Forecasting Cryptocurrencies Log-Returns a LASSO-VAR and Sentiment Approach Milos CiganovicFederico DAmario

2025-05-01 0 0 776.69KB 26 页 10玖币

侵权投诉

Forecasting Cryptocurrencies Log-Returns: a

LASSO-VAR and Sentiment Approach

Milos Ciganovic∗Federico D’Amario†

October 4, 2022

Abstract

Cryptocurrencies have become a trendy topic recently, primarily due to their dis-

ruptive potential and reports of unprecedented returns. In addition, academics in-

creasingly acknowledge the predictive power of Social Media in many ﬁelds and, more

speciﬁcally, for ﬁnancial markets and economics. In this paper, we leverage the pre-

dictive power of Twitter and Reddit sentiment together with Google Trends indexes

and volume to forecast the log returns of ten cryptocurrencies. Speciﬁcally, we con-

sider Bitcoin,Ethereum,T ether,BinanceCoin,Litecoin,EnjinCoin,Horizen,

Namecoin,P eercoin, and F eathercoin. We evaluate the performance of LASSO-VAR

using daily data from January 2018 to January 2022. In a 30 days recursive forecast,

we can retrieve the correct direction of the actual series more than 50% of the time.

We compare this result with the main benchmarks, and we see a 10% improvement

in Mean Directional Accuracy (MDA). The use of sentiment and attention variables

as predictors increase signiﬁcantly the forecast accuracy in terms of MDA but not in

terms of Root Mean Squared Errors. We perform a Granger causality test using a

post-double LASSO selection for high-dimensional VARs. Results show no “causality”

from Social Media sentiment to cryptocurrencies returns

Keywords: Cryptocurrencies, Time series analysis, Sentiment analysis,

Natural Language Processing

JEL Codes: C32, C53, C55, G17

∗Department of Economics and Law - Sapienza University of Rome. milos.ciganovic@uniroma1.it

†Department of Economics and Law - Sapienza University of Rome. federico.damario@uniroma1.it

arXiv:2210.00883v1 [q-fin.ST] 22 Sep 2022

1 Introduction

Since the introduction of Bitcoin by Nakamoto (2008), cryptocurrencies have gradually be-

come very popular among investors. In the last decade, the world witnessed an unforeseen

growth of cryptocurrencies, both in terms of their market capitalization and the number of

kinds of coins. Many reasons may justify this boom: ﬁrst of all, social media and journals re-

ported unprecedented returns of cryptocurrencies which led many, professional investors and

not, to enter this market. This mechanism has been naturally motivated by minimal global

regulation, which brought people to primarily see cryptocurrencies as means of payment for

illegal trades. Eventually, their gigantic returns stimulated an enthusiasm reminiscent of the

Gold Rush in the western U.S. Investing in cryptocurrencies can be done quickly by every-

one downloading an app on their smartphone. This led youngsters to represent, relatively,

the main investors in digital currencies1. Moreover, due to the relatively young age of the

cryptocurrency market, traditional news outlets cannot follow events timely.

Because of the reasons mentioned above, social media can be deﬁned as the primary source

of information for cryptocurrency investors. Speciﬁcally, micro-blogging websites such as

Twitter 2and Reddit 3are widely used sources for cryptocurrency information. Signiﬁcant

ﬂuctuations in cryptocurrencies’ prices and their high volatility resulted in signiﬁcant risks

associated with investment in crypto assets. This has led to heated discussions about their

place and role in the modern economy (see, for example Corbet et al. (2019), Catalini and

Gans (2020), Halaburda et al. (2020) and Auer et al. (2021)). Therefore, the issue of devel-

oping appropriate methods and models for predicting prices for digital currencies is relevant

both for the scientiﬁc community and ﬁnancial analysts, investors, and traders. A crucial

contribution in terms of modelling and forecasting cryptocurrencies’ ﬁnancial time series has

been given by Catania et al. (2019) and Catania and Grassi (2021) who develop a dynamic

model suitable for the complex dynamics of these series as well as compare several alter-

native of univariate and multivariate models for point and density forecasts. Furthermore,

many studies (see Hitam and Ismail (2018), Sun et al. (2020), Miller and Kim (2021)) show

how Machine Learning algorithms are extremely convenient in terms of computational time

and accuracy when forecasting cryptocurrencies’ time series. Lastly, increasing literature

highlights the importance of speciﬁc factors that shape cryptocurrencies’ demand helping in

1See https://www.investopedia.com/younger-generations-bullish-on-cryptocurrencies-5223563.

2See https://twitter.com/.

3See https://www.reddit.com/.

forecasting their prices, returns, and volatility. According to the eﬃcient market hypothesis

Fama (1970), market prices reﬂect all available information, thus the prediction of stock

returns should not be possible. On the other hand, considerable empirical evidences (see,

e.g. Daniel et al. (2002) for a comprehensive review) show that investors’ psychology drives

the stock market. This led many researchers to adopt sentiment indexes to improve forecast

accuracy. Glenski et al. (2019) exploit the predictive power of social signals from multiple

platforms (GitHub and Reddit) to forecast prices for three cryptocurrencies. They show

that social signals reduce error when forecasting daily coin prices and that the language

used in comments within the oﬃcial communities on Reddit are the best predictors overall.

Kraaijeveld and De Smedt (2020) show that Twitter sentiment has predictive power for the

returns of several cryptocurrencies. Aslanidis et al. (2022) and Nasir et al. (2019) highlight

the link of Google Trends with cryptocurrencies regarding their returns and volatility.

In this study, we analyze the impact of sentiment variables on cryptocurrency returns by

using a novel dataset that combines a number of social media, search engine data, and vol-

ume. We apply a state-of-the-art sentiment classiﬁcation technique to investigate whether

sentiment measures contain predictive power for returns. To the best of our knowledge,

similar sets of predictors have not been employed jointly previously. We account for the high

dimensionality of the predictor variables by using a regularization technique known as the

LASSO. This allows us to investigate (i) whether the variables constructed from our novel

dataset can help to improve log-return forecasts using a VAR approach compared to the

benchmark models; (ii) which data source and which type of sentiment or attention mea-

sure is most relevant in terms of Granger-causality in High-Dimensional VARs. Our results

show that, on average, LASSO-VAR performs better in terms of Mean Directional Accuracy

(MDA) than benchmark models. Moreover, the use of sentiment and attention variables as

predictors increase signiﬁcantly the forecast accuracy. We do not ﬁnd Granger causality from

sentiment indexes to cryptocurrencies returns. We ﬁnd, instead, Granger causality between

all cryptocurrencies except Bitcoin, Tether, and Feathercoin and from returns to the bitcoin

sentiment extracted from Twitter.

The remainder of the paper is organized as follows. Section two 2 deals with data collection,

describing the data set, its sources, and strategies to construct sentiment indexes. Section

three 3 describes the modelling strategy, the estimation, forecasting methods, and the met-

rics used for the comparative evaluation of the out-of-sample model predictions. Section four

4 summarizes some selected results. Section ﬁve 5 concludes.

2 Data Collection

This study relies on multiple data sources. First, we collect daily data of ten cryptocurrencies

from January 2018 to January 2022. We provide for the same period google trends and

sentiment indexes from Twitter and Reddit. We complete our dataset with volume for each

cryptocurrency considered.

2.1 Cryptocurrency Data

The cryptocurrencies used are reported in table 1 ranked in terms of Market Capitalization

(MC) as of January 2022. We get our data from ﬁnance.yahoo.com4.

Table 1: 10 Cryptocurrencies and their symbols, market capitalization (MCs) and rankings of MCs (as of

31 January 2022).

Cryptocurrency Symbol MC Rank by MC

Bitcoin BTC $702,864,225,136 1

Ethereum ETH $295,905,148,931 2

Tether USDT $78,188,468,450 3

Binance Coin BNB $63,930,448,963 4

Litecoin LTC $7,479,561,631 21

Enjin Coin ENJ $1,434,490,287 60

Horizen ZEN $505,212,977 120

Namecoin NMC $24,472,607 733

Peercoin PPC $13,809,105 837

Feathercoin FTC $1,927,251 1515

Many reasons brought us to choose this set of cryptocurrencies. First of all, cryptocurren-

cies emerge and disappear continually, while our selected ten currencies have been publicly-

traded consecutively. Moreover, all the currencies chosen have been created with a deﬁned

purpose representing innovative projects which brought development, progress, or value to

the blockchain technology that Bitcoin had implemented. On the other hand, our sample in-

cludes three tier currencies as in Gandal and Halaburda (2016). Bitcoin,Ethereum,T ether,

BinanceCoin, whose market capitalizations stay in the world’s top ﬁve, are “top-tier” cryp-

tocurrencies. Litecoin,EnjinCoin,Horizen, representing “middle cryptocurrencies” in

market capitalization. Namecoin,P eercoin, and F eathercoin are representative “minor

cryptocurrencies” according to market capitalization. We include Tether (USDT) in our

4See: https://it.ﬁnance.yahoo.com/criptocurrencies/.

sample for a speciﬁc reason. We know, indeed, that it is a blockchain-based cryptocurrency

whose tokens in circulation are backed by an equivalent amount of U.S. dollars, making it

a stablecoin with a price pegged to USD $1.00, which leads this currency to be very low

volatile. Tether was designed to build the necessary bridge between ﬁat currencies and cryp-

tocurrencies and oﬀer users stability, transparency, and minimal transaction charges. We

decided to include it as a counterfactual to understand whether sentiment indexes can help

the prediction of stablecoin currencies. We compute the log returns and include them in our

sample. Table 2 provides several summary statistics.

Table 2: Log-returns summary statistics for the ten cryptocurrencies during the period 1 January 2018 to

31 January 2022

BTC-USD ETH-USD USDT-USD BNB-USD LTC-USD ENJ-USD ZEN-USD NMC-USD PPC-USD FTC-USD

Mean 0.001 0.001 0 0.003 0 0.002 0 -0.001 -0.001 -0.003

Median 0.001 0.001 0 0.001 0 -0.001 -0.001 0.001 -0.001 -0.005

Min -0.465 -0.551 -0.053 -0.543 -0.449 -0.624 -0.546 -1.16 -0.665 -0.474

Max 0.172 0.231 0.053 0.529 0.291 0.768 0.38 0.75 0.567 0.409

Range 0.637 0.781 0.106 1.072 0.74 1.392 0.926 1.91 1.232 0.883

Skew -1.147 -1.099 0.3 0.3 -0.608 1.132 -0.233 -0.869 -0.174 -0.174

kurtosis 14.042 10.795 34.34 15.579 7.947 15.508 6.268 19.999 13.593 4.575

ADF -11.0753 -11.1057 -14.186 -11.1936 -11.3664 -11.201 -11.0829 -12.9631 -12.9892 -11.5868

Notes: all ADF statistics are stationary at 1% level

2.2 Google Trends Data

We collected forty three google trends searches (Table 7 in the Appendix reports the list of

all the Google trends collected).

Figure 1: Google trends words collected

Google Trends is a search trend feature that shows how frequently a given search term

is entered into Google’s search engine relative to the site’s total search volume over a given

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ForecastingCryptocurrenciesLog-Returns:aLASSO-VARandSentimentApproachMilosCiganovic*FedericoD'AmarioOctober4,2022AbstractCryptocurrencieshavebecomeatrendytopicrecently,primarilyduetotheirdis-ruptivepotentialandreportsofunprecedentedreturns.Inaddition,academicsin-creasinglyacknowledgethepredictivepo...

展开>> 收起<<

Forecasting Cryptocurrencies Log-Returns a LASSO-VAR and Sentiment Approach Milos CiganovicFederico DAmario.pdf

共26页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Forecasting Cryptocurrencies Log-Returns a LASSO-VAR and Sentiment Approach Milos CiganovicFederico DAmario

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: