Research On Generation,screening And Prevention Of Fraud Risk In Initial Coin Offering（ICO） Of Blockchain

Blockchain is a very hot topic at present and has a wide application prospect.On June 7,2021,the Ministry of industry and information technology and the office of the central network security and Information Technology Commission jointly issued the guidance on accelerating the application of blockchain technology and industrial development.The guidance proposes to promote the development of blockchain technology from policy support,industrial implementation,talent training and other aspects,and make blockchain an important support for national development by 2030,which reflects the great significance of blockchain technology for the future development of the country.Initial coin offering（ICO）is a way to raise funds for blockchain projects,with the characteristics of fast financing speed and wide range of financing,providing an efficient and convenient new way for start-ups to raise funds.However,this financing method has been abused and become a means for speculators to seek illegal interests.Although financial regulators in many countries and regions around the world have issued risk warnings to remind investors to improve risk prevention awareness and guard against their existing risks,driven by the myth of "getting rich overnight",ICO is still hot,and there are huge risks in the investment and trading of virtual tokens.Therefore,it is of certain significance to explore the screening model of fraud risk in initial coin offering.In order to achieve the research purpose,this paper has done the following work.This paper takes Hayek’s free money theory,tulip effect theory and signal theory as the theoretical basis.This paper uses Hayek’s free currency theory to reveal the essence of the current virtual currency,and uses the tulip effect theory to explain why investors are so keen on investing in ICO projects that they make irrational behavior.Based on signal theory,this paper analyzes the principal-agent problem,adverse selection and moral hazard caused by information asymmetry in the process of ICO.According to the signal theory,this paper studies the efficiency of relevant factors in the process of ICO fraud risk identification from the perspective of information disclosure.Specifically,the fraud risk identification index system of the Initial coin offering is constructed based on the five aspects of the project’s issuance data,project overview,project development progress,project team and ecological operation.In this paper,a total of 12 primary indicators were selected and 118 samples were collected manually.Among them,there were 68 fraudulent items and 50 non-fraudulent items in Initial Coin Offering.The samples were divided into training group and inspection group in a ratio of 8:2.In the training group and the test group,the proportion of the sample size of ICO fraudulent projects and the sample size of non-fraudulent projects is basically the same.Then,descriptive statistical analysis of the data shows that there are significant differences between fraudulent ICO projects and non-fraudulent ICO projects in most indicators.Using single sample K-S test for normality test,the results show that the population from which the sample comes does not accord with the normal distribution.In this case,this paper uses Mann Whitney U test（a nonparametric test method）to analyze the difference between the two groups of samples.We found that there were significant differences between ICO fraud projects and non-fraud projects.The correlation test and multicollinearity test are carried out for these 12 indicators to ensure that there is no multicollinearity among the indicators entering the logit model.Finally,seven indicators are selected to enter the construction process of the logit model.From the results of logistic regression,the significance level of three indicators is less than 0.05,which are the integrity of project information disclosure,the cumulative submission times of Git Hub and the reliability of team member information;The significance level of two indicators is less than 0.1,namely turnover rate and currency concentration.Next,compare the accuracy of the model under different thresholds.When the threshold is 0.4,the prediction effect of the model is the best,and the overall prediction accuracy reaches92.6%.Therefore,this paper selects 0.4 as the threshold of the model.Finally,ROC curve and confusion matrix are used to verify the validity of the model.The AUC value of ROC curve is close to 1,and the overall prediction accuracy of the test group has reached 87.5%.The results of the two test methods show that the blockchain Initial coin offering（ICO）fraud risk identification model constructed in this paper has a good prediction effect.Through the above work,this paper draws the following conclusions.The turnover rate,the integrity of project information disclosure,the cumulative submission times of Git Hub,the reliability of team member information and the concentration of money can significantly affect whether an ICO project is a fraud project.When these five indicators are brought into the model,the prediction accuracy of the model is also relatively high,which shows that the logit model established in this paper can be used to screen the risk of ICO fraud.This paper enriches the research results in the fields related to the Initial coin offering,and provides a quantifiable tool for identifying the fraud project of the Initial coin offering,which has certain theoretical and practical significance.