Cryptoassets, due to their versatile nature, are extremely hard to value. Some TradFi analysts argue that the DCF (Discounted Cash Flow) model be the bread and butter to all financial instruments, but actually such methods are only available to a very specific set of assets as pointed out by this article, which inspired me to jot down Medium.
Most price predictions we see on CT (Crypto Tweets) are so-called TA (Technical Analysis). I have no case against TA, but we should have some fundamental valuation as well to help ourselves fare better in volatile times. So, in this read, we will pore over fundamental valuation methods in crypto, and have some valuation exercises for each method.
There has been extensive studies on cryptoasset valuation, especially on the price of $BTC. The valuation framework I am currently aware of includes:
- Store of value ($BTC, $XRP) — Mitchnick (2018)
- Cost of production (PoW) — Hayes (2017)
- Stock-to-flow (PoW) — Plan B
- Network effect (PoW, PoS) — Alabi (2017)
- Quantity Theory of Money (PoW, PoS) — Wang (2014)
Above are all absolute fundamental valuation methods. For relative valuation analysis, we may have:
- On-Chain Metrics (PoW, PoS)
- Valuation Ratios (PoW, PoS)
For the first three methods (SoV, COP, S2F), those two are only applicable to medium-of-exchange designs and PoW chains, and the first one focuses only on cryptoassets as a payment system, so we will exclude these three for now as we want a more generally applicable framework.
Our analysis will be on the latter two methods (Network, QTM) with three L1 chains: $ETH, $LUNA, $FTM.
We are minus $BTC as all references have done the same analysis for $BTC, so doing the same work is simply redundant. For $SOL, we have no way to put hands on the blockchain statistics (wen SolScan?). The same goes for $AVAX and $NEAR, although they do show nice statistics on their respective explorers, they lack the feature to import the raw data as .csv, so we have no access to their data.
Network Effect (Metcalfe’s Law)
Metcalfe’s Law suggests that the value of network grows exponentially with the number of network participants. When there is only 1 node, it has no activity. When there are 2 nodes, we have one activity (A-B). When there are 3 nodes, we have three (A-B, A-C, B-C).
It caught public attention as the equation quite accurately described the price actions of SNS corporates, namely Facebook. Later in 2017, Ken Alabi applied the same methodology to value cryptoassets ($BTC, $ETH, $DASH), and surprisingly he found out that the model was a quite good fit.
Alabi used a slightly modified equation, but we will stick to the original Metcalfe one as it is much simpler to comprehend.
The valuation model requires a single input: N. For N, we will use the daily active addresses for each chain as a proxy to network participants. To be more precise, we will use the 30-day MA daily active addresses to make a prediction on the network growth in the form of a netoid function.
The implication of the function is that the network grows exponentially until a certain point (in above example, tm = 500), and then the growth slows down as it reaches the peak number (in above example, p = 1000).
If we have C = 1 for our Metcalfe’s Law, our network value would be V(N) = 250,000 at t = 500.
$ETH Valuation — Metcalfe’s Law
All data sources are available at EthScan.
We have ~600K daily active addresses as of Jan 2022. If we fit the netoid function to the data above, we get:
The model parameters are: p = 760,000, v = 0.0153, tm = 1,710.
We can easily project the function since we have the paramters. We have the projection over the next 3 years (1000 days), and the daily active address would grow to ~700K from ~600K today. Note that the maximum growth for the Ethereum network came at tm = 1,710, or May 2020, so the growth rate keeps decreasing until we reach our maximum at 760,000.
Next, we fit the 30-day MA daily active address data onto daily closing $ETH prices, we have C = 0.0000000007686.
The graph above shows the price using Metcalfe’s Law, or V(N) = CN². We graphed Alabi function just for illustration purpose. Looks fairly plausible to me.
We projected for 2000 days just to clearly show the convexity of the pricing function. As N approaches its maximum, its growth rate significantly slows down, so does the price growth as well. According to the model, the price of $ETH should be $5,000 by EOY, and is capped around $15,000.
Altcoin valuation — Metcalfe’s Law
For all altcoins, the main problem is that they have much less data compared to the Ethereum network, so they are really prone to overfitting.
$LUNA
For $LUNA, the maximum growth period has already passed (~ Nov 2021), and the Terra network active addresses were ~30K by then according to FlipsideCrypto.
$LUNA price by EOY would be ~$150, its price capped at $250 in the long run.
$FTM
There are two serious problems with $FTM pricing function. 1) The altcoin data problem, 2) the daily active address counting problem. FtmScan does not provide us with the daily active address data, so I used IntoTheBlock as the data source for Fantom active addresses. @FantomFDN reports 75K+ daily active addresses, but IntoTheBlock only shows ~2K active addresses, a tremendous difference. In short, we were forced to use different standards for $FTM.
I artificially toned down the Fantom network growth; I set the current time, Jan 2022, as the maximum growth period for Fantom active addresses, and the growth rate would decline henceforth.
$FTM would reach ~$5 by EOY, and its price would be capped around ~$8.
Limitations of the Network Effect Valuation
Pros for this methodology are:
- It focuses on the fundamental value of blockchains — a trustless networking service
- It provides a good fit to the available data
Cons are:
- It does not account for exogenous variables, such as macroeconomic conditions
- Daily active addresses might not be a good proxy for the network effect
The second weakness is the largest obstacle for the method as there are lots of cross-chain activities nowadays, and also daily active addresses does not account for CEX activities at all, which is still much greater than on-chain activities.
Metcalfe’s Law also has no concern for L1 traits— ex. $LUNA and $UST adoption. Maybe if we have access to all daily active addresses interacting with $UST, we have a much fine representation of the Terra chain.
Also we did not account for inflationary or deflationary tokenomics, though this could be easily incorporated had we used MC as the dependent variable, not P.
In my opinion, $ETH valuation is the most plausible one, as the Ethereum Network is the largest network by far so exception of cross-chain activities and CEX activities might be a minor flaw.
QTM Valuation
QTM is a classical macroeconomic model, criticized continuously by scholars from all fields. Nonetheless, it serves its purpose due to its simplicity.
In cryptoassets, M refers to the dependent variable “Market capitalization”, V refers to the velocity of the cryptoasset, P refers to the average price for transactions, Y refers to the total number of transactions.
The intuition behind the model is that the market capitalization should equal to the value required to fully utilize the economic value of a blockchain economy.
The first paper to apply QTM to value Bitcoin was Wang (2014), though he never actually came up with hard numbers.
The real advantage for blockchain economies to use QTM is that, unlike actual real world economies where we only have vague estimates, all transactions are recorded on blockchains so we can directly calculate the velocity of money. Pernice et al (2020) explains that out of all the velocity measures CTs use (CDD, etc.), a simple ratio of on-chain transaction volume to total coin supply captures velocity the best.
We have not derived the numbers for velocity, but Zochowski (2019) shows some nice way to calculate velocity and to eliminate self-churns.
$ETH Valuation — QTM
Now for QTM, we have no fancy graphs as we did above; we only have a call on each exogenous variable on the QTM equation.
The numerator simply means the annual transaction volume measured in USD. We got the $ETH daily transaction volume from Messari, averaged the last 30 days, and extrapolated it to 1 year by multiplying 365 — and we get $7.2B.
For velocity, we will just use what is available on Santiment API. We have 4 for $ETH velocity.
(7.2B ÷ 4) ÷ Max Supply (119M) = 5,466 USD/ETH.
Notice we just used the current circulating supply as a proxy for the maximum supply for $ETH, as there is none. Deflationary policies such as $ETH or $BNB works to “decrease” the velocity of money and thus inflating the price even if the max supply is capped at a certain number.
Altcoin Valuation — QTM
We will do $FTM first since it has similar data sources as $ETH.
$FTM
We have $FTM velocity at ~3 from the same data source as $ETH, but this time we downloaded IntoTheBlock data for Fantom daily transaction volume in USD.
The projected annual transaction volume = $22.8B, so we compute:
(22.8B ÷ 3) ÷ Max Supply (3.2B) = 2.39 USD/FTM.
If we decrease velocity to 1.5 (perhaps by burning the circulating supply), the price for $FTM would double.
$LUNA
Now, Terra Station provides us with the monthly transaction volume measured in $LUNA, so we will just use that value × $50 USD/LUNA as a proxy to the annual transaction volume. For velocity, we have 4, since, well, $ETH and $LUNA both as burning mechanism (I didn’t get lazy, trust me).
The projected annual transaction volume = $140.5B, so we have:
(140.5B ÷ 3) ÷ Max Supply (400M) = 87.6 USD/LUNA.
Limitations of QTM Valuation
Although with cryptocurrencies, we can directly measure the velocity of money, still we cannot escape from the fact that the very QTM model itself is prone to criticism.
Moreover, even with open ledgers, velocity is volatile and varies according to observation periods.
Pros
- It focuses on the value of blockchain economies
- Model inputs can be directly calculated from open ledgers
Cons
- Model inputs may significantly vary
- The same criticisms for QTM applies
Investors face immense difficulties in valuing cryptoassets as cryptoassets are brand new asset class and each cryptoasset has its own distinctive characteristics. We should not say “$DAI $10B, $UST $10B, so $MKR = $LUNA,” though both are somewhat equal in the sense that they are stablecoin projects.
The current fundamental valuation framework for cryptoassets are filled with flaws: For one, we cannot even value new projects with small track records with these “fundamental” methods (notice the drastic difference between parameters for each L1 tokens?).
Nonetheless, the literature is evolving, and remember that even in 400 years old stock markets the valuation methods are far from perfect. In volatile times, it may be good to take a step back, and to see what we are actually investing in.
In Part 2 (if there ever is one), we will cover all other methods left out in Part 1.
References
- Korbit Research Paper, https://cdn.korbit.co.kr/athena/etc/research/5/korbit_research_5_2022-01-26.pdf
- Mitchnick (2018) https://s3-us-west-1.amazonaws.com/fundamental-valuation-framework-for-cryptoassets/A+Fundamental+Valuation+Framework+for+Cryptoassets_June+2018.pdf
- Hayes (2017) https://www.sciencedirect.com/science/article/abs/pii/S0736585315301118
- Alabi (2017) https://www.sciencedirect.com/science/article/abs/pii/S1567422317300480
- Wang (2014) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2394024
- Pernice et al (2020) https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3499500
- Zochowski (2019) https://medium.com/logos-network/empirical-velocity-estimates-and-artificial-volumes-in-ethereum-9aa3d93bb60
