
PANews reported on June 15th that according to @ai_9684xtpa's analysis, ZKJ and KOGE's flash crash tonight might be a long-planned harvesting operation. Three main addresses targeted the massive trading volume and liquidity of two tokens under the Binance Alpha background, causing ZKJ and KOGE to collapse sequentially through the dual pressure of "large-scale liquidity withdrawal + continuous selling", with no exceptions.
1. Address 0x1A2...2759920:28:21 and 20:33:15 withdrew bilateral liquidity of 61130 KOGE (approximately $3.76 million) and 273017 ZKJ (approximately $532,000) in two stages
Between 20:28:58 - 20:36:57, converted 45470 KOGE to ZKJ, valued at $3.796 million, during which KOGE's on-chain transaction volume showed significant growth
Between 20:30:57 - 20:59:49, sold 1.573 million ZKJ in batches for USDT and BNB, valued at $3.052 million, with an average selling price of $1.94
At this point, KOGE and ZKJ experienced slight step-like declines but did not crash.
2. Address 0x078...8bdE720:30:33 withdrew bilateral liquidity of 33651 KOGE (approximately $2.07 million) and 709203 ZKJ (approximately $1.38 million)
Between 20:31:10 - 20:58:18, converted 36814 KOGE to ZKJ, valued at $2.26 million
Between 20:35:15 - 20:37:34, sold 1 million ZKJ, valued at $1.948 million, with an average selling price of $1.948
This address's "relay-style dumping" finally triggered a rapid decline in KOGE's price, which is the consecutive large red candles everyone saw.
3. Address 0x6aD...e2EBb20:41:55 received 772759 ZKJ from address 0x078...8bdE7 (the previous dumping address), valued at $1.5 million
Between 20:42:28 - 20:50:16, liquidated 772,000 ZKJ
The third address's main role was to cooperate, further catalyzing ZKJ's decline after KOGE's price crash, completing the harvesting of both tokens' LP and token holders
Finally: The three addresses mentioned here are the main driving addresses. In fact, Ai found more "cooperative addresses" worth hundreds of thousands of dollars, which will not be elaborated here, as the dumping methods are similar.




