Dataset 'MemeTrans' Released to Detect Risky Memecoins

- The "MemeTrans" dataset encompasses over 40,000 memecoin launches on the Solana network that have migrated to a public Decentralized Exchange (DEX). It analyzes more than 30 million transactions during the initial launchpad sale and 180 million transactions post-migration. - To identify high-risk patterns, the dataset utilizes 122 distinct features. These features analyze aspects like trading activity, holder concentration, and time-series dynamics to flag potentially malicious launches. - A key innovation of MemeTrans is its use of "bundle-level data," which can reveal multiple wallets controlled by a single entity, helping to uncover hidden coordination and manipulation. - The academic paper introducing the dataset reports that machine learning models trained on MemeTrans can effectively reduce financial losses from high-risk memecoins by an estimated 56.1%. - The need for such a tool is highlighted by research from crypto monitoring firm Solidus Labs, which found that approximately 99% of tokens on the Solana launchpad pump.fun and 93% of liquidity pools on the Raydium exchange showed characteristics of pump-and-dump schemes or rug pulls. - Scams on Solana have become increasingly prevalent, with scammers shifting from the BNB Chain to Solana due to the popularity of memecoins. In April 2024 alone, 27 presale scams on Solana resulted in over $122.5 million in losses, according to on-chain analyst ZachXBT. - The dataset specifically tracks the lifecycle of memecoins issued through the popular launchpad Pump.fun between December 1, 2024, and March 1, 2025. By October 2025, Pump.fun had been used to launch 12.8 million memecoins. - Unlike previous detection methods that focused on liquidity pool events on DEXs, MemeTrans is designed to analyze the new paradigm of launchpads where developers do not have direct control over the liquidity pool during the initial sale, making older detection methods less effective.