Document Type : Research Paper
Authors
1 Ph.D. Candidate, Department of Management and Economics, Tarbiat Modares University,Tehran, Iran.
2 Professor of Economics, Department of Economics, Yazd University, Yazd, Iran.
Abstract
Financing remains one of the most critical aspects of business growth and sustainability. The Initial Coin Offering (ICO) method, a novel approach to financing leveraging blockchain technology, has garnered attention due to its ability to attract significant capital globally within a short period and without requiring intermediaries. Understanding and analyzing the factors that influence the success or failure of ICOs is thus valuable for businesses and investors alike. This paper examines the factors impacting ICO success using logistic regression analysis, focusing on 307 completed ICO projects from 2016 to 2018. We consider two target success variables: "Total funds collected" and "Hard cap achievement percentage." Factors related to the project, campaign, social networks, and team characteristics were analyzed in separate models. Through model selection based on performance and feature prioritization using the Permutation Importance (PI) technique, the findings highlight that having a well-defined "Business model available" significantly contributes to ICO success across both models. Additionally, the top features in the first selected model under the categories of project, campaign, and social network are "White paper pages," "Token share for presale investors," and "GitHub account," respectively. In the second model, the most impactful factors are "Use of proceeds mentioned" and "Length of crowdsale" under the project and campaign categories.
Introduction
Creating a stable foundation for business growth requires sustained financial support, as financing is a central pillar of business continuity. The Initial Coin Offering (ICO) method represents an innovative financing avenue, leveraging blockchain technology to enable rapid capital acquisition from a global investor base without intermediary involvement. By examining the mechanics of ICOs and analyzing factors that influence their success or failure, this study aims to expand theoretical insights into ICOs, prevent inefficient trends, and promote the optimal utilization of this financing method.
Methods and Material
To identify factors impacting the success of ICO campaigns, this study collected data on 307 ICO projects from 2016 to 2018, categorizing them into four dimensions: project, campaign, social network, and team characteristics. Two target variables were used to assess ICO success: "Total raised capital during ICO" for the first set of four models, and "Percentage of hard cap (the maximum capital set by project founders) raised" for the second set. The logistic regression algorithm, a supervised machine learning technique for binary classification, was employed to predict the probability of success or failure. A cumulative process approach was used to create and analyze the research model.
Results and Discussion
After optimizing logistic regression structures across eight research models, the highest prediction accuracy was observed among the first four models using "Total raised capital during ICO" as the dependent variable. Models 1-3, with independent variables focusing on project, campaign, and social network characteristics, achieved 70% accuracy based on the weighted average of Precision for both success and failure groups and 72% for Recall. This result indicates the model’s high effectiveness in predicting unsuccessful ICOs.
In the second set of models, using "Percentage of hard cap raised" as the dependent variable, model 2-2 (with independent variables for project and campaign characteristics) showed the best performance, achieving 67% and 74% accuracy for precision in both groups and 93% Recall accuracy for unsuccessful groups (31% for successful groups). Consequently, models 1-3 and 2-2 were selected for their high accuracy in predicting ICO campaign success or failure.
The Permutation Importance (PI) technique identified the top influential features in each model. For model 1-3, the most critical factors were "White paper pages," "Certainty of presale token share for investors," and "Project business model accessibility." In model 2-2, the leading features included "Mentioned use of proceeds," "Availability of business model," and "Crowdsale duration." Notably, in model 1-3, the most effective features by category were "White paper pages" (project), "Token share presale investors" (campaign), and "GitHub account" (social network). For model 2-2, the prominent features were "Use of proceeds mentioned" and "Crowdsale length" within project and campaign categories, respectively.
Conclusion
The findings reveal that the most influential variables across models 1-3 and 2-2 are:
Project category: "White paper page count" and "Use of proceeds mentioned."
Campaign category: "Token share presale investors" and "Crowdsale length."
Social network category: "Active GitHub account availability."
Key features, such as "White paper pages," "Token share presale investors," and "GitHub account" in model 1-3, along with "Use of proceeds mentioned" and "Crowdsale length" in model 2-2, emerged as the most significant factors for ICO success. The prominence of campaign characteristics in both optimal models underscores their critical role in ICO outcomes. This research suggests that addressing informational gaps and identifying success factors can facilitate the responsible and effective adoption of ICOs. By focusing on these pivotal features, businesses can enhance their likelihood of success while streamlining their financing strategies through ICOs.
Keywords
)