NBER WORKING PAPER SERIES

COMMON RISK FACTORS IN CRYPTOCURRENCY

Yukun Liu
Aleh Tsyvinski
Xi Wu

Working Paper 25882

http://www.nber.org/papers/w25882

NATIONAL BUREAU OF ECONOMIC RESEARCH

1050 Massachusetts Avenue
Cambridge, MA 02138
May 2019

We thank Nicola Borri, Markus Brunnermeier, Kent Daniel, Zhiguo He, Andrew Karolyi, Alan
Kwan, Ye Li, Nikolai Roussanov, Jinfei Sheng, Michael Sockin, and Jessica Wachter for
comments. We are grateful to Colton Conley and Dean Li for their excellent research assistance.
The views expressed herein are those of the authors and do not necessarily reflect the views of the
National Bureau of Economic Research.

NBER working papers are circulated for discussion and comment purposes. They have not been
peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies
official NBER publications.

© 2019 by Yukun Liu, Aleh Tsyvinski, and Xi Wu. All rights reserved. Short sections of text, not
to exceed two paragraphs, may be quoted without explicit permission provided that full credit,
including © notice, is given to the source.
Common Risk Factors in Cryptocurrency
Yukun Liu, Aleh Tsyvinski, and Xi Wu
NBER Working Paper No. 25882
May 2019
JEL No. G12

ABSTRACT

We find that three factors – cryptocurrency market, size, and momentum – capture the cross-
sectional expected cryptocurrency returns. We consider a comprehensive list of price- and
market-related factors in the stock market, and construct their cryptocurrency counterparts. Nine
cryptocurrency factors form successful long-short strategies that generate sizable and statistically
significant excess returns. We show that all of these strategies are accounted for by the
cryptocurrency three-factor model.

Yukun Liu Xi Wu
Department of Economics Stern School of Business
Yale University New York University
New Haven, CT 06520-8268 New York, NY 10012
[email protected] [email protected]

Aleh Tsyvinski
Department of Economics
Yale University
Box 208268
New Haven, CT 06520-8268
and NBER
[email protected]
1 Introduction
The cryptocurrency market has experienced rapid growth. This market allows companies
to raise money without engaging with venture capitalists and to be traded without being
listed on stock exchanges. The entire set of coins in the crypto market ranges from well-
known currencies such as Bitcoin, Ripple, and Ethereum to much more obscure coins. There
are two views on the cryptocurrency market. The first is that most and perhaps all of the
coins represent bubbles and fraud. The second is that the blockchain technology embodied
in coins may become an important innovation and that at least some coins may be assets
that represent a stake in the future of this technology. If the latter case is true, analyzing the
cryptocurrency market from the empirical asset pricing point of view is important for at least
two reasons. The first reason is to understand whether the returns of cryptocurrencies share
similarities with other asset classes, most importantly, with equities. The second reason is
to establish a set of empirical regularities that can be used as stylized facts and important
inputs to assess and develop theoretical models of cryptocurrency.
In this paper, we study the cross-section of cryptocurrency returns. Our primary goal
is to examine this market using standard empirical asset pricing tools. We consider all of
the coins with market capitalizations above one million dollars and their returns from the
beginning of 2014 to the end of 2018. The number of such coins grew from 109 in 2014 to
1,583 in 2018.
We examine whether the characteristics that are deemed important in the cross-section
of equity returns are also present in the cryptocurrency market. We find that many of the
known characteristics in the equity market also form successful long-short trading strategies
in the cross-section of cryptocurrencies. In particular, three factors – cryptocurrency market,
size, and momentum – capture most of the cross-sectional expected returns.
The literature on the stock market established a number of factors that explain the
cross-section of stock returns. Among the factors compiled by Feng et al. (2017) and Chen
and Zimmermann (2018), we select those that are constructed based only on price and
market information – 25 such factors in total. We first describe the construction of the
cryptocurrency counterparts for all these factors in the cross-section of cryptocurrencies.
There are broadly four groups of factors: size, momentum, volume, and volatility. We also
construct a coin market return using all of the coins for which the data is readily available.
The coin market return series comprises 1,707 coins weighted by their market capitalization.
We then analyze the performance of all the 25 factors in the cryptocurrency market.

1
Each week, we sort the returns of individual cryptocurrencies into quintile portfolios based
on the value of a given factor. We track the return of each portfolio in the week that follows
and calculate the average excess return over the risk-free rate of each portfolio. We then
form the long-short strategy based on the difference between the fifth and the first quintiles.
We find that the returns of the zero-investment strategies are statistically significant for 9
out of the 25 factors. Specifically, these are: market capitalization, price, and maximum
price; one-, two-, three-, and four-week momentum; dollar volume; and standard deviation
of dollar volume. We now turn to the detailed description of the results for each group of
factors.
For the statistically significant size related strategies, a zero-investment long-short strat-
egy that longs the smallest coins and shorts the largest coins generates more than 3 percent
excess weekly returns (3.4 percent for the market capitalization, 3.9 percent for the end of
week price, and 4.1 percent for the highest price of the week strategies). For the momen-
tum strategies, a zero-investment long-short strategy that longs the coins with compara-
tively large price increases and shorts the coins with comparatively small increases generates
about 3 percent excess weekly returns (2.7 percent for one-week momentum, 3.3 percent for
two-week momentum, 4.1 percent for three-week momentum, and 2.5 percent for four-week
momentum strategies). For the volume related strategies, a zero-investment strategy that
longs the lowest volume coins and shorts the highest volume coins generates about 3 per-
cent excess weekly returns (3.2 percent for the dollar volume). For the volatility strategy, a
zero-investment strategy that longs the lowest dollar volume volatility coins and shorts the
highest dollar volume volatility coins generates about 3 percent excess weekly returns. For
all of these factors, the returns on individual quintile portfolios are almost monotonic with
the quintiles. Determining the cryptocurrency factors that predict the cross-section of the
entire cryptocurrency space is the first main result of the paper.
Next, we investigate whether these nine cross-sectional cryptocurrency return predictors
can be spanned by a small number of factors. Our second main result is to develop a factor
model for the cross-section of the cryptocurrency returns. We first consider a one-factor
model with the coin market factor only. This is, in essence, a cryptocurrency CAPM model.
The results are similar to those found in other asset classes – the model performs poorly
in pricing the cross-section of the coin returns. The alphas for most of the successful zero-
investment strategies remain large and statistically significant. The alphas for some of the
strategies decrease marginally. The explanatory power of the model is low, with the R2 s of
the long-short strategies ranging from about zero percent for the one-week momentum to

2
6.8 percent for the maximum day price strategies.
We next show that a three-factor model with the cryptocurrency market factor (CMKT),
a cryptocurrency size factor (CSMB), and a cryptocurrency momentum factor (CMOM), ac-
counts for the excess returns of all of the nine successful zero-investment strategies. Adjusted
for the cryptocurrency three-factor model, none of the alphas of the nine strategies remains
statistically significant. The CSMB factor accounts for the following strategies: market cap-
italization, price, maximum day price, dollar volume, and the standard deviation of dollar
volume. The CMOM factor accounts for the two-week, three-week, and four-week momen-
tum strategies. Both CSMB and CMOM account for the one-week momentum strategy. We
conclude that the cryptocurrency three-factor model captures the cross-section of expected
returns of cryptocurrencies.
Finally, we note several additional results. First, as the construction of the long-short
strategies relies on the ability to short coins, a natural criticism of our findings is that short
selling is either not possible or limited for most of the coins. We thus analyze each strategy
that shorts Bitcoin instead of shorting the relevant quintile portfolio. The results virtually
do not change. Second, we find that the momentum strategies perform significantly better
among the larger coins. The momentum strategy in the below median size group generates
statistically insignificant 0.6 percent weekly excess returns; the momentum strategy in the
above median size group generates statistically significant 4.2 percent weekly returns. We
also show that the stock market factor models, such as the Fama-French 3-factor, Carhart
4-factor, and the Fama-French 5-factor models, do not account for the cross-section of cryp-
tocurrency returns. Additionally, we show that the procedure that removes the unpriced
risks similar to Daniel et al. (2018) strengthens the cryptocurrency size factor but not the
cryptocurrency momentum factor. One possible explanation is that loadings on the cryp-
tocurrency momentum factor are more transient than loadings on the cryptocurrency size
factor.
We briefly discuss the relationship to the literature. Size and momentum are among
the most studied strategies in asset pricing. The size effect in the stock market is first
documented in Banz (1981). Fama and French (1992) show that size and value are important
factors in explaining the cross-section of expected stock returns. Our findings on momentum
are related to many papers on the topic such as Jegadeesh and Titman (1993), Moskowitz
and Grinblatt (1999), Moskowitz et al. (2012), Asness et al. (2013). The use of factor models
to analyze asset returns dates back to the papers of Fama and French (1993) and Fama and
French (1996). Lustig et al. (2011), Szymanowska et al. (2014), and Bai et al. (2018) develop

3
factor models for the currency, commodity, and corporate bond markets, respectively.
Yermack (2015) is one of the first papers that brings academic attention to the field of
cryptocurrency. A number of recent papers develop models of cryptocurrencies (see, e.g.,
Weber, 2016; Biais et al., 2018a; Chiu and Koeppl, 2017; Cong and He, 2018; Cong et al.,
2018a; Cong et al., 2018b; Sockin and Xiong, 2018; Schilling and Uhlig, 2018; Abadi and
Brunnermeier, 2018; Routledge and Zetlin-Jones, 2018). Several recent papers document em-
pirical facts related to cryptocurrency investments (e.g., Stoffels, 2017; Hubrich, 2017; Borri,
2018; Borri and Shakhnov, 2018a; Borri and Shakhnov, 2018b; Hu et al., 2018; Makarov and
Schoar, 2018; Liu and Tsyvinski, 2018; Li and Yi, 2018).

2 Data
We collect trading data of all cryptocurrencies available from Coinmarketcap.com. Coin-
marketcap.com is a leading source of cryptocurrency price and volume data. It aggregates
information from over 200 major exchanges and provides daily data on opening, closing, high,
low prices, volume and market capitalization (in dollars) for most of the cryptocurrencies.1
For each cryptocurrency on the website, its price is calculated by taking the volume weighted
average of all prices reported at each market. A cryptocurrency needs to meet a list of cri-
teria to be listed, such as being traded on a public exchange with an API that reports the
last traded price and the last 24-hour trading volume, and having a non-zero trading volume
on at least one supported exchange so that a price can be determined. Coinmarketcap.com
lists both active and defunct cryptocurrencies, thus alleviating concerns about survivorship
bias.
We use daily close prices to construct weekly coin returns. Specifically, we divide each
year into 52 weeks. The first week of the year consists of the first seven days of the year. The
first 51 weeks of the year consist of seven days each and the last week of the year consists
of the last eight days of the year.2 Our sample includes 1,707 coins from the beginning of
2014 to the end of 2018. The trading volume data became available in the last week of 2013,
and thus our sample period starts from the beginning of 2014. We require that the coins
have information on price, volume, and market capitalization. We further exclude coins
with market capitalizations of less than $1,000,000. To alleviate concerns for outliers, we
winsorize all non-return variables by the 1st and 99th percentiles each week.

1 Some coins are not tracked by the website because the coins’ exchanges do not provide accessible APIs.
2 The last week of 2016 consists of the last nine days of the year.

4
 The summary statistics are presented in Panel A of Table 1. The number of coins in our
sample that satisfy all the filters increases from 109 in 2014 to 1,583 in 2018. The mean
(median) market capitalization in the sample is 356.71 (8.17) million dollars. The mean
(median) daily dollar volume in our sample is 18,305.83 (103.89) thousand dollars.
We construct a cryptocurrency market return as the value-weighted return of all the
underlying available coins. The cryptocurrency excess market return (CMKT) is constructed
as the difference between the cryptocurrency market index return and the risk-free rate
measured as the one-month Treasury bill rate. The summary statistics are presented in
Panel B of Table 1. During the sample period, the average coin market index return is 1.3
percent per week, which is higher than the average Bitcoin return (1.2 percent per week)
but is lower than the average Ripple return (3.5 percent per week) or Ethereum return
(4.6 percent per week).3 The weekly standard deviation of the coin market index return is
0.117, which is slightly higher than that of Bitcoin (0.114) but much lower than those of
Ripple (0.267) and Ethereum (0.241). The coin market index returns have positive skewness
and kurtosis. Figure 1 plots the cryptocurrency market index against Bitcoin, Ripple, and
Ethereum. The values are presented as the US dollar value of investing one dollar from the
inception of the given cryptocurrency to facilitate comparisons. The figure shows strong
correlations among the cryptocurrency market index and the investment values of the major
coins.
We obtain the stock market factors for the Fama French 3-factor, Carhart 4-factor, and
Fama French 5-factor models from Kenneth French’s website.

3 Bitcoin,Ripple, and Ethereum are the three largest cryptocurrencies by market capitalization and thus
form a natural reference group.

5
 Table 1: Summary Statistics
Panel A reports the number of coins, the mean and median of market capitalization, and the mean and
median of daily trading dollar volume by year. Panel B reports the characteristics of coin market index
returns, Bitcoin returns, Ripple returns, and Ethereum returns. The coin market index returns, Bitcoin
returns, and Ripple returns start from the first week of 2014. The Ethereum returns start from the thirty-
second week of 2015.

Panel A
Year Number of Coins Market Cap (mil) Volume (thous)
Mean Median Mean Median

2014 109 239.83 3.89 1,146.09 36.24

2015 77 134.53 2.76 1,187.64 11.51
2016 155 160.06 3.39 1,789.24 23.73
2017 804 435.68 9.01 18,509.55 133.56
2018 1,583 357.20 8.91 20,829.12 124.02

Full 1,707 356.71 8.17 18,305.83 103.89

Panel B
Mean Median SD Skewness Kurtosis

Coin Market Return 0.013 0.006 0.117 0.294 4.574

Bitcoin Return 0.012 0.005 0.114 0.367 4.580
Ripple Return 0.035 -0.007 0.267 3.478 21.263
Ethereum Return 0.046 0.001 0.241 1.841 9.843

6
 Figure 1: Cryptocurrency Market Index and Major Coins
This figure plots the cryptocurrency market index against Bitcoin, Ripple, and Ethereum.

7
3 Cross-Sectional Factors
We consider a comprehensive list of the established factors in the cross-section of stock
returns, compiled by Feng et al. (2017) and Chen and Zimmermann (2018). Among these,
we select all the factors that can be directly constructed using only the information on price,
volume, and market capitalization. The reason we consider only the market-based factors is
that financial and accounting data for the cross-section of coins is either not readily available
or not applicable. We hence investigate 25 factors, which we present in Table 2. We further
group them into four broad categories: size, momentum, volume, and volatility.

3.1 Size Factors

We analyze the performance of the zero-investment long-short strategies based on the
size-related factors: market capitalization, price, maximum price, and age. Each week, we
sort individual cryptocurrencies into quintile portfolios based on the value of a given factor.
We track the return of each portfolio in the week that follows. We then calculate the average
excess returns over the risk-free rate of each portfolio, and the excess returns of the long-
short strategies based on the difference between the fifth and the first quintiles. We find
that the first three factors generate statistically significant long-short strategy returns. The
result of the zero-investment long-short strategy for age is not statistically significant and is
summarized in the last part of the section.
Table 3 presents the results. For the first three factors, the average mean excess returns
decrease from the top to the bottom quintiles. The differences in the average returns of the
highest and lowest quintiles are -3.4 percent for market capitalization, -3.9 percent for the
end of week price, and -4.1 percent for the highest price of the week. All of these differences
are statistically significant at the 5 percent level. In other words, a zero-investment strategy
that longs the smallest coins and shorts the largest coins generates about 3 percent excess
weekly returns. Of course, this strategy does not take into account trading costs and the
feasibility of short selling. We consider strategies that short Bitcoin, and present results that
long the smallest coins and short Bitcoin in Section 5. In the Appendix, we also present
results based on tercile instead of quintile portfolios.4 The results based on tercile portfolios

4 The same robustness results are presented for all other successful strategies in the Appendix.

8
are qualitatively similar.

Table 2: Factor Definitions

Category Factor Definition

Size MCAP Log last day market capitalization in the portfolio formation week
Size PRC Log last day price in the portfolio formation week
Size MAXDPRC The maximum price of the portfolio formation week
Size AGE The number of weeks that have been listed on Coinmarketcap.com
Momentum r 1,0 One-week momentum
Momentum r 2,0 Two-week momentum
Momentum r 3,0 Three-week momentum
Momentum r 4,0 Four-week momentum
Momentum r 8,0 Eight-week momentum
Momentum r 16,0 Sixteen-week momentum
Momentum r 50,0 Fifty-week momentum
Momentum r 100,0 Hundred-week momentum
Volume VOL Log average daily volume in the portfolio formation week
Volume PRCVOL Log average daily volume times price in the portfolio formation week
Volume VOLSCALED Log average daily volume times price scaled by market capitalization in
the portfolio formation week
Volatility BETA The regression coefficient βC
i
M KT in Ri − R f = α + βC M KT CMKT + i .
i i

The model is estimated using daily returns of the previous 365 days
before the formation week.
Volatility BETA2 Beta squared
Volatility IDIOVOL The idiosyncratic volatility is measured as the standard deviation of
the residual after estimating Ri − R f = αi + βC
i
M KT CMKT + i . The
model is estimated using daily returns of the previous 365 days before
the formation week.
Volatility RETVOL The standard deviation of daily returns in the portfolio formation week
Volatility RETSKEW The skewness of daily returns in the portfolio formation week
Volatility RETKURT The kurtosis of daily returns in the portfolio formation week
Volatility MAXRET Maximum daily return of the portfolio formation week
Volatility DELAY The improvement of R2 in
Ri − R f = αi + βC
i
M KT CMKT + βC M KT−1 CMKT−1 + βC M KT−2 CMKT−2 + i ,
i i

where CMKT−1 and CMKT−2 are the lagged one and two day coin
market index returns, compared to using only current coin market
excess returns. The model is estimated using daily returns of the
previous 365 days before the formation week.
Volatility STDPRCVOL Log standard deviation of dollar volume in the portfolio formation week
Volatility DAMIHUD The average absolute daily return divided by dollar volume in the
portfolio formation week

9
 Table 3: Size Factor Returns
This table reports the mean quintile portfolio returns based on the market capitalization, last day price,
and maximum day price factors. The mean returns are the time-series averages of weekly value-weighted
portfolio excess returns. *, **, *** denote significance levels at the 10%, 5%, and 1%.

Quintiles
1 2 3 4 5 5-1

MCAP Low High

Mean 0.047*** 0.018 0.013 0.013 0.013* -0.034**
t(Mean) (2.958) (1.610) (1.286) (1.439) (1.766) (-2.557)

PRC Low High

Mean 0.051*** 0.029** 0.001 0.018 0.012* -0.039**
t(Mean) (2.739) (2.118) (0.117) (1.419) (1.689) (-2.420)

MAXDPRC Low High

Mean 0.053*** 0.025* 0.002 0.021 0.012* -0.041**
t(Mean) (2.791) (1.905) (0.143) (1.568) (1.681) (-2.483)

3.2 Momentum
We analyze the performance of the zero-investment long-short strategies based on the
one-, two-, three-, four-, eight-, sixteen-, fifty-, and one hundred-week momentum factors.
Each week, we sort individual cryptocurrencies into quintile portfolios based on the value of
a given factor. All strategies are rebalanced weekly. We find that the one-, two-, three-, and
four-week momentum factors generate statistically significant long-short strategy returns.
The results of the zero-investment long-short strategies for the eight-, sixteen-, fifty, and one
hundred-week momentum are not statistically significant and are summarized in the last
part of the section.
Table 4 presents the results of the successful factors for the portfolios sorted in quintiles.
For the one-, two-, three-, and four-week momentum strategies, the average mean excess
returns increase with the quintiles. The patterns are almost universally monotonic. The

10
difference in the average returns of the highest and lowest quintiles is about 3 percent for
each horizon and statistically significant at the 5 percent level (1 percent level for three-week
momentum). In other words, a zero-investment strategy that longs the coins with compara-
tively large increases and shorts the coins with comparatively small increases generates about
3 percent excess weekly returns. The differences in the average returns of the highest and
lowest quintiles are 2.7 percent for the one-week momentum, 3.3 percent for the two-week
momentum, 4.1 percent for the three-week momentum, and 2.5 percent for the four-week
momentum.

Table 4: Momentum Factor Returns

This table reports the mean quintile portfolio returns based on the one-week, two-week, three-week, and four-
week momentum factors. The mean returns are the time-series averages of weekly value-weighted portfolio
excess returns. *, **, *** denote significance levels at the 10%, 5%, and 1%.

Quintiles
1 2 3 4 5 5-1

r 1,0 Low High

Mean -0.006 -0.002 0.010 0.042** 0.021 0.027**
t(Mean) (-0.552) (-0.176) (1.094) (2.317) (1.550) (1.994)

r 2,0 Low High

Mean -0.002 0.005 0.010 0.018* 0.030** 0.033**
t(Mean) (-0.225) (0.493) (1.155) (1.894) (2.314) (2.442)

r 3,0 Low High

Mean 0.002 0.001 0.016 0.020** 0.043*** 0.041***
t(Mean) (0.156) (0.124) (1.587) (2.091) (2.956) (2.742)

r 4,0 Low High

Mean 0.002 0.004 0.008 0.019* 0.027** 0.025**
t(Mean) (0.243) (0.435) (0.935) (1.921) (2.033) (2.002)

11
3.3 Volume Factors
We analyze the performance of the volume-related factors: volume, dollar volume, and
scaled volume. Each week, we sort individual cryptocurrencies into quintile portfolios based
on the value of a given factor. All strategies are rebalanced weekly. The dollar volume
strategy generates statistically significant long-short strategy returns. The results of the
zero-investment long-short strategies based on the other volume factors are not statistically
significant and are summarized in the last part of the section.
Table 5 presents the results for the portfolios sorted in quintiles based on the dollar
volume factor. The average mean excess returns decrease with the quintiles. The patterns
are mostly monotonic from the lowest to the highest quintiles. The difference in the average
returns of the highest and lowest quintiles is -3.2 percent for the dollar volume factor. The
difference is statistically significant at the five percent level. In other words, a zero-investment
strategy that longs the lowest dollar volume coins and shorts the highest dollar volume coins
generates about 3 percent excess weekly returns.

Table 5: Volume Factor Returns

This table reports the mean quintile portfolio returns based on the price dollar volume factor. The mean
returns are the time-series averages of weekly value-weighted portfolio excess returns. *, **, *** denote
significance levels at the 10%, 5%, and 1%.

Quintiles
1 2 3 4 5 5-1

PRCVOL Low High

Mean 0.045** 0.028** 0.017 0.018 0.013* -0.032**
t(Mean) (2.438) (2.154) (1.375) (1.456) (1.745) (-2.016)

3.4 Volatility Factors

We analyze the performance of the volatility-related factors: beta, beta squared, idiosyn-
cratic volatility, the standard deviation of returns, the skewness of returns, the kurtosis of
returns, maximum day return, delay, the standard deviation of dollar volume, and Amihud’s
illiquidity measure. Each week, we sort individual cryptocurrencies into quintile portfolios

12
on the value of a given factor. All strategies are rebalanced weekly. The standard deviation
of dollar volume measure generates statistically significant long-short strategy returns, but
the other factors do not. We summarize the insignificant volatility factors in the last part of
the section.
Table 6 presents the results for the portfolios sorted in quintiles for the standard devia-
tion of dollar volume – the only factor out of ten in this group that generates statistically
significant excess returns on the long-short strategies. For the standard deviation of dollar
volume, the average mean excess returns of the portfolios decrease monotonically with the
quintiles and are statistically significant for each quintile except quintile four. The difference
in the average returns of the highest and lowest quintiles is -3.0 percent. In other words, a
zero-investment strategy that longs the lowest dollar volume volatility coins and shorts the
highest dollar volume volatility coins generates about 3 percent excess weekly returns.5

Table 6: Volatility Factor Returns

This table reports the mean quintile portfolio returns based on the standard deviation of dollar volume
factor. The mean returns are the time-series averages of weekly value-weighted portfolio excess returns. *,
**, *** denote significance levels at the 10%, 5%, and 1%.

Quintiles
1 2 3 4 5 5-1

STDPRCVOL Low High

Mean 0.043*** 0.032** 0.021* 0.021 0.013* -0.030**
t(Mean) (2.711) (2.114) (1.687) (1.644) (1.739) (-2.269)

5 For the cross-section of cryptocurrencies, the dollar volume volatility strongly correlates with size. The
reason is that the dollar volume volatility measure is primarily driven by the differences in the price levels
of coins. In the next section, we show that the dollar volume volatility premium can be accounted for by the
cryptocurrency size factor.

13
 Table 7: Insignificant Factor Returns
This table reports the mean quintile portfolio returns based on the insignificant factors. The mean returns
are the time-series averages of weekly value-weighted portfolio excess returns. *, **, *** denote significance
levels at the 10%, 5%, and 1%.
1 2 3 4 5 5-1
Mean 0.018 0.010 0.019* 0.014 0.013* -0.005
AGE
t(Mean) (1.070) (1.034) (1.902) (1.445) (1.732) (-0.358)
Mean 0.016 0.011 0.025** 0.022** 0.022* 0.006
r 8,0
t(Mean) (1.366) (1.266) (2.016) (2.211) (1.740) (0.421)
Mean 0.017* 0.017* 0.006 0.013 0.021* 0.004
r 16,0
t(Mean) (1.693) (1.806) (0.714) (1.300) (1.661) (0.310)
Mean 0.015 0.021** 0.017 0.014 0.008 -0.008
r 50,0
t(Mean) (1.572) (2.167) (1.593) (1.447) (0.680) (-0.764)
Mean 0.032*** 0.027** 0.027* 0.024* 0.018 -0.011
r 100,0
t(Mean) (2.842) (2.556) (1.971) (1.923) (1.376) (-0.804)
Mean 0.014 0.034* 0.017* 0.014 0.013* -0.002
VOL
t(Mean) (1.237) (1.736) (1.695) (1.289) (1.780) (-0.170)
Mean 0.028* 0.035** 0.019 0.001 0.012* -0.016
VOLSCALED
t(Mean) (1.913) (2.347) (1.481) (0.120) (1.699) (-1.332)
Mean 0.019* 0.017 0.020* 0.016 0.006 -0.013
BETA
t(Mean) (1.967) (1.573) (1.817) (1.488) (0.544) (-1.256)
Mean 0.018* 0.024** 0.017 0.015 0.005 -0.012
BETA2
t(Mean) (1.847) (2.074) (1.616) (1.334) (0.484) (-1.191)
Mean 0.014* 0.027** 0.023* 0.007 0.022 0.009
IDIOVOL
t(Mean) (1.894) (2.200) (1.803) (0.564) (1.378) (0.682)
Mean 0.013 0.022** 0.026* 0.020 -0.004 -0.017
RETVOL
t(Mean) (1.549) (2.102) (1.951) (1.163) (-0.281) (-1.237)
Mean 0.011 0.002 0.020* 0.011 0.016 0.005
RETSKEW
t(Mean) (1.206) (0.270) (1.947) (1.015) (1.191) (0.383)
Mean -0.002 0.022** 0.011 0.021** 0.005 0.006
RETKURT
t(Mean) (-0.185) (2.255) (1.135) (2.009) (0.412) (0.647)
Mean 0.012 0.018* 0.013 0.030* 0.006 -0.006
MAXRET
t(Mean) (1.452) (1.691) (1.358) (1.778) (0.388) (-0.441)
Mean 0.014* 0.019* 0.018 0.018 0.012 -0.001
DELAY
t(Mean) (1.876) (1.783) (1.629) (1.338) (1.264) (-0.159)
Mean 0.013* 0.013 0.039** 0.015 0.038* 0.026
DAMIHUD
t(Mean) (1.739) (1.074) (2.159) (1.511) (1.914) (1.478)

14
3.5 Insignificant Factors
In this section, we present the table that summarizes the results for the zero-investment
strategies for the factors that do not generate statistically significant returns. There are
sixteen such factors in total: age; eight-, sixteen-, fifty-, and one hundred-week momentum;
volume, and scaled volume; beta, beta squared, idiosyncratic volatility, the standard devia-
tion of returns, the skewness of returns, the kurtosis of returns, maximum day return, delay,
and the Amihud’s illiquidity measure. Each week, we sort individual cryptocurrencies into
quintile portfolios on the value of a given factor. All strategies are rebalanced weekly.
Table 7 presents the results of the performance of the zero-investment long-short strate-
gies. None of the measures generates statistically significant long-short strategy returns. The
average mean excess returns do not change monotonically with the quintiles. The differences
in the average returns of the highest and lowest quintiles are small and statistically insignifi-
cant. For example, the sixteen-week momentum strategy generates statistically insignificant
excess returns of 0.4 percent per week on the long-short strategy.

4 Cryptocurrency Factors
In this section, we investigate whether the nine cross-sectional cryptocurrency return
predictors that we have identified can be spanned by a small number of factors. We perform
an analysis similar to that of Fama and French (1996). We first show that a one-factor model
with only the coin market return, or the cryptocurrency CAPM, cannot account for most of
the excess returns of the nine strategies. Then, we analyze two-factor models: a two-factor
model that adds the cryptocurrency size factor and a two-factor model that adds the cryp-
tocurrency momentum factor. The two-factor model with the cryptocurrency market factor
and a cryptocurrency size factor can account for the excess returns of five out of the nine
zero-investment strategies but cannot explain any of the momentum related strategies. The
two-factor model with the cryptocurrency market factor and a cryptocurrency momentum
factor can account for the four momentum related strategies but not for any of the other
strategies. Finally, we show that a three-factor model with the cryptocurrency market fac-
tor, a cryptocurrency size factor, and a cryptocurrency momentum factor explains the excess
returns of all nine strategies.
The construction of the cryptocurrency market excess returns is discussed in Section 2.
We construct the cryptocurrency size and momentum factors following the method used by

15
Fama and French (1993). Specifically, for size, each week we split the coins into three size
groups by market capitalization: bottom 30 percent (small, S), middle 40 percent (middle,
M), and top 30 percent (big, B).6 We then form value-weighted portfolios for each of the
three groups. The cryptocurrency size factor (CSMB) is the return difference between the
portfolios of the small and the big size portfolios. We construct the momentum factor
(CMOM) using the three-week momentum.7 Each week, we split the coins into three three-
week momentum groups: bottom 30 percent, middle 40 percent, and top 30 percent. Then,
we form value-weighted portfolios for each of the three three-week momentum groups. The
cryptocurrency momentum factor (CMOM) is the return difference between the top and the
bottom momentum portfolios. In the Appendix, we provide summary statistics for each of
the cryptocurrency factors.
We first consider a one-factor model with only the cryptocurrency market factor, or
the cryptocurrency CAPM. Table 8 presents the results for all the nine significant zero-
investment strategies that we have found in the previous section. The alphas for all of
the zero-investment long-short strategies remain significant. Moreover, the decreases in
magnitude are small compared to the unadjusted excess returns. The average decrease
of the zero-investment strategy alphas for the statistically significant strategies is only 8.61
percent of the orignal values. The strategies have some exposures to the coin market returns.
In particular, the zero-investment long-short strategies based on market capitalization, price,
maximum day price, dollar volume, and standard deviation of dollar volume are significantly
exposed to the coin market excess returns. The strategies based on past returns – one-week
momentum, two-week momentum, three-week momentum, and four-week momentum – are
not significantly exposed to the coin market returns. The average of the absolute value of
the statistically significant betas is 0.39 (with a range of 0.210 for the market capitalization
strategy to 0.592 for the maximum day price). However, for all the strategies, the one-factor
model does not explain a sizable portion of the excess returns, with the zero-investment
strategy R2 s ranging from about zero percent for the one-week momentum strategy to 6.8
percent for the maximum day price.

6 We use market capitalization as our main size measure because of the tradition in the stock market size
literature. The results are robust to using alternative measures of size.
7 We use three-week momentum as our main momentum measure because it generates the largest long-

short spread in the data. The results are qualitatively similar using alternative measures of momentum.

16
 Table 8: Cryptocurrency One-Factor Model

Ri − R f = αi + βC
i
M KT CMKT + i (1)

where CMKT is the cryptocurrency excess market returns. The formation of the quintile portfolios for the
nine significant strategies are discussed in Section 3. The t-statistics are reported in the parentheses. *, **,
*** denote significance levels at the 10%, 5%, and 1%. m.a.e and R¯2 are the mean of the absolute pricing
errors and the average R2 of the five portfolios, respectively.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.031** 0.005 -0.001 0.002 0.001 -0.031**
t(α) (2.346) (0.640) (-0.148) (0.257) (1.400) (-2.284)
MCAP βC M K T 1.208*** 1.074*** 0.990*** 0.984*** 0.998*** -0.210* 0.008 0.571
t(βC M K T ) (10.606) (15.302) (16.023) (18.662) (200.564) (-1.841)
R2 0.305 0.478 0.501 0.576 0.994 0.013
α 0.033** 0.013 -0.011 0.005 0.000 -0.032**
t(α) (2.153) (1.307) (-1.484) (0.502) (0.314) (-2.035)
PRC βC M K T 1.544*** 1.307*** 1.073*** 1.083*** 0.958*** -0.585*** 0.012 0.539
t(βC M K T ) (11.858) (15.688) (16.277) (12.295) (89.475) (-4.299)
R2 0.355 0.490 0.509 0.371 0.969 0.067
α 0.034** 0.010 -0.011 0.008 0.000 -0.034**
t(α) (2.218) (1.009) (-1.459) (0.722) (0.271) (-2.101)
MAXDPRC βC M K T 1.551*** 1.290*** 1.070*** 1.094*** 0.958*** -0.592*** 0.013 0.533
t(βC M K T ) (11.796) (15.753) (16.244) (11.579) (89.511) (-4.310)
R2 0.352 0.492 0.508 0.344 0.969 0.068
α -0.019** -0.014** -0.002 0.029* 0.009 0.028**
t(α) (-2.415) (-2.220) (-0.399) (1.754) (0.774) (2.012)
r 1,0 βC M K T 1.036*** 1.015*** 0.975*** 1.111*** 1.015*** -0.021 0.015 0.437
t(βC M K T ) (15.653) (18.950) (21.158) (7.865) (10.058) (-0.175)
R2 0.489 0.584 0.636 0.195 0.283 0.000
α -0.013 -0.007 -0.002 0.006 0.018 0.031**
t(α) (-1.515) (-1.000) (-0.386) (0.955) (1.649) (2.306)
r 2,0 βC M K T 0.872*** 0.998*** 0.941*** 1.048*** 1.041*** 0.169 0.009 0.484
t(βC M K T ) (11.682) (16.539) (19.940) (20.302) (11.241) (1.468)
R2 0.348 0.517 0.608 0.617 0.330 0.008
α -0.009 -0.011* 0.003 0.007 0.031** 0.040***
t(α) (-1.047) (-1.651) (0.469) (1.273) (2.431) (2.652)
r 3,0 βC M K T 0.904*** 0.994*** 0.955*** 1.037*** 1.005*** 0.101 0.012 0.459
t(βC M K T ) (11.864) (17.958) (15.591) (21.437) (9.318) (0.782)
R2 0.355 0.557 0.487 0.642 0.253 0.002

17
 Table 3 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.009 -0.008 -0.004 0.006 0.014 0.024*
t(α) (-1.333) (-1.071) (-0.770) (0.993) (1.300) (1.906)
r 4,0 βC M K T 0.939*** 0.979*** 0.959*** 1.067*** 1.064*** 0.125 0.008 0.522
t(βC M K T ) (15.586) (15.407) (22.943) (21.056) (11.353) (1.174)
R2 0.487 0.481 0.673 0.634 0.335 0.005
α 0.029* 0.014 0.003 0.004 0.001 -0.028*
t(α) (1.814) (1.389) (0.309) (0.465) (0.924) (-1.758)
PRCVOL βC M K T 1.334*** 1.193*** 1.125*** 1.114*** 0.997*** -0.337** 0.010 0.520
t(βC M K T ) (9.789) (14.131) (14.640) (14.268) (156.767) (-2.454)
R2 0.272 0.438 0.456 0.443 0.990 0.023
α 0.028** 0.017 0.006 0.007 0.001 -0.028**
t(α) (2.118) (1.374) (0.713) (0.689) (0.843) (-2.048)
STDPRCVOL βC M K T 1.223*** 1.292*** 1.192*** 1.209*** 0.993*** -0.230** 0.012 0.524
t(βC M K T ) (10.734) (12.332) (15.557) (14.870) (152.231) (-1.997)
R2 0.310 0.373 0.486 0.463 0.989 0.015

In the last two columns, we report the absolute pricing errors and average R2 s of the five
quintile portfolios for each strategy. We report the mean of the absolute pricing errors, m.a.e,
for each strategy. The mean of the absolute pricing errors is defined as the average of the
absolute value of the alphas for all the five quintile portfolios. In particular, the m.a.e ranges
from 0.8 percent for the market capitalization and the four-week momentum strategies to 1.5
percent for the one-week momentum strategy. The average R2 , R̄2 , is about 50 percent for
most of the strategies, indicating that the model explains substantial fractions of the return
variations of the individual portfolios. In other words, there is strong comovement across
different coins.

18
 Table 9: Cryptocurrency Market and Size Factor Model

Ri − R f = αi + βC
i
M KT CMKT + βCS M B CSM B + i
i
(2)

where CMKT is the cryptocurrency excess market returns and CSM B is the cryptocurrency size factor.
The formation of the quintile portfolios for the nine significant strategies are discussed in Section 3. The t-
statistics are reported in the parentheses. *, **, *** denote significance levels at the 10%, 5%, and 1%. m.a.e
and R¯2 are the mean of the absolute pricing errors and the average R2 of the five portfolios, respectively.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.006 -0.005 -0.006 -0.004 0.001 -0.005
t(α) (1.051) (-0.740) (-0.858) (-0.712) (1.413) (-0.912)
βC M K T 1.036*** 1.004*** 0.944*** 0.946*** 0.998*** -0.037
MCAP t(βC M K T ) (20.427) (17.636) (16.906) (19.744) (199.107) (-0.737) 0.004 0.755
βC S M B 1.343*** 0.548*** 0.359*** 0.300*** -0.001 -1.344***
t(βC S M B ) (32.427) (11.777) (7.873) (7.667) (-0.211) (-32.480)
R2 0.864 0.662 0.598 0.656 0.994 0.808
α 0.019 0.008 -0.017** 0.002 0.001 -0.018
t(α) (1.363) (0.792) (-2.254) (0.147) (0.690) (-1.241)
βC M K T 1.448*** 1.271*** 1.037*** 1.057*** 0.962*** -0.486***
PRC t(βC M K T ) (12.272) (15.661) (16.492) (12.095) (90.547) (-3.929) 0.009 0.584
βC S M B 0.748*** 0.279*** 0.278*** 0.197*** -0.025*** -0.773***
t(βC S M B ) (7.762) (4.206) (5.418) (2.763) (-2.884) (-7.650)
R2 0.478 0.523 0.559 0.390 0.970 0.241
α 0.020 0.004 -0.016** 0.004 0.001 -0.019
t(α) (1.440) (0.463) (-2.176) (0.386) (0.644) (-1.319)
βC M K T 1.454*** 1.253*** 1.036*** 1.069*** 0.962*** -0.493***
MAXDPRC t(βC M K T ) (12.188) (15.761) (16.389) (11.372) (90.569) (-3.941) 0.009 0.577
βC S M B 0.749*** 0.287*** 0.263*** 0.200*** -0.025*** -0.774***
t(βC S M B ) (7.686) (4.424) (5.093) (2.607) (-2.871) (-7.576)
R2 0.474 0.528 0.553 0.361 0.970 0.239
α -0.021*** -0.014** -0.006 0.028* 0.004 0.025*
t(α) (-2.737) (-2.260) (-1.176) (1.661) (0.346) (1.810)
βC M K T 1.019*** 1.012*** 0.948*** 1.103*** 0.980*** -0.039
r 1,0 t(βC M K T ) (15.462) (18.767) (21.743) (7.751) (9.850) (-0.327) 0.015 0.455
βC S M B 0.132** 0.021 0.208*** 0.068 0.274*** 0.141
t(βC S M B ) (2.460) (0.475) (5.826) (0.584) (3.369) (1.459)
R2 0.501 0.584 0.679 0.196 0.314 0.008

19
Table 9 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.016* -0.010 -0.003 0.004 0.014 0.030**
t(α) (-1.841) (-1.438) (-0.611) (0.696) (1.325) (2.230)
βC M K T 0.852*** 0.977*** 0.932*** 1.037*** 1.016*** 0.164
r 2,0 t(βC M K T ) (11.474) (16.416) (19.727) (20.098) (11.031) (1.415) 0.009 0.496
βC S M B 0.152** 0.160*** 0.068* 0.084** 0.190** 0.039
t(βC S M B ) (2.498) (3.295) (1.751) (1.982) (2.528) (0.408)
R2 0.363 0.536 0.613 0.623 0.347 0.009
α -0.013 -0.014** 0.000 0.006 0.026** 0.039**
t(α) (-1.445) (-2.170) (0.058) (1.085) (2.093) (2.566)
βC M K T 0.880*** 0.972*** 0.935*** 1.030*** 0.975*** 0.095
r 3,0 t(βC M K T ) (11.668) (17.920) (15.447) (21.211) (9.097) (0.729) 0.012 0.477
βC S M B 0.186*** 0.168*** 0.159*** 0.055 0.234*** 0.048
t(βC S M B ) (3.012) (3.782) (3.207) (1.379) (2.672) (0.456)
R2 0.377 0.581 0.507 0.645 0.274 0.003
α -0.012* -0.012 -0.006 0.005 0.011 0.023*
t(α) (-1.716) (-1.615) (-1.159) (0.790) (0.997) (1.831)
βC M K T 0.921*** 0.952*** 0.946*** 1.059*** 1.041*** 0.120
r 4,0 t(βC M K T ) (15.418) (15.348) (22.851) (20.834) (11.143) (1.120) 0.009 0.537
βC S M B 0.141*** 0.204*** 0.100*** 0.063 0.179** 0.038
t(βC S M B ) (2.893) (4.023) (2.967) (1.516) (2.344) (0.431)
R2 0.503 0.512 0.684 0.637 0.349 0.006
α 0.008 0.006 -0.006 -0.001 0.001 -0.007
t(α) (0.646) (0.659) (-0.728) (-0.101) (1.110) (-0.575)
βC M K T 1.190*** 1.140*** 1.065*** 1.078*** 0.998*** -0.193*
PRCVOL t(βC M K T ) (11.093) (14.422) (15.499) (14.236) (156.423) (-1.780) 0.004 0.623
βC S M B 1.116*** 0.408*** 0.465*** 0.276*** -0.008 -1.124***
t(βC S M B ) (12.738) (6.324) (8.281) (4.460) (-1.483) (-12.726)
R2 0.555 0.514 0.571 0.483 0.990 0.402
α 0.011 0.007 -0.001 -0.001 0.001 -0.010
t(α) (1.018) (0.581) (-0.133) (-0.134) (1.124) (-0.926)
βC M K T 1.102*** 1.221*** 1.140*** 1.156*** 0.995*** -0.107
STDPRCVOL t(βC M K T ) (12.379) (12.623) (16.116) (15.315) (152.720) (-1.191) 0.004 0.632
βC S M B 0.946*** 0.550*** 0.403*** 0.416*** -0.012** -0.957***
t(βC S M B ) (13.011) (6.964) (6.967) (6.749) (-2.176) (-13.040)
R2 0.586 0.473 0.568 0.545 0.989 0.409

20
 We then consider a two-factor model with the cryptocurrency market factor and the
cryptocurrency size factor. Table 9 presents the results for all nine strategies. The long-
short alphas for most of them, with the exception of the momentum strategies, are no longer
significant. For example, the absolute value of the alpha for dollar volume drops from 2.8
percent under the one-factor model to an insignificant 0.7 percent under the two-factor
model. All non-momentum strategies have significant exposures to the cryptocurrency size
factor. Among the non-momentum strategies, the absolute values of their size factor loadings
range from 1.344 for the market capitalization factor to 0.773 for the last day price factor.
In other words, the small coins are also more illiquid and have lower trading volume, similar
to results in the stock market. Many strategies have significant loadings on CMKT, with the
exception of the market capitalization, the standard deviation of dollar volume, the one-,
two-, three- and four-week momentum factors. For all non-momentum strategies, the model
explains substantial fractions of the return variations beyond what the coin market factor
explains. Among the non-momentum strategies, the zero-investment long-short strategy
R2 s range from 23.9 percent for the strategy based on the maximum day price factor to
more than 80 percent for the strategy based on the market capitalization factor. However,
this two-factor model based on the cryptocurrency market and size falls short in explaining
any of the momentum based strategies. The alphas on the momentum based strategies are
all statistically significant adjusting for this two-factor model. Compared to those of the
one-factor model, the means of absolute pricing errors decrease dramatically for the non-
momentum strategies. For example, the m.a.e of the dollar volume strategy reduces from
1.0 percent in the one-factor model to 0.4 percent in the two-factor model controlling for
the cryptocurrency market and size factors – a 60 percent decrease. The means of absolute
pricing errors do not materially change for the momentum strategies controlling for the
two-factor model.
We next consider an alternative two-factor model by combining the cryptocurrency mar-
ket factor and the cryptocurrency momentum factor. Table 10 presents the results for all
nine zero-investment long-short strategies adjusting for the alternative two-factor model.
This two-factor model performs well in capturing the excess returns of the four momen-
tum factors – one-, two-, three-, and four-week momentum factors. After controlling for
this alternative two-factor model, the alphas for all four momentum strategies are no longer
statistically significant. For example, the alpha of the one-week momentum strategy drops
from 2.8 percent under the one-factor model to 0.7 percent under this alternative two-factor
model. All four momentum strategies have statistically significant exposures to the momen-

21
tum factor. For these four strategies, their momentum factor loadings range from 0.582
for the one-week momentum to 0.985 for the three-week momentum. All non-momentum
strategies, with the exception of dollar volume, have significant exposures to the market. On
the other hand, none of the momentum strategies significantly exposes to the cryptocurrency
market factor. For the momentum strategies, this alternative two-factor model explains a
substantial fraction of the return variations in contrast to the market one-factor model or the
market and size two-factor model. The zero-investment strategy R2 s range from 23.2 percent
for the one-week momentum to 56.1 percent for the three-week momentum. However, the
model underperforms in explaining the return variations of the non-momentum strategies
compared to the two-factor model with the cryptocurrency market and the cryptocurrency
size factors. The alphas of the non-momentum strategies, with the exception of the dollar
volume strategy, remain statistically significant. Compared to the one-factor model, the
means of absolute pricing errors largely decrease for the momentum factors. For example,
the m.a.e of the two-week momentum strategy reduces from 0.9 percent in the one-factor
model to 0.2 percent in the two-factor model.
Finally, we consider a three-factor model that combines the cryptocurrency market, size,
and momentum factors. Table 11 presents the results for all nine strategies. Adjusted
for the cryptocurrency three-factor model, none of the alphas for the nine strategies remains
statistically significant. We now turn to the discussion of exposures to the three factors. The
one-week momentum long-short strategy is statistically significantly exposed to both the size
and momentum factors. The market capitalization and standard deviation of dollar volume
zero-investment long-short strategies are statistically significantly exposed to the size factor
only but not to the market or momentum factors. The two-, three- and four-week momentum
zero-investment long-short strategies are statistically significantly exposed to the momentum
factor only but not to the market or size factors. The following strategies are statistically
significantly exposed to both the market and size factors: price, maximum day price, and
dollar volume. None of the strategies is exposed to the market factor only. In other words,
both size and momentum are important in explaining the cross-section of expected returns
of cryptocurrencies. Compared to the one-factor model, the means of absolute pricing errors
largely decrease for all of the nine strategies.

22
 Table 10: Cryptocurrency Market and Momentum Factor Model

Ri − R f = αi + βC
i
M KT CMKT + βCS M B CMOM + i
i
(3)

where CMKT is the cryptocurrency excess market returns and CMOM is the cryptocurrency momentum
factor. The formation of the quintile portfolios for the nine significant strategies are discussed in Section
3. The t-statistics are reported in the parentheses. *, **, *** denote significance levels at the 10%, 5%,
and 1%. m.a.e and R¯2 are the mean of the absolute pricing errors and the average R2 of the five portfolios,
respectively.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.034** 0.004 0.001 0.003 0.001 -0.033**
t(α) (2.503) (0.509) (0.179) (0.458) (1.341) (-2.444)
βC M K T 1.218*** 1.071*** 0.991*** 0.989*** 0.998*** -0.219*
MCAP t(βC M K T ) (10.657) (15.184) (15.956) (18.693) (199.519) (-1.920) 0.009 0.572
βC M O M -0.076 0.027 -0.007 -0.036 0.000 0.076
t(βC M O M ) (-1.040) (0.609) (-0.179) (-1.067) (0.155) (1.047)
R2 0.308 0.478 0.501 0.578 0.994 0.017
α 0.035** 0.011 -0.012 0.005 0.001 -0.034**
t(α) (2.257) (1.082) (-1.490) (0.520) (0.888) (-2.088)
βC M K T 1.552*** 1.300*** 1.072*** 1.084*** 0.961*** -0.591***
PRC t(βC M K T ) (11.870) (15.556) (16.178) (12.244) (90.786) (-4.319) 0.013 0.540
βC M O M -0.063 0.055 0.008 -0.008 -0.020*** 0.043
t(βC M O M ) (-0.755) (1.040) (0.188) (-0.143) (-2.979) (0.491)
R2 0.356 0.492 0.509 0.371 0.970 0.068
α 0.036** 0.007 -0.012 0.008 0.001 -0.035**
t(α) (2.290) (0.761) (-1.468) (0.689) (0.850) (-2.124)
βC M K T 1.557*** 1.282*** 1.069*** 1.094*** 0.961*** -0.596***
MAXDPRC t(βC M K T ) (11.792) (15.619) (16.145) (11.512) (90.851) (-4.316) 0.013 0.534
βC M O M -0.050 0.062 0.009 0.006 -0.020*** 0.030
t(βC M O M ) (-0.600) (1.189) (0.203) (0.095) (-3.005) (0.344)
R2 0.353 0.495 0.508 0.344 0.970 0.068
α -0.013* -0.009 0.001 0.027 -0.006 0.007
t(α) (-1.651) (-1.507) (0.118) (1.570) (-0.527) (0.552)
βC M K T 1.057*** 1.031*** 0.985*** 1.102*** 0.963*** -0.094
r 1,0 t(βC M K T ) (16.404) (19.671) (21.553) (7.772) (10.364) (-0.905) 0.011 0.473
βC M O M -0.168*** -0.125*** -0.077*** 0.070 0.414*** 0.582***
t(βC M O M ) (-4.089) (-3.762) (-2.664) (0.771) (7.010) (8.781)
R2 0.520 0.606 0.646 0.196 0.399 0.232

23
Table 10 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.003 -0.001 0.001 0.001 0.003 0.006
t(α) (-0.350) (-0.134) (0.157) (0.088) (0.346) (0.569)
βC M K T 0.908*** 1.019*** 0.951*** 1.030*** 0.990*** 0.082
r 2,0 t(βC M K T ) (13.062) (17.485) (20.365) (20.644) (11.736) (0.876) 0.002 0.542
βC M O M -0.286*** -0.170*** -0.083*** 0.145*** 0.399*** 0.685***
t(βC M O M ) (-6.476) (-4.579) (-2.799) (4.583) (7.428) (11.435)
R2 0.440 0.553 0.620 0.646 0.450 0.344
α 0.004 -0.002 0.007 0.005 0.008 0.004
t(α) (0.462) (-0.399) (0.980) (0.885) (0.769) (0.435)
βC M K T 0.950*** 1.023*** 0.968*** 1.030*** 0.926*** -0.024
r 3,0 t(βC M K T ) (14.006) (20.119) (15.939) (21.324) (10.364) (-0.276) 0.005 0.553
βC M O M -0.362*** -0.229*** -0.103*** 0.059* 0.623*** 0.985***
t(βC M O M ) (-8.386) (-7.095) (-2.663) (1.918) (10.971) (18.005)
R2 0.494 0.630 0.501 0.647 0.493 0.561
α 0.000 -0.000 -0.001 0.004 -0.004 -0.004
t(α) (0.031) (-0.013) (-0.141) (0.639) (-0.376) (-0.422)
βC M K T 0.973*** 1.006*** 0.970*** 1.060*** 1.001*** 0.028
r 4,0 t(βC M K T ) (17.809) (16.732) (23.553) (20.932) (12.419) (0.373) 0.002 0.592
βC M O M -0.266*** -0.218*** -0.085*** 0.057* 0.495*** 0.760***
t(βC M O M ) (-7.652) (-5.698) (-3.249) (1.756) (9.648) (15.751)
R2 0.583 0.540 0.686 0.638 0.513 0.496
α 0.011 0.005 0.013 0.007 -0.003 -0.014
t(α) (1.233) (0.980) (1.304) (0.944) (-0.314) (-1.114)
βC M K T 1.137*** 0.888*** 1.061*** 1.006*** 1.114*** -0.023
PRCVOL t(βC M K T ) (15.660) (19.048) (12.723) (16.897) (14.675) (-0.227) 0.008 0.509
βC M O M -0.235*** -0.145*** -0.025 0.099*** 0.299*** 0.534***
t(βC M O M ) (-5.099) (-4.895) (-0.467) (2.611) (6.184) (8.252)
R2 0.505 0.595 0.389 0.542 0.514 0.211
α 0.029** 0.012 0.008 0.006 0.001 -0.028**
t(α) (2.118) (0.954) (0.885) (0.629) (0.912) (-2.044)
βC M K T 1.225*** 1.274*** 1.198*** 1.208*** 0.994*** -0.232**
STDPRCVOL t(βC M K T ) (10.698) (12.205) (15.581) (14.775) (151.535) (-2.002) 0.011 0.527
βC M O M -0.016 0.139** -0.047 0.012 -0.002 0.015
t(βC M O M ) (-0.226) (2.086) (-0.956) (0.240) (-0.445) (0.198)
R2 0.311 0.383 0.488 0.464 0.989 0.015

24
 Table 11: Cryptocurrency Three-Factor Model

Ri − R f = αi + βC
i
M KT CMKT + βCS M B CSM B + βC H M L CMOM + i
i i
(4)

where CMKT is the cryptocurrency excess market returns, CSM B is the cryptocurrency size factor, and
CMOM is the cryptocurrency momentum factor. The formation of the quintile portfolios for the nine
significant strategies are discussed in Section 3. The t-statistics are reported in the parentheses. *, **, ***
denote significance levels at the 10%, 5%, and 1%. m.a.e is the mean of the absolute pricing errors. R¯2 is
the average R2 of the five portfolios.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.007 -0.007 -0.006 -0.003 0.001 -0.007
t(α) (1.224) (-0.974) (-0.865) (-0.528) (1.354) (-1.091)
βC M K T 1.040*** 0.998*** 0.943*** 0.949*** 0.998*** -0.042
t(βC M K T ) (20.436) (17.487) (16.802) (19.735) (198.002) (-0.818)
MCAP βC S M B 1.341*** 0.550*** 0.359*** 0.298*** -0.001 -1.342*** 0.005 0.756
t(βC S M B ) (32.355) (11.833) (7.856) (7.620) (-0.205) (-32.409)
βC M O M -0.032 0.045 0.005 -0.026 0.000 0.032
t(βC M O M ) (-0.981) (1.256) (0.133) (-0.857) (0.146) (0.997)
R2 0.865 0.664 0.598 0.657 0.994 0.809
α 0.020 0.005 -0.017** 0.002 0.002 -0.019
t(α) (1.435) (0.531) (-2.290) (0.149) (1.309) (-1.258)
βC M K T 1.453*** 1.262*** 1.035*** 1.058*** 0.964*** -0.488***
t(βC M K T ) (12.254) (15.517) (16.370) (12.031) (92.059) (-3.927)
PRC βC S M B 0.746*** 0.282*** 0.279*** 0.197*** -0.026*** -0.772*** 0.009 0.585
t(βC S M B ) (7.722) (4.260) (5.423) (2.754) (-3.065) (-7.620)
βC M O M -0.038 0.065 0.017 -0.002 -0.021*** 0.017
t(βC M O M ) (-0.511) (1.254) (0.427) (-0.029) (-3.154) (0.221)
R2 0.478 0.526 0.560 0.390 0.971 0.242
α 0.021 0.002 -0.017** 0.004 0.002 -0.020
t(α) (1.476) (0.175) (-2.214) (0.337) (1.269) (-1.302)
βC M K T 1.458*** 1.243*** 1.034*** 1.067*** 0.964*** -0.494***
t(βC M K T ) (12.152) (15.616) (16.268) (11.292) (92.112) (-3.924)
MAXDPRC βC S M B 0.748*** 0.291*** 0.264*** 0.201*** -0.026*** -0.774*** 0.009 0.578
t(βC S M B ) (7.652) (4.489) (5.099) (2.609) (-3.053) (-7.552)
βC M O M -0.026 0.072 0.017 0.012 -0.021*** 0.005
t(βC M O M ) (-0.341) (1.421) (0.428) (0.206) (-3.180) (0.060)
R2 0.474 0.532 0.553 0.361 0.971 0.239

25
Table 11 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.015** -0.010 -0.003 0.025 -0.012 0.003
t(α) (-1.970) (-1.535) (-0.657) (1.470) (-1.080) (0.274)
βC M K T 1.041*** 1.029*** 0.958*** 1.093*** 0.924*** -0.117
t(βC M K T ) (16.198) (19.486) (22.122) (7.650) (10.171) (-1.126)
r 1,0 βC S M B 0.124** 0.014 0.204*** 0.072 0.297*** 0.173** 0.013 0.491
t(βC S M B ) (2.360) (0.328) (5.776) (0.617) (4.014) (2.043)
βC M O M -0.164*** -0.125*** -0.071** 0.072 0.424*** 0.587***
t(βC M O M ) (-4.022) (-3.739) (-2.581) (0.796) (7.378) (8.914)
R2 0.531 0.606 0.687 0.198 0.435 0.245
α -0.006 -0.004 -0.000 -0.001 -0.001 0.005
t(α) (-0.677) (-0.572) (-0.068) (-0.218) (-0.073) (0.430)
βC M K T 0.890*** 0.999*** 0.943*** 1.017*** 0.962*** 0.072
t(βC M K T ) (12.847) (17.352) (20.136) (20.443) (11.529) (0.764)
r 2,0 βC S M B 0.136** 0.151*** 0.063* 0.092** 0.212*** 0.076 0.002 0.554
t(βC S M B ) (2.415) (3.224) (1.656) (2.260) (3.123) (0.986)
βC M O M -0.282*** -0.165*** -0.081*** 0.148*** 0.405*** 0.687***
t(βC M O M ) (-6.429) (-4.522) (-2.736) (4.711) (7.679) (11.466)
R2 0.452 0.571 0.624 0.653 0.470 0.347
α 0.000 -0.006 0.004 0.004 0.003 0.002
t(α) (0.053) (-0.924) (0.565) (0.680) (0.273) (0.236)
βC M K T 0.928*** 1.002*** 0.948*** 1.022*** 0.890*** -0.037
t(βC M K T ) (13.819) (20.117) (15.779) (21.089) (10.160) (-0.432)
r 3,0 βC S M B 0.166*** 0.155*** 0.153*** 0.058 0.268*** 0.102 0.003 0.570
t(βC S M B ) (3.039) (3.829) (3.130) (1.470) (3.758) (1.451)
βC M O M -0.356*** -0.224*** -0.098** 0.061** 0.632*** 0.988***
t(βC M O M ) (-8.385) (-7.114) (-2.574) (1.983) (11.399) (18.089)
R2 0.512 0.651 0.519 0.650 0.519 0.564
α -0.002 -0.004 -0.003 0.003 -0.008 -0.005
t(α) (-0.358) (-0.557) (-0.532) (0.418) (-0.809) (-0.591)
βC M K T 0.956*** 0.981*** 0.957*** 1.051*** 0.974*** 0.018
t(βC M K T ) (17.645) (16.687) (23.437) (20.700) (12.221) (0.234)
r 4,0 βC S M B 0.127*** 0.192*** 0.096*** 0.066 0.206*** 0.079 0.004 0.606
t(βC S M B ) (2.876) (4.019) (2.883) (1.599) (3.174) (1.273)
βC M O M -0.262*** -0.212*** -0.082*** 0.059* 0.501*** 0.763***
t(βC M O M ) (-7.632) (-5.690) (-3.171) (1.827) (9.943) (15.810)
R2 0.596 0.567 0.696 0.642 0.531 0.499

26
 Table 11 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α 0.006 0.002 -0.005 0.000 0.001 -0.005
t(α) (0.433) (0.193) (-0.582) (0.024) (1.151) (-0.362)
βC M K T 1.181*** 1.125*** 1.069*** 1.082*** 0.998*** -0.183*
t(βC M K T ) (10.968) (14.298) (15.483) (14.221) (155.617) (-1.688)
PRCVOL βC S M B 1.120*** 0.415*** 0.463*** 0.274*** -0.008 -1.128*** 0.003 0.626
t(βC S M B ) (12.770) (6.471) (8.236) (4.425) (-1.492) (-12.759)
βC M O M 0.069 0.116** -0.029 -0.030 -0.001 -0.070
t(βC M O M ) (1.014) (2.329) (-0.665) (-0.624) (-0.324) (-1.025)
R2 0.557 0.525 0.572 0.484 0.990 0.405
α 0.010 0.001 0.000 -0.002 0.001 -0.009
t(α) (0.945) (0.067) (0.017) (-0.239) (1.207) (-0.848)
βC M K T 1.100*** 1.200*** 1.145*** 1.152*** 0.995*** -0.105
t(βC M K T ) (12.290) (12.499) (16.107) (15.193) (152.008) (-1.160)
STDPRCVOL βC S M B 0.947*** 0.559*** 0.401*** 0.417*** -0.012** -0.958*** 0.003 0.635
t(βC S M B ) (12.986) (7.143) (6.923) (6.757) (-2.194) (-13.017)
βC M O M 0.015 0.157** -0.034 0.026 -0.002 -0.017
t(βC M O M ) (0.258) (2.582) (-0.747) (0.545) (-0.540) (-0.295)
R2 0.586 0.486 0.569 0.545 0.989 0.409

5 Other Results
In this section, we describe four sets of additional results: using Bitcoin for short port-
folios, the analysis of the Fama-MacBeth regressions, using the stock market factors, and
hedging unpriced risks.

5.1 Using Bitcoin for Short Portfolios

One concern with constructing the zero-investment strategies in cryptocurrencies is that
shorting is not readily available for most of the coins. Table 12 presents the analysis of the
strategies that short Bitcoin rather than shorting the relevant factor quintiles. The results
are qualitatively similar to those of Section 4. The reason is that most of the relevant factor
quintiles behave similarly to Bitcoin. The exceptions are the momentum factors for which
the lowest quintiles behave differently from Bitcoin. As a result, the mean returns of the
one-, two-, and four-week momentum strategies are no longer statistically significant, and
the returns to the Bitcoin zero-investment strategies are somewhat different. We also report

27
the results adjusting for the one- and three-factor cryptocurrency models. For the one-
factor cryptocurrency model, the only major difference is that the alpha of the dollar volume
strategy is no longer statistically significant. Consistent with the previous section, none
of the alphas remain statistically significant controlling for the three-factor cryptocurrency
model.

5.2 Additional Cross-Sectional Results

We first present the results of the cross-sectional regressions using the Fama-MacBeth
method in Table 13. We only report the factors that form successful long-short strategies.
Panel A shows the results for the size related factors. All of them are individually statistically
significant but not jointly significant. This is consistent with the fact that these factors are
highly correlated. Panel B shows the results for the volume related factors. The dollar
volume factor is individually statistically significant. Panel C shows the results for the
volatility factor, which is statistically significant. Panel D shows that the momentum factors
are not statistically significant in the Fama-MacBeth regressions. This is different from what
we have found in the previous section for the value-weighted portfolio strategies. A potential
reason for this discrepancy is that, in essence, the Fama-MacBeth regressions consider each
observation equally and thus are close to strategies formed on equally weighted portfolios. In
Panel E, we show that the momentum strategies perform strongly for the larger coins, defined
as coins with more than 10 million dollar market capitalization. We further confirm this in
the Appendix. We double sort first on market capitalization into two groups at the median.
Then, within each size group, we sort on the past three-week returns into five groups. We
find that the long-short momentum strategy in the below median size group only generates
0.6 percent weekly returns which is not statistically significant. In contrast, the long-short
momentum strategy in the above median size group generates statistically significant 4.2
percent weekly returns. This implies that the momentum strategy works better for the
larger coins in the cryptocurrency market. This is in sharp contrast to the equity market
where momentum strategies work better among smaller stocks (see Hong et al., 2000).

28
 Table 12: Bitcoin for Short Portfolios
CMKT is the cryptocurrency excess market returns, CSM B is the cryptocurrency size factor, and CMOM
is the cryptocurrency momentum factor. The formation of nine quintiles are discussed in Section 3. The
t-statistics are reported in the parentheses. *, **, *** denote significance levels at the 10%, 5%, and 1%.

Cons t CMKT t CSMB t CMOM t R2

Mean 0.034** (2.423)
MCAP C-1 0.031** (2.216) 0.141 (1.464) 0.008
C-3 0.004 (0.601) 0.071* (1.719) 1.431*** (33.786) -0.024 (-0.716) 0.819
Mean 0.039** (2.295)
PRC C-1 0.031* (1.864) 0.397*** (3.531) 0.046
C-3 0.015 (0.978) 0.357*** (3.542) 0.850*** (8.237) -0.025 (-0.306) 0.247
Mean 0.040** (2.358)
MAXDPRC C-1 0.032* (1.924) 0.405*** (3.571) 0.047
C-3 0.016 (1.018) 0.363*** (3.572) 0.852*** (8.182) -0.012 (-0.151) 0.245
Mean 0.009 (0.694)
r 1,0 C-1 0.007 (0.559) 0.085 (0.991) 0.004
C-3 -0.016 (-1.345) 0.027 (0.345) 0.378*** (4.804) 0.428*** (6.949) 0.215
Mean 0.018 (1.498)
r 2,0 C-1 0.016 (1.321) 0.102 (1.277) 0.006
C-3 -0.005 (-0.439) 0.050 (0.684) 0.295*** (3.986) 0.411*** (7.083) 0.205
Mean 0.030** (2.248)
r 3,0 C-1 0.036*** (2.607) 0.092 (0.800) 0.003
C-3 0.006 (0.506) 0.004 (0.054) 0.348*** (4.536) 0.636*** (10.602) 0.338
Mean 0.014 (1.204)
r 4,0 C-1 0.012 (1.000) 0.122 (1.505) 0.009
C-3 -0.012 (-1.109) 0.061 (0.869) 0.290*** (4.051) 0.506*** (9.027) 0.277
Mean 0.032* (1.896)
PRCVOL C-1 0.027 (1.607) 0.254** (2.199) 0.018
C-3 0.001 (0.041) 0.186** (2.067) 1.209*** (13.087) 0.079 (1.090) 0.412
Mean 0.030** (2.109)
STDPRCVOL C-1 0.027* (1.853) 0.187* (1.908) 0.014
C-3 0.005 (0.468) 0.133* (1.751) 1.033*** (13.304) 0.023 (0.379) 0.417

29
 Table 13: Fama-MacBeth Cross-Sectional Regression
This table reports the Fama-MacBeth regression results. Each factor is first sorted into five portfolios at
the end of each week. Panel A, B, C, and D are based on the sample of coins with market capitalizations
of more than 1 million dollars. Panel E is based on the sample of coins with market capitalizations of more
than 10 million dollars. The t-statistics of the coefficient estimates are reported in the parentheses.*, **, ***
denote significance levels at the 10%, 5%, and 1%.

r ett +1
MCAP -0.011* -0.000
(-1.890) (-0.210)
PRC -0.008*** -0.027
Panel A
(-3.120) (-1.136)
MAXDPRC -0.008*** 0.019
(-3.256) (0.802)
PRCVOL -0.006**
Panel B
(-2.546)
> 1 mil STDPRCVOL -0.007***
Panel C
(-2.894)
r 1,0 -0.002 -0.004
(-0.692) (-1.427)
r 2,0 0.000 0.003
(0.020) (0.751)
Panel D
r 3,0 -0.000 -0.002
(-0.148) (-0.499)
r 4,0 -0.001 0.000
(-0.274) (0.107)
r 1,0 0.005* -0.006
(1.668) (-1.520)
r 2,0 0.007** 0.005
(2.095) (0.866)
> 10 mil Panel E
r 3,0 0.007** 0.006
(1.978) (1.086)
r 4,0 0.003 -0.000
(0.693) (-0.061)

30
 Table 14: Fama-French Three-Factor Model

Ri − R f = αi + βiM KT MKT + βSi M B SM B + βiH M L H M L + i (5)

where MKT is the excess stock market returns, SM B is the Fama-French size factor, and H M L is the Fama-
French value factor. The formation of the quintile portfolios for the nine significant strategies are discussed
in Section 3. The t-statistics are reported in the parentheses. *, **, *** denote significance levels at the
10%, 5%, and 1%. m.a.e and R¯2 are the mean of the absolute pricing errors and the average R2 of the five
portfolios.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.044*** 0.017 0.013 0.012 0.012 -0.032**
t(α) (2.737) (1.469) (1.220) (1.278) (1.626) (-2.372)
βM K T 0.859 0.981 0.311 0.615 0.404 -0.455
t(β M K T ) (0.976) (1.573) (0.551) (1.178) (1.002) (-0.615)
MCAP βS M B -0.298 -0.966 0.124 -0.412 -0.682 -0.384 0.020 0.008
t(βS M B ) (-0.210) (-0.961) (0.136) (-0.490) (-1.050) (-0.323)
βH M L -1.398 0.589 -0.434 -0.642 -0.222 1.176
t(β H M L ) (-1.014) (0.603) (-0.491) (-0.786) (-0.352) (1.016)
R2 0.008 0.013 0.002 0.008 0.007 0.007
α 0.049** 0.027** 0.000 0.017 0.011 -0.037**
t(α) (2.551) (1.999) (0.026) (1.295) (1.560) (-2.256)
βM K T 1.254 0.503 0.624 1.125 0.396 -0.858
t(β M K T ) (1.202) (0.669) (1.029) (1.578) (1.010) (-0.944)
PRC βS M B -1.295 -1.512 -0.403 -0.837 -0.592 0.702 0.021 0.008
t(βS M B ) (-0.770) (-1.249) (-0.412) (-0.729) (-0.938) (0.480)
βH M L -1.176 0.396 -0.395 1.071 -0.131 1.045
t(β H M L ) (-0.720) (0.337) (-0.416) (0.960) (-0.213) (0.734)
R2 0.008 0.007 0.005 0.014 0.006 0.006
α 0.050*** 0.024* 0.001 0.020 0.011 -0.039**
t(α) (2.603) (1.793) (0.051) (1.457) (1.553) (-2.319)
βM K T 1.180 0.499 0.609 1.111 0.395 -0.785
t(β M K T ) (1.122) (0.673) (1.006) (1.482) (1.008) (-0.855)
MAXDPRC βS M B -1.565 -1.195 -0.381 -0.636 -0.592 0.972 0.021 0.007
t(βS M B ) (-0.924) (-1.001) (-0.391) (-0.526) (-0.938) (0.658)
βH M L -1.207 0.293 -0.257 1.083 -0.131 1.076
t(β H M L ) (-0.733) (0.252) (-0.271) (0.923) (-0.213) (0.749)
R2 0.008 0.005 0.004 0.012 0.006 0.006

31
Table 14 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.008 -0.003 0.009 0.042** 0.021 0.029**
t(α) (-0.732) (-0.321) (0.953) (2.306) (1.495) (2.094)
βM K T 0.766 0.472 0.367 1.600 -0.381 -1.148
t(β M K T ) (1.291) (0.883) (0.745) (1.592) (-0.498) (-1.522)
r 1,0 βS M B 0.047 -1.340 -0.775 1.171 -1.934 -1.981 0.017 0.014
t(βS M B ) (0.049) (-1.558) (-0.978) (0.724) (-1.568) (-1.631)
βH M L -1.394 -0.381 -0.517 2.795* -0.588 0.806
t(β H M L ) (-1.499) (-0.456) (-0.671) (1.777) (-0.490) (0.682)
R2 0.016 0.011 0.006 0.026 0.013 0.027
α -0.004 0.005 0.009 0.018* 0.029** 0.034**
t(α) (-0.402) (0.457) (0.965) (1.827) (2.218) (2.498)
βM K T 0.856 0.373 -0.020 0.186 -0.190 -1.046
t(β M K T ) (1.442) (0.665) (-0.041) (0.346) (-0.261) (-1.414)
r 2,0 βS M B -0.139 0.308 -1.085 -1.294 -1.933* -1.794 0.013 0.010
t(βS M B ) (-0.146) (0.341) (-1.389) (-1.497) (-1.654) (-1.505)
βH M L -0.783 -0.125 -0.640 0.416 -1.140 -0.357
t(β H M L ) (-0.842) (-0.142) (-0.843) (0.496) (-1.003) (-0.308)
R2 0.011 0.003 0.010 0.010 0.015 0.022
α 0.001 0.001 0.014 0.018* 0.043*** 0.042***
t(α) (0.130) (0.087) (1.373) (1.924) (2.961) (2.799)
βM K T 0.436 0.094 0.615 0.892* -0.798 -1.234
t(β M K T ) (0.714) (0.174) (1.115) (1.718) (-0.997) (-1.507)
r 3,0 βS M B 1.219 -0.373 -0.484 -0.891 -1.864 -3.083** 0.015 0.009
t(βS M B ) (1.241) (-0.431) (-0.545) (-1.065) (-1.446) (-2.338)
βH M L -0.474 -0.520 -0.137 0.247 -0.177 0.297
t(β H M L ) (-0.496) (-0.619) (-0.158) (0.304) (-0.141) (0.232)
R2 0.011 0.002 0.005 0.014 0.015 0.038
α 0.002 0.003 0.007 0.017* 0.026* 0.024**
t(α) (0.210) (0.265) (0.843) (1.749) (1.969) (1.975)
βM K T 0.206 0.711 0.332 0.717 -0.551 -0.757
t(β M K T ) (0.379) (1.251) (0.706) (1.331) (-0.750) (-1.121)
r 4,0 βS M B 0.623 -0.071 -0.850 -0.977 -1.752 -2.375** 0.011 0.009
t(βS M B ) (0.712) (-0.077) (-1.120) (-1.126) (-1.479) (-2.183)
βH M L -0.027 -0.279 0.405 -0.338 -1.659 -1.633
t(β H M L ) (-0.031) (-0.314) (0.550) (-0.401) (-1.441) (-1.544)
R2 0.003 0.007 0.007 0.010 0.020 0.036

32
 Table 14 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α 0.042** 0.026** 0.015 0.017 0.012 -0.030*
t(α) (2.222) (1.987) (1.190) (1.366) (1.607) (-1.826)
βM K T 1.691 0.759 1.009 0.495 0.393 -1.299
t(β M K T ) (1.648) (1.047) (1.506) (0.733) (0.973) (-1.453)
PRCVOL βS M B -1.596 -0.898 -0.803 -0.098 -0.678 0.918 0.022 0.008
t(βS M B ) (-0.965) (-0.769) (-0.744) (-0.090) (-1.043) (0.638)
βH M L -0.563 -0.855 -0.105 -0.509 -0.232 0.331
t(β H M L ) (-0.350) (-0.752) (-0.101) (-0.482) (-0.368) (0.236)
R2 0.012 0.007 0.009 0.003 0.007 0.009
α 0.040** 0.029* 0.019 0.020 0.012 -0.029**
t(α) (2.508) (1.902) (1.491) (1.528) (1.601) (-2.098)
βM K T 1.302 1.322 0.894 0.721 0.393 -0.909
t(β M K T ) (1.475) (1.556) (1.300) (1.007) (0.978) (-1.215)
STDPRCVOL βS M B -1.251 -1.581 -1.137 -0.172 -0.678 0.573 0.024 0.009
t(βS M B ) (-0.880) (-1.156) (-1.027) (-0.149) (-1.047) (0.475)
βH M L -0.517 -0.734 -0.667 -0.417 -0.221 0.296
t(β H M L ) (-0.374) (-0.552) (-0.620) (-0.372) (-0.351) (0.253)
R2 0.010 0.013 0.010 0.005 0.007 0.006

5.3 Stock Market Factors

We next investigate whether the nine successful long-short strategies can be explained
by the stock market risk factors. Previous research (e.g., Asness et al., 2013) finds that
value and momentum strategies comove strongly across different asset classes. Hence, the
cryptocurrency strategies may comove with their corresponding counterparts in the equity
market as well. In this section, we present results controlling for the Fama-French three-
factor model. In the Appendix, we also present results based on the Carhart four-factor and
the Fama-French five-factor models. The results are qualitatively similar using any of the
stock market factor models.
Table 14 presents the results based on the Fama-French three-factor model. Overall,
the Fama-French three-factor model adjusted alphas of the strategies are quantitatively
similar to the unadjusted excess returns. For example, the adjusted alpha for the market
capitalization long-short strategy is -3.2 percent per week with a t-statistic of -2.372. The
unadjusted average excess returns of the long-short strategy is -3.4 percent per week with a
t-statistic of -2.557. The adjusted alpha for the three-week momentum long-short strategy

33
is 4.2 percent per week with a t-statistic of 2.799. The unadjusted average excess returns of
the long-short strategy is 4.1 percent per week with a t-statistic of 2.742.

5.4 Hedged Strategies

Recent empirical asset pricing literature has found that the common practice to create
factor-portfolios by sorting on characteristics associated with average returns captures both
priced and unpriced risks. Daniel et al. (2018) develop a method to hedge the unpriced
risks in the stock market using covariance information estimated from past returns. In this
section, we apply their method to our factors and evaluate whether we can further strengthen
the performance of our cryptocurrency factors.
We follow the procedure in Daniel et al. (2018) and provide an example based on the
cryptocurrency size factor. Detailed descriptions of the theoretical motivation and empirical
account can be found in Daniel et al. (2018). We first rank all cryptocurrencies by their
previous week market capitalization. Break-points are selected at the 30 percent and 70
percent marks. Then, all cryptocurrencies are assigned to one of the three bins. Next,
each of the three bins is further sorted into three equal bins based on the coins’ expected
covariances with the cryptocurrency size factor. Therefore, cryptocurrencies with similar
size but different loadings on size factor are assigned into different bins. We estimate the
expected covariance between coin returns and the size factor using the rolling past 365 days
of data. Finally, the hedge-portfolio for the cryptocurrency size factor is constructed as going
long on an equal-weighted portfolio of the low size-factor-loading portfolios and short on an
equal-weighted portfolio of the high size-factor-loading portfolios. We find that the hedge-
portfolio does not carry statistically significant return spreads for either the cryptocurrency
size strategy or momentum strategy, similar to findings in the stock market (See Daniel
and Titman, 1997). Therefore, the construction ensures that the hedge-portfolio captures
unpriced risks. We construct the cryptocurrency momentum hedge-portfolio in the same
way.
We use the squared Sharpe-ratio to evaluate the performance of the strategies. For the
cryptocurrency size factor, we find considerable gains from hedging the unpriced risks. The
squared weekly Sharpe-ratio goes from 0.057 for the unhedged strategy to 0.076 for the
hedged strategy when the hedge-portfolio is available. The gains are economically large.
However, for the cryptocurrency momentum factor, the adjustment does not increase the
squared Sharpe-ratio of the momentum strategy. One possibility for the lack of improvement

34
of the cryptocurrency momentum strategy is that the expected loadings on the momentum
factors change faster and are more transient than those on the size factors.

6 Conclusion
The results of this paper show that the cross-section of cryptocurrencies can be meaning-
fully analyzed using standard asset pricing tools. We document that, similar to other asset
classes (see, e.g., Asness et al., 2013), size and momentum factors are important in cap-
turing the cross-section of cryptocurrency returns. Moreover, a parsimonious three-factor
model that can be constructed using the market information is successful in pricing the
strategies in the cryptocurrency market. The paper thus establishes a set of stylized facts
on the cross-section of cryptocurrencies that can be used to assess and develop theoretical
models.

35
References
Abadi, J. and M. Brunnermeier (2018): “Blockchain economics,” Working Paper.

Amaya, D., P. Christoffersen, K. Jacobs, and A. Vasquez (2015): “Does realized

skewness predict the cross-section of equity returns?” Journal of Financial Economics,
118, 135–167.

Amihud, Y. (2002): “Illiquidity and stock returns: Cross-section and time-series effects,”
Journal of Financial Markets, 5, 31–56.

Ang, A., R. J. Hodrick, Y. Xing, and X. Zhang (2006): “The cross-section of volatility
and expected returns,” The Journal of Finance, 61, 259–299.

Asness, C., A. Frazzini, R. Israel, T. J. Moskowitz, and L. H. Pedersen (2018):

“Size matters, if you control your junk,” Journal of Financial Economics.

Asness, C. S., T. J. Moskowitz, and L. H. Pedersen (2013): “Value and momentum

everywhere,” The Journal of Finance, 68, 929–985.

Bai, J., T. G. Bali, and Q. Wen (2018): “Common risk factors in the cross-section of
corporate bond returns,” Journal of Financial Economics.

Bali, T. G., N. Cakici, and R. F. Whitelaw (2011): “Maxing out: Stocks as lotteries
and the cross-section of expected returns,” Journal of Financial Economics, 99, 427–446.

Banz, R. W. (1981): “The relationship between return and market value of common
stocks,” Journal of Financial Economics, 9, 3–18.

Barberis, N., A. Shleifer, and R. Vishny (1998): “A model of investor sentiment,”

Journal of Financial Economics, 49, 307–343.

Barberis, N. and R. Thaler (2003): “A survey of behavioral finance,” Handbook of the

Economics of Finance, 1, 1053–1128.

Barry, C. B. and S. J. Brown (1984): “Differential information and the small firm
effect,” Journal of Financial Economics, 13, 283–294.

Biais, B., C. Bisiere, M. Bouvard, and C. Casamatta (2018a): “The blockchain folk
theorem,” Working Paper.

Biais, B., C. Bisiere, M. Bouvard, C. Casamatta, and A. J. Menkveld (2018b):

“Equilibrium bitcoin pricing,” Working Paper.

Bianchi, D. (2017): “Cryptocurrencies as an asset class: An empirical assessment,” Work-

ing Paper.

36
Bianchi, D. and A. Dickerson (2018): “Trading volume in cryptocurrency markets,”
Working Paper.

Borri, N. (2018): “Conditional tail-risk in cryptocurrency markets,” Working Paper.

Borri, N. and K. Shakhnov (2018a): “The cross-section of cryptocurrency returns,”

Working Paper.

——— (2018b): “Cryptomarket discounts,” Working Paper.

Brennan, M. J., T. Chordia, and A. Subrahmanyam (1998): “Alternative factor

specifications, security characteristics, and the cross-section of expected stock returns,”
Journal of Financial Economics, 49, 345–373.

Carhart, M. M. (1997): “On persistence in mutual fund performance,” The Journal of

Finance, 52, 57–82.

Chen, A. and M. Velikov (2017): “Accounting for the anomaly zoo: A trading cost
perspective,” Working Paper.

Chen, A. Y. and T. Zimmermann (2018): “Publication bias and the cross-section of

stock returns,” Working Paper.

Chiu, J. and T. V. Koeppl (2017): “The economics of cryptocurrencies–bitcoin and

beyond,” Working Paper.

Chordia, T., A. Subrahmanyam, and V. R. Anshuman (2001): “Trading activity and

expected stock returns,” Journal of Financial Economics, 59, 3–32.

Cong, L. W. and Z. He (2018): “Blockchain disruption and smart contracts,” Working

Paper.

Cong, L. W., Z. He, and J. Li (2018a): “Decentralized mining in centralized pools,”

Working Paper.

Cong, L. W., Y. Li, and N. Wang (2018b): “Tokenomics: Dynamic adoption and
valuation,” Working Paper.

Corbet, S., B. Lucey, and L. Yarovaya (2018a): “Datestamping the Bitcoin and
Ethereum bubbles,” Finance Research Letters, 26, 81–88.

Corbet, S., B. M. Lucey, A. Urquhart, and L. Yarovaya (2018b): “Cryptocurren-

cies as a financial asset: A systematic analysis,” Working Paper.

Corbet, S., A. Meegan, C. Larkin, B. Lucey, and L. Yarovaya (2018c): “Exploring

the dynamic relationships between cryptocurrencies and other financial assets,” Economics
Letters, 165, 28–34.

37
Daniel, K., D. Hirshleifer, and A. Subrahmanyam (1998): “Investor psychology and
security market under- and overreactions,” The Journal of Finance, 53, 1839–1885.

Daniel, K. and T. J. Moskowitz (2016): “Momentum crashes,” Journal of Financial

Economics, 122, 221–247.

Daniel, K., L. Mota, S. Rottke, and T. Santos (2018): “The cross-section of risk
and return,” Working Paper.

Daniel, K. and S. Titman (1997): “Evidence on the characteristics of cross sectional

variation in stock returns,” The Journal of Finance, 52, 1–33.

——— (2006): “Market reactions to tangible and intangible information,” The Journal of
Finance, 61, 1605–1643.

De Bondt, W. F. and R. Thaler (1985): “Does the stock market overreact?” The
Journal of Finance, 40, 793–805.

Fama, E. F. (1965): “The behavior of stock-market prices,” The Journal of Business, 38,
34–105.

——— (1998): “Market efficiency, long-term returns, and behavioral finance,” Journal of
Financial Economics, 49, 283–306.

Fama, E. F. and K. R. French (1992): “The cross-section of expected stock returns,”

The Journal of Finance, 47, 427–465.

——— (1993): “Common risk factors in the returns on stocks and bonds,” Journal of Fi-
nancial Economics, 33, 3–56.

——— (1995): “Size and book-to-market factors in earnings and returns,” The Journal of
Finance, 50, 131–155.

——— (1996): “Multifactor explanations of asset pricing anomalies,” The Journal of Fi-
nance, 51, 55–84.

——— (2016a): “Choosing factors,” Working Paper.

——— (2016b): “Dissecting anomalies with a five-factor model,” Review of Financial Stud-
ies, 29, 69–103.

Fama, E. F. and J. D. MacBeth (1973): “Risk, return, and equilibrium: Empirical

tests,” Journal of Political Economy, 81, 607–636.

Feng, G., S. Giglio, and D. Xiu (2017): “Taming the factor zoo,” Working Paper.

George, T. J. and C.-Y. Hwang (2004): “The 52-week high and momentum investing,”
The Journal of Finance, 59, 2145–2176.

38
Gervais, S., R. Kaniel, and D. H. Mingelgrin (2001): “The high-volume return
premium,” The Journal of Finance, 56, 877–919.

Gilbert, S. and H. Loi (2018): “Digital currency risk,” International Journal of Eco-
nomics and Finance, 10, 108.

Herskovic, B., A. Moreira, and T. Muir (2018): “Hedging risk factors,” Working
Paper.

Hong, H., T. Lim, and J. C. Stein (2000): “Bad news travels slowly: Size, analyst
coverage, and the profitability of momentum strategies,” The Journal of Finance, 55,
265–295.

Hong, H. and J. C. Stein (1999): “A unified theory of underreaction, momentum trading,

and overreaction in asset markets,” The Journal of Finance, 54, 2143–2184.

Hou, K., C. Xue, and L. Zhang (2015): “Digesting anomalies: An investment approach,”
Review of Financial Studies, 28, 650–705.

Hu, A., C. A. Parlour, and U. Rajan (2018): “Cryptocurrencies: Stylized facts on a

new investible instrument,” Working Paper.

Hubrich, S. (2017): “’Know when to hodl ’em, know when to fodl ’em’: An investigation
of factor based investing in the cryptocurrency space,” Working Paper.

Jegadeesh, N. (1990): “Evidence of predictable behavior of security returns,” The Journal

of Finance, 45, 881–898.

Jegadeesh, N. and S. Titman (1993): “Returns to buying winners and selling losers:
Implications for stock market efficiency,” The Journal of Finance, 48, 65–91.

——— (1995): “Overreaction, delayed reaction, and contrarian profits,” Review of Financial
Studies, 8, 973–993.

——— (2001): “Profitability of momentum strategies: An evaluation of alternative expla-

nations,” The Journal of Finance, 56, 699–720.

Li, J. and W. Mann (2018): “Initial coin offering and platform building,” Working Paper.

Li, J. and G. Yi (2018): “Toward a factor structure in crypto asset returns,” Working
Paper.

Liu, Y. and A. Tsyvinski (2018): “Risks and returns of cryptocurrency,” Working Paper.

Lustig, H., N. Roussanov, and A. Verdelhan (2011): “Common risk factors in cur-
rency markets,” Review of Financial Studies, 24, 3731–3777.

39
Makarov, I. and A. Schoar (2018): “Trading and arbitrage in cryptocurrency markets,”
Working Paper.

Moskowitz, T. J. and M. Grinblatt (1999): “Do industries explain momentum?” The

Journal of Finance, 54, 1249–1290.

Moskowitz, T. J., Y. H. Ooi, and L. H. Pedersen (2012): “Time series momentum,”

Journal of Financial Economics, 104, 228–250.

Novy-Marx, R. (2012): “Is momentum really momentum?” Journal of Financial Eco-

nomics, 103, 429–453.

Pagnotta, E. (2018): “Bitcoin as decentralized money: Prices, mining rewards, and net-
work security,” Mining Rewards, and Network Security.

Pagnotta, E. and A. Buraschi (2018): “An equilibrium valuation of bitcoin and decen-
tralized network assets,” Working Paper.

Routledge, B. and A. Zetlin-Jones (2018): “Currency stability using blockchain tech-

nology,” Working Paper.

Schilling, L. and H. Uhlig (2018): “Some simple bitcoin economics,” Working Paper.

Sockin, M. and W. Xiong (2018): “A model of cryptocurrencies,” Working Paper.

Stoffels, J. (2017): “Asset pricing of cryptocurrencies and momentum based patterns,”

Working Paper.

Szymanowska, M., F. De Roon, T. Nijman, and R. Van Den Goorbergh (2014):

“An anatomy of commodity futures risk premia,” The Journal of Finance, 69, 453–482.

Weber, W. E. (2016): “A Bitcoin standard: Lessons from the gold standard,” Working
Paper.

Yang, H. (2018): “Behavioral anomalies in cryptocurrency markets,” Working Paper.

Yermack, D. (2015): “Is Bitcoin a real currency? An economic appraisal,” in Handbook of

Digital Currency, Elsevier, 31–43.

40
Online Appendix

Table A.1: Summary Statistics of Factors

Panel A reports mean, median, standard deviation, skewness, and kurtosis of the cryptocurrency market
excess returns, the cryptocurrency size factor returns, and the cryptocurrency momentum factor returns.
Panel B reports the correlation matrix of the cryptocurrency market excess returns, the cryptocurrency size
factor returns, the cryptocurrency momentum factor returns, the Bitcoin returns, the Ethereum returns, and
the Ripple returns.

Panel A
Mean Median SD Skewness Kurtosis
Market Excess Return 0.013 0.005 0.117 0.292 4.574
Size Return 0.020 -0.005 0.143 3.067 19.616
Mom Return 0.038 0.034 0.184 0.791 7.868

Panel B
CMKT CSIZE CMOM Bitcoin Ethereum Ripple
Market Excess Return 1.000
Size Return 0.187 1.000
Mom Return -0.001 -0.033 1.000
Bitcoin Return 0.927 0.055 -0.030 1.000
Ethereum Return 0.531 0.146 -0.019 0.367 1.000
Ripple Return 0.549 0.268 0.022 0.376 0.289 1.000

Table A.2: Double Sort on Size and Momentum

This table shows results based on double sorting on both the size and momentum factors. Each coin is first
sorted into one of two size portfolios. Within each size portfolio, the coins are further sorted into one of five
momentum portfolios. *, **, *** denote significance levels at the 10%, 5%, and 1%.

Momentum
1 2 3 4 5 5-1
Low High
Low 0.032** 0.015 0.020* 0.015 0.039** 0.006
(2.05) (1.29) (1.72) (1.26) (2.05) (0.30)
Size
High -0.002 -0.00 0.018 0.019** 0.040** 0.042***
(-0.15) (-0.01) (1.60) (1.98) (2.48) (2.57)

41
 Table A.3: Carhart Four-Factor

Ri − R f = αi + βiM KT MKT + βSi M B SM B + βiH M L H M L + βiMOM MOM + i (6)

where MKT is the excess stock market returns, SM B is the Fama-French size factor, H M L is the Fama-
French value factor, and MOM is the momentum factor. The formation of the quintile portfolios for the
nine significant strategies are discussed in Section 3. *, **, *** denote significance levels at the 10%, 5%,
and 1%. m.a.e and R¯2 are the mean of the absolute pricing errors and the average R2 of the five portfolios.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.043*** 0.017 0.012 0.012 0.012 -0.031**
βM K T 0.937 0.998 0.333 0.672 0.400 -0.536
βS M B -0.177 -0.941 0.159 -0.322 -0.689 -0.512
MCAP 0.019 0.008
βH M L -1.023 0.667 -0.326 -0.365 -0.244 0.779
βM O M 0.607 0.126 0.175 0.449 -0.036 -0.643
R2 0.009 0.013 0.003 0.010 0.007 0.009
α 0.049** 0.027** 0.000 0.017 0.011 -0.038**
βM K T 1.210 0.522 0.637 1.112 0.410 -0.800
βS M B -1.364 -1.483 -0.383 -0.857 -0.571 0.792
PRC 0.021 0.008
βH M L -1.392 0.485 -0.335 1.008 -0.067 1.325
βM O M -0.351 0.144 0.096 -0.104 0.103 0.454
R2 0.009 0.008 0.005 0.014 0.006 0.006
α 0.050*** 0.024* 0.000 0.020 0.011 -0.039**
βM K T 1.157 0.515 0.631 1.077 0.409 -0.748
βS M B -1.601 -1.169 -0.346 -0.690 -0.571 1.030
MAXDPRC 0.021 0.008
βH M L -1.319 0.373 -0.148 0.913 -0.064 1.255
βM O M -0.183 0.128 0.176 -0.276 0.107 0.290
R2 0.009 0.005 0.005 0.013 0.006 0.006
α -0.009 -0.003 0.008 0.042** 0.021 0.029**
βM K T 0.927 0.467 0.408 1.604 -0.337 -1.265*
βS M B 0.298 -1.348 -0.711 1.178 -1.866 -2.165*
r 1,0 0.017 0.017
βH M L -0.613 -0.406 -0.319 2.816 -0.376 0.237
βM O M 1.265* -0.041 0.320 0.032 0.343 -0.922
R2 0.028 0.011 0.007 0.026 0.014 0.031
α -0.005 0.005 0.009 0.017* 0.029** 0.034**
βM K T 0.979 0.329 -0.064 0.270 -0.121 -1.100
βS M B 0.053 0.239 -1.153 -1.163 -1.826 -1.879
r 2,0 0.013 0.013
βH M L -0.186 -0.337 -0.854 0.823 -0.807 -0.620
βM O M 0.967 -0.345 -0.347 0.658 0.539 -0.428
R2 0.018 0.004 0.012 0.014 0.017 0.022

42
Table A.3 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α 0.001 0.001 0.014 0.018* 0.042*** 0.041***
βM K T 0.494 0.082 0.596 0.952* -0.636 -1.130
βS M B 1.309 -0.391 -0.514 -0.798 -1.610 -2.920**
r 3,0 0.015 0.012
βH M L -0.193 -0.578 -0.230 0.537 0.611 0.804
βM O M 0.455 -0.095 -0.152 0.468 1.277 0.823
R2 0.013 0.002 0.005 0.016 0.022 0.041
α 0.002 0.002 0.007 0.017* 0.026* 0.024*
βM K T 0.264 0.738 0.304 0.718 -0.426 -0.690
βS M B 0.713 -0.028 -0.895 -0.976 -1.556 -2.270**
r 4,0 0.011 0.011
βH M L 0.253 -0.148 0.264 -0.335 -1.054 -1.307
βM O M 0.453 0.212 -0.230 0.005 0.981 0.529
R2 0.005 0.007 0.008 0.010 0.025 0.038
α 0.049** 0.027** 0.000 0.017 0.011 -0.038**
βM K T 1.210 0.522 0.637 1.112 0.410 -0.800
βS M B -1.364 -1.483 -0.383 -0.857 -0.571 0.792
PRCVOL 0.021 0.008
βH M L -1.392 0.485 -0.335 1.008 -0.067 1.325
βM O M -0.351 0.144 0.096 -0.104 0.103 0.454
R2 0.009 0.008 0.005 0.014 0.006 0.006
α 0.040** 0.029* 0.019 0.019 0.012 -0.028**
βM K T 1.380 1.305 0.873 0.796 0.389 -0.992
βS M B -1.129 -1.608 -1.170 -0.054 -0.686 0.443
STDPRCVOL 0.024 0.009
βH M L -0.139 -0.818 -0.772 -0.053 -0.246 -0.107
βM O M 0.612 -0.137 -0.170 0.590 -0.042 -0.654
R2 0.011 0.013 0.010 0.006 0.007 0.008

43
 Table A.4: Fame-French Five-Factor

Ri − R f = αi + βiM KT MKT + βSi M B SM B + βiH M L H M L + βRMW

i
RMW + βC
i
M ACM A + i (7)

where MKT is the excess stock market returns, SM B is the Fama-French size factor, H M L is the Fama-French
value factor, and RMW is the Fama French profitability factor, and CM A is the Fama French investment
factor. The formation of the quintile portfolios for the nine significant strategies are discussed in Section 3.
*, **, *** denote significance levels at the 10%, 5%, and 1%. m.a.e and R¯2 are the mean of the absolute
pricing errors and the average R2 of the five portfolios.

1 2 3 4 5 5-1 m.a.e R¯2

α 0.045*** 0.017 0.013 0.013 0.012 -0.033**
βM K T 0.903 0.762 0.164 0.555 0.357 -0.546
βS M B -0.554 -1.326 0.001 -0.747 -0.743 -0.188
MCAP βH M L -1.992 0.716 -0.143 -1.020 -0.163 1.829 0.020 0.011
βR M W -1.147 -1.932 -0.744 -1.636 -0.341 0.806
βC M A 1.789 -0.129 -0.723 1.243 -0.123 -1.912
R2 0.010 0.018 0.004 0.015 0.007 0.009
α 0.049** 0.028** 0.001 0.017 0.011 -0.038**
βM K T 1.395 0.368 0.758 1.240 0.305 -1.090
βS M B -1.593 -1.813 -0.619 -0.766 -0.614 0.980
PRC βH M L -2.178 0.335 -1.227 0.800 0.145 2.322 0.021 0.011
βR M W -1.231 -1.561 -0.853 0.461 -0.206 1.025
βC M A 2.935 0.352 2.416 0.705 -0.742 -3.676
R2 0.012 0.010 0.011 0.015 0.007 0.012
α 0.051*** 0.025* 0.001 0.020 0.011 -0.040**
βM K T 1.386 0.328 0.670 1.240 0.304 -1.082
βS M B -1.856 -1.503 -0.669 -0.548 -0.614 1.242
MAXDPRC βH M L -2.419 0.343 -0.964 0.793 0.146 2.565 0.022 0.010
βR M W -1.119 -1.635 -1.271 0.553 -0.209 0.910
βC M A 3.506 0.050 2.119 0.744 -0.745 -4.251
R2 0.013 0.008 0.011 0.013 0.007 0.013
α -0.008 -0.003 0.008 0.041** 0.022 0.029**
βM K T 0.701 0.530 0.344 2.065* -0.548 -1.249
βS M B -0.135 -1.344 -0.790 1.675 -2.363* -2.228*
r 1,0 βH M L -1.484 -0.587 -0.466 2.075 -0.759 0.724 0.016 0.017
βR M W -0.924 0.048 -0.098 2.879 -2.189 -1.265
βC M A 0.359 0.570 -0.130 1.676 0.732 0.373
R2 0.017 0.011 0.006 0.031 0.018 0.029

44
Table A.4 Continued 1 2 3 4 5 5-1 m.a.e R¯2
α -0.004 0.005 0.008 0.017* 0.030** 0.034**
βM K T 0.632 0.282 0.057 0.341 -0.336 -0.968
βS M B -0.324 0.012 -0.991 -0.926 -2.365* -2.041
r 2,0 βH M L -0.334 -0.327 -0.741 0.525 -1.386 -1.053 0.013 0.014
βR M W -1.122 -1.482 0.525 1.889 -2.180 -1.058
βC M A -1.120 0.736 0.221 -0.520 0.940 2.061
R2 0.014 0.007 0.011 0.017 0.020 0.025
α 0.002 0.001 0.013 0.018* 0.044*** 0.042***
βM K T 0.304 0.106 0.563 0.916 -0.695 -0.999
βS M B 0.901 -0.478 -0.378 -0.723 -2.170 -3.071**
r 3,0 βH M L -0.572 -0.746 0.229 0.459 -1.060 -0.488 0.016 0.012
βR M W -1.632 -0.479 0.434 0.810 -1.306 0.326
βC M A 0.464 0.685 -1.066 -0.682 2.614 2.149
R2 0.015 0.003 0.007 0.016 0.020 0.040
α 0.002 0.003 0.007 0.017* 0.027** 0.025**
βM K T 0.181 0.671 0.317 0.720 -0.259 -0.439
βS M B 0.602 -0.228 -0.801 -0.818 -1.813 -2.415**
r 4,0 βH M L 0.024 -0.414 0.541 -0.070 -2.770* -2.794** 0.011 0.011
βR M W -0.130 -0.780 0.206 0.743 0.049 0.179
βC M A -0.125 0.467 -0.402 -0.832 3.088 3.213
R2 0.003 0.008 0.008 0.012 0.026 0.043
α 0.043** 0.028** 0.014 0.017 0.012 -0.031*
βM K T 1.450 0.719 1.156 0.886 0.327 -1.123
βS M B -2.261 -1.451 -0.652 -0.139 -0.749 1.512
PRCVOL βH M L -0.891 -1.677 -0.344 -1.923 -0.130 0.761 0.023 0.014
βR M W -3.382 -2.623 0.874 0.263 -0.411 2.971
βC M A 1.306 2.596 0.565 3.908* -0.235 -1.541
R2 0.018 0.018 0.011 0.014 0.007 0.015
α 0.041** 0.031** 0.018 0.021 0.012 -0.030**
βM K T 1.176 1.124 1.160 0.873 0.333 -0.843
βS M B -1.662 -2.271 -1.038 -0.484 -0.744 0.918
STDPRCVOL βH M L -0.798 -1.250 -1.406 -1.477 -0.126 0.672 0.025 0.014
βR M W -2.065 -3.442 0.767 -1.279 -0.379 1.687
βC M A 1.021 1.836 1.972 3.101 -0.221 -1.242
R2 0.013 0.023 0.013 0.012 0.007 0.010

45
 Table A.5: Crypto Factor – Tercile
CMKT is the cryptocurrency excess market returns, CSM B is the cryptocurrency size factor, and CMOM
is the cryptocurrency momentum factor. The formation of the quintile portfolios for the nine significant
strategies are discussed in Section 3. The t-statistics are reported in the parentheses. *, **, *** denote
significance level at the 10%, 5%, and 1%.
Cons t CMKT t CSMB t CMOM t R2
Mean -0.015* (-1.810)
MCAP C-1 -0.013 (-1.633) -0.121* (-1.741) 0.012
C-3 0.004** (1.970) -0.006 (-0.369) -0.893*** (-72.831) -0.004 (-0.429) 0.994
Mean -0.027** (-2.080)
PRC C-1 -0.022* (-1.721) -0.441*** (-4.094) 0.062
C-3 -0.010 (-0.815) -0.365*** (-3.621) -0.544*** (-6.630) -0.047 (-0.738) 0.201
Mean -0.027** (-2.099)
MAXDPRC C-1 -0.022* (-1.741) -0.436*** (-4.070) 0.061
C-3 -0.010 (-0.819) -0.359*** (-3.589) -0.545*** (-6.685) -0.051 (-0.807) 0.203
Mean 0.041*** (3.719)
r 1,0 C-1 0.041*** (3.633) 0.060 (0.628) 0.002
C-3 0.021** (2.072) -0.019 (-0.219) 0.127* (1.852) 0.479*** (8.947) 0.246
Mean 0.029*** (2.698)
r 2,0 C-1 0.030*** (2.997) 0.168** (1.978) 0.015
C-3 0.007 (1.031) 0.089 (1.484) -0.014 (-0.286) 0.618*** (16.236) 0.519
Mean 0.031*** (2.836)
r 3,0 C-1 0.030*** (2.696) 0.120 (1.287) 0.006
C-3 -0.003 (-0.828) 0.005 (0.147) -0.009 (-0.346) 0.897*** (44.227) 0.886
Mean 0.027** (2.518)
r 4,0 C-1 0.025** (2.340) 0.157* (1.742) 0.012
C-3 0.000 (0.029) 0.073 (1.218) -0.046 (-0.943) 0.692*** (18.278) 0.576
Mean -0.020* (-1.686)
PRCVOL C-1 -0.017 (-1.408) -0.300*** (-2.920) 0.032
C-3 0.002 (0.231) -0.181** (-2.184) -0.805*** (-11.921) -0.113** (-2.153) 0.385
Mean -0.021* (-1.764)
STDPRCVOL C-1 -0.018 (-1.494) -0.284*** (-2.845) 0.031
C-3 0.001 (0.820) 0.998*** (171.436) -0.006 (-1.170) -0.002 (-0.588) 0.992

46