Jenny Jun has recently written several AI strategies, and her biggest feeling is that the market is constantly changing. The strategy you wrote last month may not be applicable this month. For example, Johnny Jun had a volatile strategy with a win rate close to 100% before, which was basically in a resting state after BTC broke through. Not issuing orders is an advantage, but also a disadvantage, and there are still areas worth optimizing. A good AI platform and strategy must be constantly iterated, just like the newly launched GPT Astra, which is also constantly upgraded, and the cryptocurrency industry is no exception. Have you not seen how various AI strategy platforms are like fish crossing a river, and how difficult it is for people to succeed? Therefore, I find the self evolving upgrade of NeoTrade @ NeoSoulAI very interesting and worth paying attention to. Their solution is that the agent looks back at the history they have run, leaving only the signals that actually worked, writes them back to memory, and continues running with these contexts in the next round. Official summary: Same Agent, more context, less repetition of the same lesson. But what I value more is how it ensures that this review is not just a formality. They keep all the information, timestamps, configurations, skill versions, orders, and transactions related to the decision-making process on file. Afterwards, you don't have to ask the agent what you were thinking at the time, you can directly access the records. And all attempts, including failed versions, must be recorded in the experiment, otherwise the system will only display the filtered successful results, making it easy to mistake accidental performance for progress. I think in the field of AI strategy, the next challenge is not about whose model is strong, but about whose verification mechanism is honest and whose optimization is reliable. Stop talking, Jenny is planning to optimize my strategy!