FSS Enables Real-Time Monitoring and Analysis of Unfair Trading in Virtual Assets

Second Phase of Virtual Asset Unfair Trading System Upgrade

The Financial Supervisory Service (FSS) can now analyze suspicious transactions more quickly by utilizing artificial intelligence (AI) based on a real-time monitoring system to prevent unfair trading in the virtual asset market.

On May 3, the FSS announced that it has completed the second phase of upgrades to its investigation system for unfair trading in virtual assets, which includes these enhancements. This follows the first phase in January, which introduced features such as the automatic detection of suspicious activity intervals.

The upgrades are being implemented to prevent unfair trading in the virtual asset market, where transactions occur 24 hours a day and price volatility is high. In particular, as trading using application programming interfaces (APIs) becomes more prevalent and unfair trading tactics become more sophisticated, the upgrades aim to strengthen monitoring and post-trade analysis systems.

Through the second phase of upgrades, a real-time monitoring system for virtual assets has been established. Not only can the FSS collect and analyze data such as order books, trade execution information, and market alert issuance histories from five domestic virtual asset exchanges like Upbit and Bithumb, but it can also do so from three overseas exchanges, including Binance and Coinbase, via their APIs. The FSS also plans to focus on developing and expanding the application of functions that detect unfair trading and identify suspicious items within the first half of this year.

An AI-powered trading analysis platform has also been established to automatically identify groups of suspected unfair trading based on comparisons of trading methods, order timing, and trading media. Previously, unfair trading activities had to be analyzed manually, but moving forward, the system will automatically analyze such behaviors, group accounts exhibiting similar activities, and identify them as suspected groups.

Performance tests of the trading analysis platform showed that it was able to accurately classify account groups involved in market manipulation cases. However, the FSS acknowledged that there are still limitations in precisely classifying small groups, and plans to improve performance by optimizing system parameters as more investigation cases involving various types of suspected groups accumulate in the future.

An FSS official stated, "The real-time monitoring system will enable us to identify stocks showing signs of unfair trading in advance by developing automatic detection algorithms for suspicious items and text analysis functions using large language models (LLMs). For the trading analysis system, we also plan to add features that can identify wallets or accounts requiring further investigation through on-chain and fund flow analysis."

This content was produced with the assistance of AI translation services.