Recently, I tested all the AI prediction models. Basically, it's really hard to predict the underdogs. Right now, Qianwen and GPT are the best, especially Qianwen—it got it right several times when others didn't. I trained my GPT with my database and some knowledge, and now it's super accurate at making on-the-spot judgments about certain levels. This all takes training. I have to say, in the past, every time I encountered different situations, I had to calculate and convert things myself. Now I just feed it to my AI, and it knows instantly.