Non-Fungible Tokens (NFT). The Analysis of Risk and
Return.
Mieszko Mazur, IESEG School of Management
This version: 31 October 2021
Abstract
This study examines the risk and return characteristics of the NFT-based startups listed on the
cryptocurrency exchange. Our investigation is motivated by the recent surge in the NFT activity
on the part of creators, investors, and traders. We begin by proposing novel classification of the
existing NFTs that range from NFT blockchains through NFT metaverse to NFT DeFi. Next,
we establish that NFTs: 1) earn 130% on the first-listing-day; 2) yield an average investment
multiple of 40 (roughly 4,000%) over long-term, which is four times higher than bitcoin during
the same period; 3) deliver positive and significant alpha and exhibit above-average beta. We
also show that the NFT segment of the cryptocurrency market leads market recovery following
the mid-2021 crash and generate a return of close to 350%. In the final analysis of the paper,
we find that NFT infrastructure integrated within the existing blockchains increase market
valuations of these networks.
1
This paper was minted as an NFT (ERC-1155) on OpenSea under the contract address
0x2953399124F0cBB46d2CbACD8A89cF0599974963
1
,1 Introduction
Non-fungible token (NFT) is a way to record, verify, and track the ownership of a unique asset,
either physical or digital. Consequently, NFTs can be utilized to represent a work of art, futures
contract, music score, book, real estate, etc. – any type of object that could be considered unique
or rare. NFTs are minted, stored and transferred on a blockchain, and therefore cannot be seized or
tampered with by bad actors. On the other hand, NFTs can provide an instant proof of authenticity
and provenance, thus eliminating the problem of counterfeiting. In the first half of 2021, NFT sales
rose to a record $2.5 billion2 .
Just like the ERC-20 token revolutionized fundraising through Initial Coin Offerings (ICO)
in 2017, the ERC-721 token has been transforming the way investors interact with nonfungible assets,
whose elasticity of supply is close to zero - or in some instances - absolutely zero. Both types of tokes
were first minted on the Ethereum blockchain, however, recently they are also created on other types
of blockchains, some of them fully dedicated to NFTs (e.g., Flow, Ethernity, Efinity).
In this paper, we look at the investability of the NFT startups traded in the cryptocurrency
markets. First, we propose the novel classification of the NFT firms driven by the advancements
in the NFT technology3 . Most of these projects are linked to the NFT gaming, NFT decentralized
finance (DeFi), and NFT-dedicated blockchains – networks built solely for the purpose of serving the
NFT primary and secondary markets, as well as other NFT applications. Next, by analogy to the
IPO, we move on to the analysis of the first-day trading characteristics. We find that, at the time
of listing, the average NFT has roughly a unicorn status with 977 ($mil) of market capitalization.
The average first trading day return is an astounding 130% on a raw basis, and the first trading day
volume is about 333% of the average daily volume. To put these numbers into context, first day
returns to IPO are an order of magnitude lower. For example, Loughran and McDonald (2013) report
the first-day return of 35% at the mean (they consider it high), whereas Aggarwal et al. (2002) find
the first-day post-IPO volume of about 130%.
Further, we examine long-run price behavior of the NFTs. We calculate raw buy-and-hold
returns starting one day after the listing day – an approach that eliminates the impact of large
2
https://www.reuters.com/technology/nft-sales-volume-surges-25-bln-2021-first-half-2021-07-05/
3
Throughout the paper we refer to the NFT startups as NFT-based projects or simply NFTs. These are young
organizations that incorporate NFT technology into their main product line(s). In some instances, the NFT startups
are represented solely by open-source protocols. In lieu of the initial share offering, as in the standard entrepreneurship
model (e.g., Shleifer and Wolfenzon, 2002), NFT startups issue cryptocurrencies or tokens (in our analysis, we make
no distinction between the two).
2
,returns observed on the first day of trading. We find that NFTs yield spectacularly high returns in
the long-run. Because the returns are abnormally high, we adopt the approach used in the venture
capital (VC) industry and present the results as investment multiples, defined as the net proceeds
from selling the cryptocurrency divided by the net cost of buying the cryptocurrency. The average
value of the multiple in our sample is 6 (600% in percentage terms). It implies that the typical NFT
generates the return six times greater than the cost of investment.
Among the best performing NFTs are Axie Infinity Shards (AXS) that delivers an invest-
ment multiple of 535, Theta (THETA) with 60, and four other startups that obtain a multiple of 10
or higher. To put these numbers into perspective, a renowned venture capitalist Peter Thiel, real-
ized an investment multiple of 20 when selling Facebook shares in 2012 after an eight-year holding
period4 . In contrast to cumbersome and exclusive venture investments, available only for the certain
types of investment vehicles, NFTs are also available for retail clientele. Remarkably, compared to
typical winners firm in a VC portfolio (see e.g., Harris, Jenkinson, and Kaplan, 2014), NFTs appear
to yield much higher returns in a much shorter time frame. It is worth emphasizing that our analysis
is based on the publicly available cryptocurrency trading data. Knowing that the prices at which
tokens are sold to institutional investors in a private pre-sales are significantly lower than the prices
on the first-listing day, the true magnitude of the NFT investment multiples must be substantially
higher. Finally, our sample includes only about 23% NFTs that generate long-term losses.
The levels of volatility observed for the NFTs in our sample may seem excessive with
the standard deviation of daily returns of 11% (175% annualized). By comparison, the annualized
volatility of natural gas is 51%, oil 30%, and S&P 500 index 15% (Huang, Li, Wang, and Zhou,
2020). We also find that most volatile NFTs, yield the lowest returns. A similar pattern is observed
for public equities (Campbell and Hentschel, 1992), whereby volatility tends to rise after the prices
fall. However, in the context of cryptocurrencies, this finding cannot be in any way explained by the
leverage effect, as the NFT startups are debt-free.
In the tests that follow, we examine the returns on the risk-adjusted basis by using Sharpe
ratios and market-adjusted returns. A closer look at the Sharpe ratios reveal a sample average of
0.32. Interestingly, this figure compares with the historical average of 0.3 reported for publicly traded
stocks (Fama and French, 2002; Ferreira and Santa-Clara, 2011). Nevertheless, over 25% of the NFTs
in our sample have Sharpe ratios greater than 1 (e.g., AXS, THETA, MANA, ERN). Sharpe ratio,
4
https://www.wsj.com/articles/SB10000872396390443713704577601832028619176
3
, however, does not reflect the magnitude of the return. An asset with a low return and a low volatility
can have a relatively high Sharpe ratio. As an alternative, we also examine market-adjusted returns.
The performance of the NFTs on the market-adjusted basis is remarkable, with the average return of
83% for the holding period of about 300 days. Roughly 60% of NFTs in our sample deliver positive
returns and over half of these earn market-adjusted returns greater than 200%.
Next, we estimate the NFT alphas and betas. We follow Aggarwal, Green and Ren (2018)
and use capital asset pricing model (CAPM), as a single-factor model seems to work better for
alternative investments. We find that several NFTs deliver positive and significant alphas, whereas
most NFTs have betas greater than 1. At the NFT portfolio level, alpha is positive and highly
significant and the portfolio beta is roughly 1.1. Beta greater than 1 reflects greater volatility of
NFTs with respect to bitcoin price movements. This finding makes sense, since if bitcoin were to
fail as a technology then most likely other networks would fail as well – they are less resilient to
common risk factors (lower security) and have significantly shorter history. Positive and significant
alpha implies higher than expected returns due to additional risks to which NFTs might be exposed.
These risks may arise from new and untested technology used by NFT-driven startups, prototype
nature of their products and business models, regulatory uncertainty, and many others.
In the subsequent analysis, we construct the equal-weighted NFT price index. The objective
is to indicate visually the performance of the NFT portfolio against the backdrop of the return to
bitcoin – the oldest, most valuable, and trusted cryptocurrency. We are particularly interested in
gauging how NFTs respond to the bitcoin drawdown that occurred in the mid 2021, when bitcoin lost
about 55% of its market value (based on the intraday minimums and maximums). The magnitude of
the drawdown compares in percentage terms with the bitcoin price drop in March 2020 triggered by
COVID-19. First, we find that the portfolio of NFTs outperforms bitcoin in the long-run within the
time-window starting on the initial listing day and ending on the last day of the sample period. The
NFT investment multiple is close to 40 (4000%), compared to 10 (1000%) earned by bitcoin over the
same time span. Second and equally important, the NFT portfolio outperforms bitcoin following the
crash of mid-2021. The realized multiples are much smaller (3.5 vs. 1.5), however, what is perhaps
more interesting is that NFT segment leads the crypto market recovery after the market bottoms
out in the summer of 2021.
In the final test, we perform an event study to measure the valuation effect on the blockchains
4
Return.
Mieszko Mazur, IESEG School of Management
This version: 31 October 2021
Abstract
This study examines the risk and return characteristics of the NFT-based startups listed on the
cryptocurrency exchange. Our investigation is motivated by the recent surge in the NFT activity
on the part of creators, investors, and traders. We begin by proposing novel classification of the
existing NFTs that range from NFT blockchains through NFT metaverse to NFT DeFi. Next,
we establish that NFTs: 1) earn 130% on the first-listing-day; 2) yield an average investment
multiple of 40 (roughly 4,000%) over long-term, which is four times higher than bitcoin during
the same period; 3) deliver positive and significant alpha and exhibit above-average beta. We
also show that the NFT segment of the cryptocurrency market leads market recovery following
the mid-2021 crash and generate a return of close to 350%. In the final analysis of the paper,
we find that NFT infrastructure integrated within the existing blockchains increase market
valuations of these networks.
1
This paper was minted as an NFT (ERC-1155) on OpenSea under the contract address
0x2953399124F0cBB46d2CbACD8A89cF0599974963
1
,1 Introduction
Non-fungible token (NFT) is a way to record, verify, and track the ownership of a unique asset,
either physical or digital. Consequently, NFTs can be utilized to represent a work of art, futures
contract, music score, book, real estate, etc. – any type of object that could be considered unique
or rare. NFTs are minted, stored and transferred on a blockchain, and therefore cannot be seized or
tampered with by bad actors. On the other hand, NFTs can provide an instant proof of authenticity
and provenance, thus eliminating the problem of counterfeiting. In the first half of 2021, NFT sales
rose to a record $2.5 billion2 .
Just like the ERC-20 token revolutionized fundraising through Initial Coin Offerings (ICO)
in 2017, the ERC-721 token has been transforming the way investors interact with nonfungible assets,
whose elasticity of supply is close to zero - or in some instances - absolutely zero. Both types of tokes
were first minted on the Ethereum blockchain, however, recently they are also created on other types
of blockchains, some of them fully dedicated to NFTs (e.g., Flow, Ethernity, Efinity).
In this paper, we look at the investability of the NFT startups traded in the cryptocurrency
markets. First, we propose the novel classification of the NFT firms driven by the advancements
in the NFT technology3 . Most of these projects are linked to the NFT gaming, NFT decentralized
finance (DeFi), and NFT-dedicated blockchains – networks built solely for the purpose of serving the
NFT primary and secondary markets, as well as other NFT applications. Next, by analogy to the
IPO, we move on to the analysis of the first-day trading characteristics. We find that, at the time
of listing, the average NFT has roughly a unicorn status with 977 ($mil) of market capitalization.
The average first trading day return is an astounding 130% on a raw basis, and the first trading day
volume is about 333% of the average daily volume. To put these numbers into context, first day
returns to IPO are an order of magnitude lower. For example, Loughran and McDonald (2013) report
the first-day return of 35% at the mean (they consider it high), whereas Aggarwal et al. (2002) find
the first-day post-IPO volume of about 130%.
Further, we examine long-run price behavior of the NFTs. We calculate raw buy-and-hold
returns starting one day after the listing day – an approach that eliminates the impact of large
2
https://www.reuters.com/technology/nft-sales-volume-surges-25-bln-2021-first-half-2021-07-05/
3
Throughout the paper we refer to the NFT startups as NFT-based projects or simply NFTs. These are young
organizations that incorporate NFT technology into their main product line(s). In some instances, the NFT startups
are represented solely by open-source protocols. In lieu of the initial share offering, as in the standard entrepreneurship
model (e.g., Shleifer and Wolfenzon, 2002), NFT startups issue cryptocurrencies or tokens (in our analysis, we make
no distinction between the two).
2
,returns observed on the first day of trading. We find that NFTs yield spectacularly high returns in
the long-run. Because the returns are abnormally high, we adopt the approach used in the venture
capital (VC) industry and present the results as investment multiples, defined as the net proceeds
from selling the cryptocurrency divided by the net cost of buying the cryptocurrency. The average
value of the multiple in our sample is 6 (600% in percentage terms). It implies that the typical NFT
generates the return six times greater than the cost of investment.
Among the best performing NFTs are Axie Infinity Shards (AXS) that delivers an invest-
ment multiple of 535, Theta (THETA) with 60, and four other startups that obtain a multiple of 10
or higher. To put these numbers into perspective, a renowned venture capitalist Peter Thiel, real-
ized an investment multiple of 20 when selling Facebook shares in 2012 after an eight-year holding
period4 . In contrast to cumbersome and exclusive venture investments, available only for the certain
types of investment vehicles, NFTs are also available for retail clientele. Remarkably, compared to
typical winners firm in a VC portfolio (see e.g., Harris, Jenkinson, and Kaplan, 2014), NFTs appear
to yield much higher returns in a much shorter time frame. It is worth emphasizing that our analysis
is based on the publicly available cryptocurrency trading data. Knowing that the prices at which
tokens are sold to institutional investors in a private pre-sales are significantly lower than the prices
on the first-listing day, the true magnitude of the NFT investment multiples must be substantially
higher. Finally, our sample includes only about 23% NFTs that generate long-term losses.
The levels of volatility observed for the NFTs in our sample may seem excessive with
the standard deviation of daily returns of 11% (175% annualized). By comparison, the annualized
volatility of natural gas is 51%, oil 30%, and S&P 500 index 15% (Huang, Li, Wang, and Zhou,
2020). We also find that most volatile NFTs, yield the lowest returns. A similar pattern is observed
for public equities (Campbell and Hentschel, 1992), whereby volatility tends to rise after the prices
fall. However, in the context of cryptocurrencies, this finding cannot be in any way explained by the
leverage effect, as the NFT startups are debt-free.
In the tests that follow, we examine the returns on the risk-adjusted basis by using Sharpe
ratios and market-adjusted returns. A closer look at the Sharpe ratios reveal a sample average of
0.32. Interestingly, this figure compares with the historical average of 0.3 reported for publicly traded
stocks (Fama and French, 2002; Ferreira and Santa-Clara, 2011). Nevertheless, over 25% of the NFTs
in our sample have Sharpe ratios greater than 1 (e.g., AXS, THETA, MANA, ERN). Sharpe ratio,
4
https://www.wsj.com/articles/SB10000872396390443713704577601832028619176
3
, however, does not reflect the magnitude of the return. An asset with a low return and a low volatility
can have a relatively high Sharpe ratio. As an alternative, we also examine market-adjusted returns.
The performance of the NFTs on the market-adjusted basis is remarkable, with the average return of
83% for the holding period of about 300 days. Roughly 60% of NFTs in our sample deliver positive
returns and over half of these earn market-adjusted returns greater than 200%.
Next, we estimate the NFT alphas and betas. We follow Aggarwal, Green and Ren (2018)
and use capital asset pricing model (CAPM), as a single-factor model seems to work better for
alternative investments. We find that several NFTs deliver positive and significant alphas, whereas
most NFTs have betas greater than 1. At the NFT portfolio level, alpha is positive and highly
significant and the portfolio beta is roughly 1.1. Beta greater than 1 reflects greater volatility of
NFTs with respect to bitcoin price movements. This finding makes sense, since if bitcoin were to
fail as a technology then most likely other networks would fail as well – they are less resilient to
common risk factors (lower security) and have significantly shorter history. Positive and significant
alpha implies higher than expected returns due to additional risks to which NFTs might be exposed.
These risks may arise from new and untested technology used by NFT-driven startups, prototype
nature of their products and business models, regulatory uncertainty, and many others.
In the subsequent analysis, we construct the equal-weighted NFT price index. The objective
is to indicate visually the performance of the NFT portfolio against the backdrop of the return to
bitcoin – the oldest, most valuable, and trusted cryptocurrency. We are particularly interested in
gauging how NFTs respond to the bitcoin drawdown that occurred in the mid 2021, when bitcoin lost
about 55% of its market value (based on the intraday minimums and maximums). The magnitude of
the drawdown compares in percentage terms with the bitcoin price drop in March 2020 triggered by
COVID-19. First, we find that the portfolio of NFTs outperforms bitcoin in the long-run within the
time-window starting on the initial listing day and ending on the last day of the sample period. The
NFT investment multiple is close to 40 (4000%), compared to 10 (1000%) earned by bitcoin over the
same time span. Second and equally important, the NFT portfolio outperforms bitcoin following the
crash of mid-2021. The realized multiples are much smaller (3.5 vs. 1.5), however, what is perhaps
more interesting is that NFT segment leads the crypto market recovery after the market bottoms
out in the summer of 2021.
In the final test, we perform an event study to measure the valuation effect on the blockchains
4
