Crypto asset valuation
Ecosystem overview for data scientists
Note: nothing herein is financial advice. This article is made only for informative purposes, not to be used for investment decisions.
Key Points
- Crypto protocols with their native assets provide an open and data-rich asset class with venture-scale economics and advantages of public market liquidity
- We still don't have established metrics for long-term valuation which provides a window of opportunity
- There's an active area of research in modeling crypto protocols as complex systems
- Decomposing the value chain enables an increased number of iterations on a specific idea maze. This speeds up the collective discovery of a product-market fit.
- Investing in the asset class provides asymmetric opportunities
Introduction
As a data scientist, it’s hard not to be excited about the open nature of the crypto ecosystem. Open knowledge and open data imply a trove of insights to be found that can guide our investment decisions.
At the very least, it’s worth diving into it because of its multi-disciplinary approach to how we can change how we coordinate human activity. Not only we can learn about how our human systems can be improved, but we can also about how and why they function the way they do.
The mainstream's interest largely follows the recent price trends. In this piece, I go beyond the short-term price speculation of crypto prices and explore the various aspects of how to look at crypto asset valuation with a long-term perspective. This is not meant to be a deep dive into any of the topics, more of a high-level overview of the ecosystem. For a deeper dive, I provide sources with the mentioned concepts.
Investment analysis
Most of the data science related articles about crypto markets focus on technical analysis to predict the future price. If the historical patterns show a future increase/decrease in price, this provides a buy/sell signal. Meanwhile, fundamental analysis focuses on the estimation of the intrinsic value of the asset. If the estimated intrinsic value of an asset is higher/lower than the market price, this provides a buy/sell signal.
Why is this distinction important for the crypto markets? They are still immature and most of the investment (trading) is speculative and will stay this way as long as real-world problems are not solved with the technology. Market developments will enable a similar valuation of assets as it is done in other, more mature asset classes (stocks, commodities, FX, etc.)
Why is technical analysis more popular? The most obvious answer is the availability of the data. Even if the blockchain data is openly available, it still needs to be parsed in a meaningful format. The price information is easily accessible. Not only that, we can apply the same (or similar) approaches for analysis as any other asset class. It can be modeled as a supervised machine learning or a deep reinforcement learning problem. Or we might go with technical indicators, transferring knowledge from traditional finance. To build an LSTM neural network on the underlying prices, we don't need any domain knowledge apart from knowing how to optimize the model architecture (which is a complex, but different problem). With that, the focus is more on out-smarting others (crowd psychology) instead of evaluating the underlying value of the asset.
Meanwhile, fundamental analysis depends on domain knowledge and it's hard to find an area that spans across more areas than crypto. It comprises economics, game theory, cryptography, and distributed systems, among others.
The fundamental analysis can be decomposed into two parts:
- estimation of the future market potential of solving a business problem
- estimation of the probability of individual solutions to capture investment returns for solving it
How individual solutions succeed at solving a business problem over time is an evolutionary process of what Balaji Srinivasan calls the idea maze.
Idea maze
A historical non-crypto example of an idea maze is a social networking product. If we were to estimate that the future total addressable market (TAM) will be large, how would we distribute the probability of success across different alternatives? Would Facebook get as much attributed probability as its percentage of the total addressable market (TAM) is today? Or would MySpace or Hi5 be seen as having more future potential? Or maybe we would conclude that another product will capture the market that hasn't even been launched back then yet? How well is the current market leader solving the business problem and how strong and valuable are its network effects? These are some questions that might be worth asking.
This might be a good intellectual exercise, but mostly useless. There are two reasons for this: a lack of deal flow and closed data. The first reason is that most people (retail investors) don't have access to invest in an early-stage business. This is reserved for angel investors and VCs as they have both the funds and incentive to hold assets illiquid for a long period for a small probability of a large upside. The second reason is that the metrics with which we could evaluate the business were not open to retail investors. These companies operate in private markets so they don't have an incentive to share their usage metrics. As more and more economic growth comes from tech companies and they take longer and longer to go public, this provides a difficult challenge: not much return left for retail investors to capitalize on.
In crypto, many talk about the "killer dApp"(decentralized application) that will bring crypto technology to the mainstream. Meanwhile, there are still crucial challenges that need to be overcome which relate to UX (such as on-ramp) and scalability of the computational throughput.
There have been a rise in decentralized finance (DeFi) applications in the last year which have $1B total value locked (TVL). The value is transparent, and we can see how the value is distributed across different projects. Meanwhile, this is a fraction of the current $300 billion total industry market cap.
Future development depends on many unpredictable factors. And as always with technology, timing matters. Certain solutions to the idea maze might not be successful today and succeed in 10 years from now. Regardless, one thing is clear: as the industry's market cap increases, so will the mainstream's interest in its returns. So without speculating about the actual killer dApp, it makes sense to explore one obvious idea maze that crypto has multiple unique approaches to tackle: social investing.
The use-case for solving the idea maze of social investing is to provide a platform to connect people good at investing and the people that don't have the resources (time/energy/skill) to invest actively themselves. That's not to say that we can't build solutions that are not based on crypto technology. eToro's CopyTrader does it. Meanwhile, just like the internet democratized the spread of information, crypto can democratize the creation and management of financial products. DeFi enables the creation of programmable, composable and interoperable assets and we already have working examples of that. Stablecoins and lending protocols built on top of them already have multiple variations. So does social investing.
Set Protocol built a DeFi alternative to eToro's product and added many additional features. Melon, another DeFi solution, focuses on building decentralized on-chain asset management. At the moment, hedge funds pay large amounts for administrative tasks that might be encoded in smart contracts and automate costly workflows with software. With DeFi, it will be possible to set up a crypto fund in minutes for a fraction of the cost, regardless of location. ICONOMI is another provider with centralized finance (CeFi) solution that enables the creation and following of different crypto strategies.
These are some examples of exploration of the social investing idea maze. Markets are often ruled by winner-take-all dynamics because of network effects, and that's also a source of the long-term crypto protocol moat. That's not to say that this will happen here. These products are targeted at both separate and intersecting use-cases.
Metrics
Over the last two decades, software businesses have changed how we provide services across many verticals. With books like Lean analytics and frameworks like AARRR, we now have established models of how to think about SaaS business valuation. We understand both the value of tracking and improving the growth funnel. We now know understand that retention matters and the importance of unit economics.
One of the early lessons in building startups was that a lot of the metrics that we might think are important are just vanity metrics. The number of views on a page or the total number of customers doesn't give us much valuable information about the actual performance of the business. There is little to none actionable insight, only a false signal of success.
How does this relate to crypto? We don't have established metrics of success yet. Crypto networks are complex and interlinked. They operate on different layers, built on top of each other. Just like the strength of a chain is determined by the weakest link, the security of the upper protocol layers is determined by the security of the lower ones. And the higher up the layers we go, the more user-facing the protocols become.
In DeFi, the interlinked nature of crypto protocols is described as money legos. By decomposing the value chain, we might see an increased number of iterations on a specific idea maze. With that, speed up the collective effort for finding product-market fit with crypto applications.
Besides that, there is a distinction between crypto protocol creating and capturing value via its native asset. All the internet's economy today is built on top of TCP/IP protocol. Yet its creators didn't capture any portion of that value that was created on top of it. Certain protocols are built on top of other protocols that (at the moment) couldn't scale with a large increase in the number of users. Sometimes the protocol's network is not secure against malicious attacks (attack vectors) or is not sufficiently decentralized. This is known as the blockchain trilemma that crypto networks need to tackle in the trade-off between scalability, security, and decentralization.
There are different views on how crypto protocols should function. One view is that they are just new kinds of businesses whose aim should be to optimize their success metrics (e.g. revenue). That implies that they should be evaluated on the same grounds as stocks: their business fundamentals. Another view is they only provide the systems of logic to coordinate between different actors in the network. As coordinators, they should minimize the value they capture, not maximize it (unlike businesses).
Besides that, crypto projects have unique challenges to tackle. Product/market fit, community participation, and sufficient decentralization are like three legs of a stool. They all need to function for a crypto project to succeed.
There has also been a lot of involvement in defining meaningful crypto metrics. Crypto protocol's value increases with the number of participants because of network effects. Metcalfe's law quantified it for telecommunications networks such that the value of a network grows proportionally to the square number of participants. Another metric that captured a lot of the ecosystem's attention is the equation of exchange (MV = PQ) taken from monetary economics. It implies that the crypto asset's valuation can be estimated by solving for M (size of the monetary base) which grows with both the increased quantity of digital resource (Q) and its price (P) and decreases with the increase of the asset's velocity (V). Other more crypto native metrics include NVT and INET. Chris Burninske, a pioneer thinker in the space, wrote both an early post and a book on this topic.
Slowly certain practices will become established and taken for granted. Just as web developers today need not think about pointers (low-level programming abstraction) when developing web apps, future blockchain developers won't need to think about many of the complexities of developing blockchain applications today.
Simulating protocol behavior
The design of a purpose-driven token brings challenges that arise from its goal: to build a self-organizing system. It exhibits behavior that cannot be predicted by the sum of its parts.
Consider the study of planetary motion. A high schooler can calculate the motion of a planet around the sun using pen and paper, while the three-body problem lacks any closed-form solution.
Complex emergent behavior needs to be taken into account when analyzing crypto protocols. cadCAD is a Python library that enables modeling by simulating the impact that a set of actions might have on a system. It can answer the "what if" of the system design and analysis. This paper provides a good technical introduction to the topic.
Why does this matter to valuation? Because for a system to be effective and reliable, we need to understand how different inputs affect its output. The last thing we want is to either use or invest in a protocol whose behavior under different parameters is a black box to us.
Domain knowledge
It's clear by now that knowledge about crypto spans across many domains. How to keep up with the evolution of the ecosystem?
Unlike most industries, the crypto community's members often share their knowledge and views with the public.
It would be unfair to not start with blockchain developers. Vitalik Buterin (creator of Ethereum) is one of the greatest thinkers in the space. Meanwhile, his thoughts might not directly apply to a data scientist. His writing might be more useful to deepen the knowledge of deeper technical aspects of protocol design and industry trends in the next 5–10 years.
Venture capital (VC) firms that specialize in crypto are also often more active in writing than those who don't. Placeholder, a16z crypto, Outlier Ventures, and DragonFly write on how they see the progression of crypto and also share their thoughts on why they invested in certain crypto projects (their investment theses). For a data scientist, the main value is gaining a better understanding of both value creation and value capture, combined with different valuation mental models.
Crypto analytics companies often write about the insights that their platforms can uncover from the mountains of available data. These might be the most relevant for a data scientist. IntoTheBlock writes in-depth behavioral analyses of different protocols.
Other similar companies include Santiment, Alethio, Dune Analytics, and TokenAnalyst. These focus on analyzing observational data. The open nature of blockchain data enables insights that can't be gathered for any other asset class.
Companies like Gauntlet and BlockScience (authors of cadCAD) focus more on the simulation of protocol behavior using agent-based simulations. The former recently published a research report on the financial risk of Compound (DeFi application).
Conclusion
As a retail investor, investing in crypto assets is possible at earlier stages than it is the case with traditional tech companies. The projects’ metrics are open and investments are liquid (no lock-up period). Or put differently, crypto projects enable venture-scale economics with all the advantages of public market liquidity.
This enables investors to capitalize on asymmetric risk (large upside, limited downside). This especially applies to data scientists who can use their skill of data-driven decision-making leveraged by business and technical knowledge.
This knowledge will increasingly be monetizable as the industry matures and grows in its market cap. We will establish valuation frameworks and become better at analyzing and simulating the emergent behavior of crypto protocols. But as always in investing, being non-consensus right matters most, as that's where the returns are.
