Adrian Reid sorts his worst trades in RealTest, then screenshots each one to see entry and exit points. Screenshots reveal context that trade lists miss, letting him generate five coded improvements like adding index and regime filters. He plans to use Claude for brainstorming but wants to build his own image-analysis skills.
So in RealTest, I sorted the trades from worst to best. I went and looked at the worst 5 trades, and I copied the image of the trade where it shows the entry and it shows the exit. Now I want to be honest, it didn't come up with rocket science, but nor did I expect it to. But it did analyze them pretty quickly, and it certainly did save me some time, and it gave me 5 testable ideas that were fully coded, which is cool. One of the ideas was just use an index filter your mug because you didn't have one, a regime filter in your system. So I did that and it made it a lot better. It can speed it up. I would definitely do this with Claude in the future, but I would also do it myself as well. Because I think once you've got some practice at this, you can probably go a bit deeper. And I think I can probably also create a deeper skill to analyze the images and look at what's going right and wrong in the trades, but I haven't done that bit yet. If you give the trade list, it knows, it knows what you give it, right? So it knows the entry date, the entry price, the exit date, the entry price. It doesn't know what happened before, during, and after the trade. Whereas if you give it the image, which is zoomed out, I think you could do both, but I wouldn't give it just the trade list because it's not enough information. Can you brainstorm how to improve a system by just looking at the trade list? Probably not really. You might get some ideas, but it will be incomplete. So that, that's why I use the images. Yeah. Now, this is the first time that I've done this with brainstorming improvement ideas. So I'm gonna come up with better ideas, no doubt, but this is a starting point.
