosystem perceives a “SignalRank index company” as a stamp of approval.
Conclusion
Howard Marks cites Yale’s David Swensen as arguing that investors need to have “uncomfortably idiosyncratic portfolios” to generate returns for LPs. SignalRank has a differentiated indexing strategy relative to most VC strategies.
We also recognize that many people believe indexing is not how VC should be played. VC is about backing the crazy founders to drive innovation. Hell, indexing may even by un-American – what happened to being the home of the brave?
Vanguard encountered similar pushback when indexing was kicking off in the public markets. See Leuthold Group’s poster in the 1970s (supposedly a joke by Leuthold employees).
Figure 2. Help stamp out index funds!
Yet 2024 saw passive investment strategies overtake active investment strategies for the first time in terms of AUM. This is because low cost indices have consistently outperformed most active managers in the public markets.
We are seeing a similar story play out in the private markets. By reducing adverse selection and avoiding losses, SignalRank aims to deliver low cost, scalable & consistent investment products that offer investors access to the next cycle of generation defining private companies.