The ICO Model

Initial Coin Offerings (ICO) had a breakout year in 2017, raising over $3 billion. With the Telegram ICO promising to raise well over $1 billion, 2018 will continue and accelerate this trend. But while the money is flowing in, controversy and uncertainty surround ICOs. Critics say that ICOs are scams at worst, or a tulip bubble at best. Even investors who see the value of ICOs are unsure how to value token sales, and how to distinguish promising projects from wannabes.

In this post, I’ll provide a framework for understanding ICOs, including the basis of their value, and how to price them. The goal is to help investors make sense of the ICO landscape, and to make better investment decisions.

From IPO to ICO

It’s useful to compare an ICO to the traditional Initial Public Offering (IPO) model, and see ways in which they are the different (or the same). The IPO model is fairly intuitive: you take a company, and divide into millions of tiny fractions called shares. Each share is entitled to its own fraction of the company’s profit, and the company’s profit in turn determines the value of the shares. This profit can be current profit/dividends in the case of a mature company like GE, or the expectation of future profit/dividends in the case of a growing company like Amazon.

The P/E ratio is a common technique to value companies in the IPO model, where P is price per share, and E is earnings per share. You have P/E = k, where k is some constant value. A mature company will often have a “k” value of 20, which just means that it generates a profit each year equal to 5% of the share price. We can solve the equation for P to get the pricing formula for the IPO model:

P = kE

That is, share price equals earnings times a constant. This is easy to understand, and has an implication for the kinds of companies that are attractive for an IPO. Companies looking to IPO can’t just generate revenue, they have to generate a profit. To see the importance of this, compare giant American companies: Walmart and Microsoft. Walmart has 5x the revenue of Microsoft, but it’s value is half of Microsoft’s. The reason is simple: Walmart profits are half of Microsoft’s, and profits, not revenue, drive stock prices.

Utility Token Basics

Before we do a deep dive into ICO pricing, let’s review some basics about how they work. ICOs don’t sell shares of a company like IPOs do. Instead, they sell what are called “utility tokens”: digital coins that consumers will use for some sort of service. FileCoin, for example, sold tokens that can be used to buy disk storage, while TensorCoin is creating tokens that can be used for AI computation. A token doesn’t grant you access to profits, but access to a service.

Proponents of ICO’s base the value story on usage. If there is a token for X, they argue, and the usage of X goes up, then the token’s value must also go up, because of all the demand to buy tokens and make use of X.

Critics of ICO’s have a counter argument. That’s all well and good, they say, but there is a critical mistake: people won’t want to hold X. With the advent of instant exchange services such as ShapeShift, people will hold all their money in a popular cryptocurrency like Bitcoin, and instantly convert to/from utility tokens when they need them. The conversions to/from each utility token will cancel out in terms of buys/sells, and the value of utility tokens for X will be near zero, even if X is extremely popular.

Who’s right? Well, let’s dig into the details and find out.

The Basis for Token Value

Let’s start with a simple example, and work our way up from there. Imagine we have a token called StorageCoin, which is used to pay for disk storage. The developers of StorageCoin have made a network for providing storage, and they’re ready to launch it. The mint a billion StorageCoins (STC), and make them available in the open market. What will be the value?

Well, along comes Alice. She has a terabyte of data and she wants to store it somewhere. Let’s suppose the market price for storing that terabyte of data is $100. Alice, because she’s an early adopter, wants to store her data using StorageCoins, and she needs to buy some STC to do that. But how much? After all, the StorageCoin network just launched, and there’s no bid history, and no one has ever used it before. Alice decides to go for all the marbles: she puts in a bid of $100 for all one billion STC. The StorageCoin founders, who need some money to pay their cable bill, get excited and sell all one billion STC to Alice.

So now Alice has all the STC in existence, and she is ready to buy some storage. She connects her computer to the StorageCoin network, and puts in the following request: “I will pay one billion STC for someone to store a terabyte of data for me.” Now comes an important bit: will anyone accept this request? After all, no one know what STC is worth.

Fortunately, Alice is in luck. Bob is connected to the StorageCoin network, and he also monitors STC prices. He saw that someone (Alice) just paid $100 for one billion STC on an exchange, and concludes that one billion STC is worth $100. He accepts Alice’s request for storage, since she is paying the market price.

Alice and Bob make their exchange, and we arrive at the following situation: Bob is storing Alice’s terabyte of data, and he also has in his possession one billion STC. Here is the second critical bit: how can Bob go from STC to dollars? If critics of utility tokens are right, then the buys and sells should cancel out, and the STC are actually worth zero. Did Alice just swindle Bob?

No. The ICO critics are wrong, if an important fact holds true. If there is ongoing demand for storage, then the STC tokens hold value, and Bob can sell them for $100. Think of this: Alice paid $100 for the STC because she needed them to use the StorageCoin network. Bob just has to wait for another user to come along, and he can sell his tokens to this new user for $100.

The ICO critics miss a critical piece of the puzzle when they claim that all token values will trend to zero: utility tokens have an expected future value. This expected future value is in knowing that someone will come along in the near future, and buy the utility tokens for X so they can use service X. As long as service X is needed, market participants can rest assured they will be able to get a market price for their tokens.

A Framework for Token Value

Our thought experiment shows that the utility tokens sold in ICO’s have some value, but that still leaves an important question: what is that value? We’d like to have a model for token value, similar to P/E ratio for ICO companies.

Let’s return to our example from before, where Alice and Bob are transacting on the StorageCoin network. In this example, we established a price floor for STC tokens, and showed that utility token prices do not tend toward zero. But that leaves a question: where will they end up?

Let’s imagine a third actor enters the market: Charlie. Charlie also a consumer, and he wants to buy a terabyte of storage as well. With Alice and Charlie combined, there is now a demand for $200 worth of goods, double from before. Does this mean the price of STC tokens doubles?

Not quite. Imagine the following scenario: Alice buys all billion STC tokens for $100, and uses them to buy storage from Bob. Bob then takes those STC tokens, and sells them for $100 to Charlie, who uses them to buy storage again. Even though the output of our token economy has doubled, the price of tokens hasn’t changed! What’s going on?

The answer lies in something called the velocity of money. The velocity of money is the number of times a token is used for exchange in a given time period. In our example, each STC token was used two times over a month, which defines its token velocity. Thus, even though transaction volume double, token velocity doubled as well. The transaction value was diluted, since each token was used two times, and there was no price increase. This sounds like bad news for utility tokens and ICOs. If increasing usage doesn’t translate to increasing price, the entire model is broken.

Fortunately, that’s not the whole story. But before discussing how utility tokens can escape the Curse of Velocity, let’s formalize the problem. Whereas the formal statement of a stock shares price is P = kE, the formal value of a utility token can be understood using the equation of exchange:

MV = TP

where M is the monetary base (the number of tokens), V is the velocity of tokens (how many times in a time period is a token used?), T is the number of transaction (how many purchase take place in a time period?), and P is the price level (what is the value of the thing being purchase?). We can use algebra to solve for token price:

Pt = T / MV

Note that Pt is the token price, and it is the inverse of P, which is the price of goods.

This equation reveals both the problem, and the solution to it. In our example, the token velocity was directly proportional to the number of transactions. If the number of transactions in a time period is 10, then each token is used 10 times. If the number of transactions is a million per second, then each token is used a million times a second. The price of tokens never goes above some baseline value.

So how can we escape the Curse of Velocity? First, let’s remember a common criticism of bitcoin: that blocks take 10 minutes to confirm. A similar property exists in Ethereum. In short, there is a limit to how high token velocity can go, limited by the nature of decentralized systems. In our StorageCoin economy, imagine that the block speed is 10 minutes, just as bitcoin’s speed is. If Alice and Charlie want to buy storage within 10 minutes of each other, they’ll only be able to buy half of all the STC tokens available, and the price will double to $200.

In short, one way to capture value in tokens is to decrease token velocity. This can be done with protocol limits and other mechanisms to slow down the re-use of tokens. Specifically, well designed token economics will separate increases in transaction volume from increases in token velocity.

In addition to token velocity, let’s consider M, the monetary base, as well. The term monetary base is a bit unclear, but it means this: the tokens available for transacting. This is a related concept to velocity: if a token is stuck in a 10 minute block confirmation, it’s not available for transacting, and the effective monetary base decreases. But there are other mechanisms to decrease the effective monetary base. For example, proof of stake blockchains reward individuals who hold tokens from network fees. If the reward is high enough, individuals will withhold their tokens from the market, decreasing the effective supply.

There are more techniques for capturing value in tokens, and we will cover them in a subsequent blog post, and also add more detail and color to the mechanisms we just described.

Want to learn more about TensorCoin? Subscribe to the newsletter for updates