I'm studying quantitative finance, and honestly, it's really hard on my own. Are there any Italians here?
I'd love to find someone to talk to, study with, and maybe even create a project with..
Does anyone else run into this problem when using AI (especially Claude) for quantitative finance projects?
I use AI mainly to write code for my projects. The frustrating part is that even relatively simple ideas, sometimes involving only basic mathematics, quickly turn into huge codebases full of bugs, inconsistencies, and unexpected issues.
It often feels like the model can't keep all the different parts of the project connected. Fixing one problem creates two more somewhere else, and after a while you're stuck in an endless debugging loop.
Instead of converging toward a working implementation, the project seems to drift further away from a correct solution with every iteration.
It's like an infinite loop where each project never sees a proper end.
Is this a limitation of current AI coding models, or am I approaching these projects in the wrong way? How do you structure your workflow to avoid this?
Does anyone else run into this problem when using AI (especially Claude) for quantitative finance projects?
I use AI mainly to write code for my projects. The frustrating part is that even relatively simple ideas, sometimes involving only basic mathematics, quickly turn into huge codebases full of bugs, inconsistencies, and unexpected issues.
It often feels like the model can't keep all the different parts of the project connected. Fixing one problem creates two more somewhere else, and after a while you're stuck in an endless debugging loop.
Instead of converging toward a working implementation, the project seems to drift further away from a correct solution with every iteration.
It's like an infinite loop where each project never sees a proper end.
Is this a limitation of current AI coding models, or am I approaching these projects in the wrong way? How do you structure your workflow to avoid this?
Does anyone else run into this problem when using AI (especially Claude) for quantitative finance projects?
I use AI mainly to write code for my projects. The frustrating part is that even relatively simple ideas, sometimes involving only basic mathematics, quickly turn into huge codebases full of bugs, inconsistencies, and unexpected issues.
It often feels like the model can't keep all the different parts of the project connected. Fixing one problem creates two more somewhere else, and after a while you're stuck in an endless debugging loop.
Instead of converging toward a working implementation, the project seems to drift further away from a correct solution with every iteration.
It's like an infinite loop where each project never sees a proper end.
Is this a limitation of current AI coding models, or am I approaching these projects in the wrong way? How do you structure your workflow to avoid this?