Phase 1: Advancing the Trading Analysis Platform

Financial authorities will use artificial intelligence (AI) to detect unfair trading of virtual assets in real time, down to the second. This is expected to improve both the accuracy and speed of future investigations into unfair trading.


As former President Donald Trump won the U.S. presidential election, Bitcoin continued its soaring run, surpassing $76,000 for the first time ever and reaching a new all-time high. On the 7th, the price was displayed on the market board at the Upbit Customer Center in Gangnam-gu, Seoul. Photo by Jo Yongjun

As former President Donald Trump won the U.S. presidential election, Bitcoin continued its soaring run, surpassing $76,000 for the first time ever and reaching a new all-time high. On the 7th, the price was displayed on the market board at the Upbit Customer Center in Gangnam-gu, Seoul. Photo by Jo Yongjun

View original image

The Financial Supervisory Service announced on February 2 that it has independently developed an algorithm (Suspicious Section Auto-Extraction Algorithm) that automatically detects sections suspected of price manipulation, which were previously identified manually. The agency plans to apply this algorithm to abnormal trading detection and the identification of suspicious cases.


This system features a 'sliding window grid search' that divides a suspect's trading period into multiple detailed sections and automatically executes abnormal trading detection for every section. As a result, it can swiftly analyze hundreds of thousands of sub-second intervals and identify all suspicious sections, regardless of the frequency or duration of price manipulation.


Performance tests showed that the new algorithm detected more suspicious sections than human investigators. It also uncovered sections that were difficult for investigators to identify, thereby improving both the accuracy and speed of investigations.


The Financial Supervisory Service plans to gradually expand AI analysis functions by the end of this year to detect unfair trading of virtual assets at an early stage and handle cases quickly. To this end, the 170 million won budget secured this year for server expansion will be used to improve the performance of AI-related platforms.


Following this first phase, the second phase will use clustering algorithms to automatically extract groups of suspicious accounts in response to organized price manipulation through collusion. In the third phase, a large language model (LLM) based on open-source technology will be used to analyze news articles and other content to identify which assets could be targets for fraud. The fourth phase will analyze on-chain data such as suspects' wallet addresses and fund transactions, and establish a tracking support system to prioritize accounts for analysis among the vast number of accounts.



A Financial Supervisory Service official stated, "We will enhance our AI-based investigation system to swiftly detect unfair trading in the virtual asset market. We are committed to thoroughly investigating and taking strict action against any unfair trading we uncover, in order to protect users and establish a sound market order."


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

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing