Chapter 11

Quantitative Models & Metrics—Error of Projection

“People who watch their weight, golf scores, and fuel bills seem to shun quantitative evaluation of their investment management skills although it involves the most important client in the world—themselves.”

—Warren Buffett

Numbers can be scary.

Cold.

Calculated.

Emotionless.

“Facts don’t care about your feelings,” and neither does quantitative analysis. Fortunately, we have options. No model is perfect. Every model is flawed. The key is finding a model that lines up decently well with reality such that when the present reality is disconnected from what we know to be true, we get to call it out. Place our bets. Stare the market straight in the eye and say, “You are overvaluing this investment” or “You haven’t accounted for this piece of information.”

The goal of this chapter is to examine the numbers that matter. Quantitative measurements give us the ability to compare blockchains to each other, as well as make predictions and projections about the future. While some of these sections appear to be mathematically intensive, all these models are publicly available by searching “CarterLWoetzel GitHub” online—you don’t have to make any of these models from scratch. The purpose of each section in this chapter is twofold: explaining the underlying logic of each model in addition to giving you simple examples of each model in action. It will be up to you to utilize the resources provided on my GitHub, as well as finding other models that exist out in the wild.

In my mind, two types of models exist: traditional and weak. Traditional models generate a target price based on dividends given to the owners of an asset, such as a stock or bond. Weak models are used to estimate the value of non-dividend producing investments. In the cryptocurrency space, dividends are uncommon. This is shifting as many blockchains move to the “proof of stake” consensus model, which provides a dividend to those who “stake” their crypto to secure the network. Currently, weak models are more common and accessible for cryptocurrency price projection. Regardless, we will propose a variety of models—many of which are looked down upon and others that are widely accepted.

Comparison Metrics

You are comparing two blockchains to each other, trying to figure out which cryptocurrency is more valuable. How do you tell which blockchain is more valuable quantitatively?

Figure 16: Comparison Metrics

As mentioned earlier, Coinmetrics as well as Blockchain Charts have all of this data publicly available. Marketing and PR can’t hide from these quantitative comparisons between different cryptocurrencies. Number of nodes is especially important because the more distributed the network, the safer the network is from 51 percent attacks, censorship, and network downtime. If a blockchain network is doing better on all of these metrics and is less valued than a different blockchain, it is a strong signal it is undervalued.

Net Worth Market Capitalization Model

The “Net Worth Market Capitalization Model,” created by Daniel Sangyoon Kim, is based on the assumption that in the near future, crypto will be part of every wealthy American’s portfolio. This weak model is not based on dividends, but a qualitative prediction based on cryptocurrency’s relevance as an asset class.

Figure 17: Network Capitalization Model

Here is a simple example using Bitcoin:

•Projected 0.025 percent of American’s portfolio (in the top 1 percent of income) will invest in cryptocurrency

•1.3 million households in the top 1 percent account for $11 trillion

•Bitcoin holds 50 percent of the total cryptocurrency market capitalization

•21 million total Bitcoin in circulation

•$5,200 current price of Bitcoin

•Projected Bitcoin market capitalization = $11 trillion * 0.025 percent * 50 percent = $137.5 billion

•Projected Bitcoin price = $137.5 billion / 21 million Bitcoin = $6,547

•Projected Bitcoin price $6,547 > current Bitcoin price $5,200

Because the projected value is greater than the current price, this quantitative model gives you a target sell price of $6,547 (a 21 percent return on investment). Kyle Samani of Multicoin Capital makes an argument for a projected $50–$100 trillion case in net worth invested based on Bitcoin use cases as digital gold, Bitcoin deflating the monetary premium of a variety of assets, Bitcoin as a replacement for offshore bank accounts, Bitcoin as a means for securing the world’s assets, and new economic activity facilitated on the Bitcoin network.

I hope the error of prediction is evident in this kind of model. Yet this model holds some truth: a certain number of people in the future will invest a certain amount of capital into cryptocurrency as an asset class. Every cryptocurrency holds a certain amount of the market share of the crypto sphere and therefore, based on the total capital invested into the asset class, will result in a price for every single cryptocurrency based on their respective market share. If you can have broadly accurate predictions of what these numbers look like in the future, you can make decisions quantitatively right now—radically error prone, but slightly helpful.

Industry Share-Steal Model

A weak model I have created is built on the idea that a blockchain and its respective cryptocurrency will gobble up a certain amount of a market share based on the use cases enabled by the blockchain. The benefit of this model is that it creates a logical ceiling or floor on valuation.

Take NEO, a cryptocurrency that aims to be a peer-to-peer digital currency aiming for cheap fees and a better smart contract design than Ethereum. Because the NEO blockchain is largely focused on transaction facilitation, we can somewhat fairly compare it to a company such as PayPal.

NEO has approximately eighteen thousand transactions per day. Compare this to PayPal, which has five million transactions per day. We will calculate NEO’s “stealing” of PayPal’s market cap based on the ratio of NEO’s transactions to PayPal’s transactions, which is 18,000/5,000,000 = 0.36 percent. Assuming NEO takes 0.36 percent of PayPal’s market capitalization (cap) using (Market Capitalization Stolen * M) / Ƈ, where M is equal to PayPal’s market cap ($182.02 billion) and Ƈ is the amount of NEO in circulation, we generate the following price:

Projected NEO Price = (0.0036 * $182,010,000,000 (PayPal Market Cap)/(100,000,000 NEO in circulation)= $6.55

Compare this to the current price of NEO of $9.55, and suddenly this model starts to give us some decent clarity. Perhaps market share use cases other than transaction facilitation account for the discrepancy. In the future, if NEO were being sold at twenty dollars while still only generating eighteen thousand transactions per day, we can firmly say we have quantitative evidence showing that the market has diverged from reality. If the market is valuing NEO drastically more without any increase in adoption (daily transactions), then we hedge our bets against the current NEO price assuming there is no new qualitative information to justify the price increase.

The much more dependable valuation of PayPal can be used as a hazy mirror to value different cryptocurrencies that are focused largely on transactions. This same process can be applied to blockchains focused on impacting other specific industries by creating a quantitative marker based on a reliable, publicly traded stock.

Note that this same model can be used to make projections—if adoption in the future will increase NEO’s number of total transactions to eighty thousand per day, suddenly the underlying valuation needs to reflect this increase in adoption. While traditional models are based on the growth of many future cash flows, weak models for cryptocurrency use transactions as the fundamental building block.

Comparable Tokens Model

Ernst & Young released “The Valuation of Crypto-Assets” in 2019. Within this primer, the comparable tokens model is described:

“ Consider the token market capitalizations achieved in recent, comparable ICOs as a proxy for the total value of the subject tokens issued, similar to the benchmarking approach taken in the valuation of early-stage companies raising VC funding rounds.”

Essentially, you take the market cap of the subject cryptocurrency you are examining ( is an excellent site for this), make a market capitalization comparison to another cryptocurrency, and pose the following question: “On the basis of my qualitative risk analysis and the quantitative metrics of these two blockchains, is the subject token properly priced in relation to another similar crypto asset?”

With blockchain, the comparable tokens model appears to be one of the most appealing models because it takes your qualitative investigation far more seriously than other models. So while this is about as subjective as a model could be, it is once again better than nothing. Here is an example of using this strategy:

Figure 18: Comparable Token Model Example

Using an extra simplified scorecard, which should include the comparison metrics listed earlier but doesn’t for this example, we examine the qualitative risk scores. According to this (fictional) qualitative scorecard, TRON’s market capitalization should increase by 10.64 percent because of the difference in risk, despite the fact the market does not currently price the respective cryptocurrencies this way. While this is a pretty absurd pricing model, this is surprisingly how many venture capitalists (as mentioned by Ernst & Young) operate at the earliest stages of valuation.

Discounted Transaction Cash Flow Model

The investment value of a blockchain is the total revenue generated by transactions, distributed to those who secure the network by staking or mining the cryptocurrency of said blockchain. Revenue is created and given to those who secure the network per transaction appended to the blockchain ledger. Every transaction is priced by a user but executed only if the miner or staker has determined the fee that will be received is worth it in relation to their costs (electricity costs to run a node, risk, and target rate of return). If the proposed user fee is worth it, the miner/staker adds the transaction to the block of transactions that will be appended to the ledger if the cryptographic solution is found.

As such, the stakers (those who lock their cryptocurrency into a smart contract to secure the network using the Proof of Stake Consensus model) or the miners (who expend electricity to append transactions using the PoW model) are the ones who decide the baseline value of any given cryptocurrency. Note that the market buyers and sellers continue to trade for certain prices, but because the miners and stakers are the ones who generate and receive cryptocurrency for maintaining the blockchain, their valuation of any cryptocurrency carries far more clout than a trader in the long run.

The more people adopt a blockchain, the more transactions that are executed. As a result of more people fighting to have their transactions be a part of the most recent block to be appended to the ledger, the greater the transaction fee. This is because miners and stakers can drive a harder bargain. All of this is classic supply and demand.

My model assumes growth in average fees is proportional to the growth in total transactions on a yearly basis. This tends to underestimate average fee growth, as the spurts in demand can be very unstable, strongly impacting the mean fee price. As time progresses, total transactions and the average price will more than likely decouple as innovation in blockchain technology will aim to decrease fees, which would increase total transactions to a blockchain as it becomes cheaper to transact and use because of the blockchain design improvements.

In addition, active currency in circulation is the set of crypto that is being used for its intended purpose (smart contracts and transactions). Miners and stakers are only concerned with the total level of activity that generates them revenue, and therefore will largely ignore currency that sits inactive. Active currency in circulation can be hard to hunt down for smaller cryptocurrencies, so beware of the significant impact on this model if total currency in circulation is used instead.

For those of you not familiar with the traditional discounted cash flow model, doing some basic research on the present value of future cash flows might be worthwhile. I will gloss over these basics for now as they are taught in a variety of courses and textbooks. The discounted transaction cash flow model (DTCF) I propose is as follows:

Figure 19: Discounted Transaction Cash Flow Model

•T = Total Transactions for year n

•F = Average Transaction Fee for year n

•r = Miner/Staker Target Rate of Return

Every year, the total transactions per year and the average transaction fee per year will grow by the projected total transaction growth rate. After this summation is done for n projected time periods, the total discounted cash flow is divided by the total active currency in circulation to arrive at your projected target price. Note that this formula is identical to the discounted cash flow model, with a slightly different structure to reflect revenue coming from transactions and fees of a blockchain to miners and stakers.

Figures 20, 21: DTCF Model Data

The 7 percent target rate of return is what is set to be in place for the Ethereum network with proof-of-stake and is approximately the rate of return miners are operating under with the current PoW consensus model. The Ethereum blockchain saw a 2,500 percent increase in total transactions from 2016 to 2017. This slowed down to 65 percent growth from 2018 to 2019. As such, I decided this particular Ethereum price valuation should underestimate growth per year at a mere 50 percent to give a realistic picture of the next five years. Depending on this growth rate, the value of Ethereum can range quite a bit. Seventy percent growth rate in total transactions per year puts Ethereum at $732 for Value Per Ethereum “share” compared to the current price of $203 (as of writing this). In addition, the starting average transaction fee of $0.23 in 2020 also impacts the valuation significantly.

Having created this model, I can say with confidence it has plenty of errors. As with all investment models, our goal is to match the value of an asset based on quantitative metrics and future growth projections using information publicly available.

I hope you can now see how flimsy and chaotic the art of quantitative modeling appears to be. As blockchains continue to mature, I expect we will see far better models appear. Cryptocurrency is a strange hybrid of currency, store of value, and digital fuel. It will power the future, that much I am certain of.

How shall we accurately measure its worth?

This remains a frustratingly complex mystery that will continue to be chipped away at as cryptocurrency sees more adoption from institutional and retail investors.

---