PaperlayerView exact revision on arXiv

Source-linked AI summary

The ICO Phenomenon and Its Relationships with Ethereum Smart Contract Environment

Gianni Fenu, Lodovica Marchesi, Michele Marchesi, Roberto Tonelli

arXiv:1803.01394v1cs.CY

TL;DR

ICOs grew rapidly as a way to finance projects, and the paper examines which factors are associated with their success. It analyzes ICO, token-market, and Ethereum data, defining success using funds raised and token-price behavior. ICO-bench ratings, some countries of origin, and platform were associated with success in the reported analysis.

  • Problem

    ICO fundraising expanded rapidly, motivating analysis of ICO characteristics and factors associated with success.

    Source: §I · pp. 1–2

  • Method

    The authors analyze 1,388 ICOs, token financial data, and Ethereum token data, defining success from fundraising and token-price behavior.

    Source: §I · p. 1, B. Defining ICO success · p. 3

  • Results

    ICO-bench ratings, some countries of origin, and platform were associated with ICO success; more than 1,000 ICOs used Ethereum, with over 16 million transfers and about 5.5 million holders.

    Source: §I · p. 2

  • Takeaways & Limitations

    The analysis found that team size did not seem relevant to success, while ratings and Ethereum token management were associated with it.

    Source: §IV · pp. 6–7

Abstract

from arXiv · show

Initial Coin Offerings (ICO) are public offers of new cryptocurrencies in exchange of existing ones, aimed to finance projects in the blockchain development arena. In the last 8 months of 2017, the total amount gathered by ICOs exceeded 4 billion US$, and overcame the venture capital funnelled toward high tech initiatives in the same period. A high percentage of ICOS is managed through Smart Contracts running on Ethereum blockchain, and in particular to ERC-20 Token Standard Contract. In this work we examine 1388 ICOs, published on December 31, 2017 on icobench.com Web site, gathering information relevant to the assessment of their quality and software development management, including data on their development teams. We also study, at the same date, the financial data of 450 ICO tokens available on coinmarketcap.com Web site, among which 355 tokens are managed on Ethereum blochain. We define success criteria for the ICOs, based on the funds actually gathered, and on the behavior of the price of the related tokens, finding the factors that most likely influence the ICO success likeliness.

I. INTRODUCTION

The paper examines ICO characteristics and success factors using ICO, token-market, and Ethereum data. It reports that ICO-bench ratings, some countries of origin, and the platform are associated with success.

I. INTRODUCTION

  • In the last eight months of 2017, ICOs raised more than US$4 billion, exceeding venture capital for high-tech initiatives in the same period.ICOs offer new cryptocurrencies in exchange for existing ones to finance projects, mostly in blockchain development.Source: §I · p. 1
  • A high percentage of ICOs use Ethereum smart contracts, particularly the ERC-20 standard, which supports creating tokens and transferring them under ownership controls.The contract stores token-owner addresses and balances and permits transfers when the sender demonstrates control of the associated private key.Source: §I · p. 1
  • The authors note that scientific-journal research on ICOs was scarce and cite prior work on ICO features, risks, and success factors.They describe icobench.com as a popular ICO-listing site that evaluates listed offerings and provides an API for gathering information.Source: §I · p. 1
  • The study examines 1,388 ICOs from icobench.com and financial data for 450 tokens from coinmarketcap.com, including 355 Ethereum tokens.It gathers information on ICO quality, software development management, and development teams, and defines success using funds raised and token-price behavior.Source: §I · pp. 1–2
  • ICO-bench ratings, some countries of origin, and the platform were associated with ICO success in the multivariate analysis.The introduction also reports more than 1,000 ICOs on Ethereum, over 16 million token transfers, and about 5.5 million holders.Source: §I · p. 2

II. METHOD

The study's method comprises gathering ICO data, establishing which data to analyze and how to define ICO success, and analyzing the data for factors associated with success.

II. METHOD

  • The method has three steps: gather ICO data, determine the data and criteria for assessing success, and analyze factors associated with success.The following subsections describe these steps.Source: §II · p. 2

A. Retrieving ICO data from the Internet

The study combines ICO listings, token-market information, and Ethereum transaction and holder data. It links records across sources using token names and symbols, with a manual check of uncertain matches.

A. Retrieving ICO data from the Internet

  • The three data sources were ICO listings, token financial data from cryptocurrency exchanges, and token transaction and holder data from Ethereum via ethplorer.io.Source: A. Retrieving ICO data from the Internet · p. 2
  • ICO records were gathered from icobench.com's API and included ratings, dates, fundraising figures, platform, team, milestones, and category, with some fields missing.The database contained textual, categorical, and numeric variables.Source: A. Retrieving ICO data from the Internet · p. 2
  • For listed tokens, the study collected names, symbols, supply, capitalization, Ethereum addresses when available, and daily price, volume, and market-cap series.The price series came from coinmarketcap.com using the Beautiful Soup scraping library.Source: A. Retrieving ICO data from the Internet · p. 2
  • For tokens with an Ethereum address, ethplorer.io APIs supplied total token supply, transfer counts, and holder counts; individual transactions and holders were outside the study's scope.Source: A. Retrieving ICO data from the Internet · p. 2
  • Records from icobench.com and coinmarketcap.com were linked by token name and symbol, with a Levenshtein-distance match below 0.6 used when names differed but symbols matched.The authors manually checked those matches and other possible matches involving same-name or same-symbol records.Source: A. Retrieving ICO data from the Internet · p. 2

B. Defining ICO success

The study classifies eligible ICOs as successful or failed using fundraising thresholds and, when sufficient 2017 price data are available, a market-cap decline criterion. It excludes some offerings that are incomplete or have insufficient data.

B. Defining ICO success

  • The authors extend a soft-cap-based success definition because soft-cap data may be missing and some ICOs proceed despite not reaching the soft cap.They use a binary successful-or-failed variable and exclude offerings that lack data or remain in progress.Source: B. Defining ICO success · p. 3
  • ICOs raising less than US$80,000 were classified as failed, those raising more than US$200,000 as successful unless criterion 3 applied, and those between the thresholds as undecided and excluded.Source: B. Defining ICO success · p. 3
  • The analysis excluded ICOs ending in 2018, except those stopped after raising money in 2017, and those with no money raised and no end date.Source: B. Defining ICO success · p. 3
  • For tokens with price data covering all of December 2017, an ICO was classified as failed if market capitalization fell by more than 75% from the start of quotation.Market capitalization was computed as a moving average of 20 consecutive days to filter daily variations.Source: B. Defining ICO success · p. 3

C. Analysis of the factors influencing the success

The analysis tests selected ICO characteristics against binary success or failure using a multivariate Logit model. It evaluates each variable’s significance and its association with the success odds.

C. Analysis of the factors influencing the success

III. RESULTS AND DISCUSSION

The results section reports datasets gathered on December 31, 2017, including ICOs and token samples. It also includes a figure on ICO rating distributions.

III. RESULTS AND DISCUSSION

  • 1,387 ICOs listed on icobench.com were gathered on December 31, 2017.Source: §III · p. 4
  • 450 coinmarketcap.com tokens were gathered, including 355 Ethereum-managed tokens with data reported on ethplorer.io.Source: §III · p. 4
  • Fig. 3 presents the distribution of ICO ratings as a moving average of three rating scores.Source: Figure 3 · p. 4

A. Descriptive statistics of the ICO data

The descriptive analysis spans ICO origins, business categories, ratings, team sizes, and token platforms. Ethereum and ERC-20 feature prominently, and the studied ERC-20 tokens show highly uneven transfer and holder distributions.

A. Descriptive statistics of the ICO data

  • The USA and Russia Federation were the most active ICO-origin countries; 139 ICOs did not declare a country.Singapore, Switzerland, Estonia, and Slovenia were also described as relatively active despite their small size.Source: A. Descriptive statistics of the ICO… · p. 4
  • Most ICOs named platforms for decentralized business, while 233 were new kinds of cryptocurrencies and other categories spanned nearly all business sectors.The analysis reports only each ICO’s first listed category, assumed to be its most expressive business target.Source: A. Descriptive statistics of the ICO… · p. 4
  • Ratings climbed from about 1.2 to a peak of 3.9, then fell steeply, with very few ratings at or above 4.5.The reported distribution uses a centered moving average of three rating values.Source: A. Descriptive statistics of the ICO… · p. 4
  • The overall ICO-team size distribution peaked around 7–8 people, but included business staff and advisors, not just software developers.Where developers were part of a team, they typically accounted for 20%–50%; some teams had only business and marketing staff before success.Source: A. Descriptive statistics of the ICO… · p. 4, Figure 4 · p. 4
  • Ethereum was by far the most-used platform; 193 ICOs did not declare a platform, while Waves was second with 67.ERC-20 on Ethereum was reported in 787 cases, the standard was undeclared in 581, and ERC223 appeared in 12.Source: A. Descriptive statistics of the ICO… · pp. 4–5, Figure 5 · p. 4
  • 1,082 ICOs relied on Ethereum, whose public blockchain was described as performing quite well despite the load.Their overall value was estimated at 30–40 billion US$ at the time of evaluation.Source: A. Descriptive statistics of the ICO… · p. 5
  • Across 355 ERC-20 Ethereum tokens, transfer and holder counts had means well above medians, high standard deviations, and very large maxima.Their log-log CCDFs tended toward straight lines on the right, a characteristic described as typical of power-law distributions.Source: Figure 6 · p. 4, A. Descriptive statistics of the ICO… · p. 5, Figure 7 · p. 5

B. Multivariate Analysis of the factors influencing the success

In the multivariate analysis, icobench.com rating is the strongest success correlate; country, category, and platform also show associations, while team size does not.

B. Multivariate Analysis of the factors influencing the success

  • p-value 0.0001, coefficient 1.6991, and Z-Wald 3.84 make icobench.com rating the strongest success correlate; each unit raises the success odds about fivefold.The rating is on a 0-to-5 scale.Source: B. Multivariate Analysis of the… · p. 6
  • rtProfile (p-value 0.0515) and rtVision (p-value around 0.1) also contribute, with unit increases corresponding to odds-ratio factors of about 0.4 and 0.7, respectively.rtProfile is a robot-assigned index mainly influenced by rating.Source: B. Multivariate Analysis of the… · p. 6
  • p-value 0.0004 and an odds-ratio factor of about 0.4 associate Ethereum with lower success odds on average than other platforms.The analysis notes that rare platform categories can produce extreme odds ratios without statistical significance.Source: B. Multivariate Analysis of the… · p. 6
  • After setting aside platforms observed once or a few times, Waves is the only common-platform comparison to Ethereum and shows no significant success contribution.Rare platform categories can yield exceptionally high or low odds ratios despite lacking statistical significance.Source: B. Multivariate Analysis of the… · p. 6
  • Slovenia and the USA contribute odds-ratio factors of about 5.2 and 2.2, with p-values of about 0.016 and 0.038, respectively.Israel and China are also described as favorable, but with less significance.Source: B. Multivariate Analysis of the… · p. 6
  • The software category positively contributes to success, with a p-value of 0.0325 and an odds-ratio factor of 3.1.Gambling and business have lower statistical significance.Source: B. Multivariate Analysis of the… · p. 6
  • Team size does not appear to determine success or failure, including when tested using several size-category schemes.The analysis includes the overall team, not only software developers.Source: B. Multivariate Analysis of the… · p. 6

IV. CONCLUSIONS

The study analyzes ICO characteristics and success associations alongside Ethereum token activity. It finds that a few tokens account for most observed blockchain workload and identifies several factors correlated with ICO success.

IV. CONCLUSIONS

  • The study examined 1,388 ICOs, finding that most used Ethereum and the ERC-20 standard.It drew ICO information from icobench.com, token prices from coinmarketcap.com, and Ethereum transaction information from ethplorer.io.Source: §IV · p. 6
  • Token transfers and holders follow a fat-tailed distribution resembling a power law, with a few tokens accounting for most blockchain workload.The authors describe this distribution as common in technological and financial data.Source: §IV · p. 6
  • The multivariate analysis associates success with origin in Slovenia or the USA, higher icobench.com ratings, and Ethereum, while team size appears irrelevant.The authors say the rating association seems mainly due to the robot’s advice rather than human experts’ advice.Source: §IV · pp. 6–7
  • Future work will gather ICO data from additional sources to check validity and examine Ethereum transactions alongside token price and volume.Source: §IV · p. 7
Loading 1803.01394v1…