Have fully working proprietary high frequency (JS written) bot that tracks newly created pools real time and enters within first seconds of pool creation interacting directly with Raydium program (not relying on their trade api). Most pools with good amount of SOL deposited have price action similar to boom and bust index - slow steady growth then sudden drop. Really struggle to determine points of time when large sell will happen. Front running is likely impossible considering solana block time (400 ms). There is a tradeoff between holding time and frequency of trades - capturing 5% on each pool reduces risk of facing dump but requires more than 15-20 successful trades to be above breakeven. Any ideas how to work around this problem? Gathering historical data and trying to get average pump duration (assuming no drift to shorter pump duration due to degradation of the market)? Or maybe some on chain data that may help to evaluate odds or contains signals that dump time is approaching? Rugcheck stuff is useless in my view since locked liquidity prevents liquidity removal, not sale of tokens Any thoughts are appreciated.