Recently, I've been working on small projects in three directions:

For data, the main goal is to create an auxiliary tool for product research, so that I can quickly validate some ideas later on.

On the strategy side, I'm exploring Poly's mechanisms as I go. Although I've written several strategies, actually running arbitrage is still quite difficult to implement (for reasons you understand). However, this process has definitely deepened my understanding of the underlying logic.

I'm most satisfied with the information processing part. Two years ago, I tried using ChatGPT for this, but it was complicated and the results were mediocre. The new version I'm working on now is much better. Next, I'm considering integrating a Feishu group chat summary feature, which should make it even more practical.