Anyone using AI for their modeling?
Trending Content
Career Resources
WSO Free Modeling Series - Now Open Through
May 31, 2024
9 Free Financial Modeling Lessons
Get instant access to lessons taught by experienced private equity pros and bulge bracket investment bankers including financial statement modeling, DCF, M&A, LBO, Comps and Excel Modeling.
+Bonus: Get 27 financial modeling templates in swipe file
+Bonus: Get 27 financial modeling templates in swipe file
As a former programmer in IB, no. financial modeling is meant to be auditable and reproducible by peers, superiors, 3rd parties, etc. and AI is generally a black box - the exact opposite of that.
I have heard that programming (not even AI, just like number crunching in python or something) is becoming more slowly more common at l/s hedge funds, but even then it doesn't replace core financial modeling in excel (is at most a way to supplement and confirm an existing thesis). And you'd probably need a really forward thinking and technical PM to not tell you to fck off when you propose to pull out python for anything
Helpful - couldn't you use it to help your modeling though? e.g., debug or do the more mundane things (e.g., copy paste)?
I use it all the time for answering questions about how to structure things out. It's great for questions, but never for building out the entire model. If I'm trying to get my balance sheet to balance, for example, and it's just not, then I will go through each assumption and ask where it go wrong, and it helps me debug, like having another Associate next door that I chat with for some help. I'd love to see something where I put in my assumptions and it kicks out an actual excel, though, that would be sick.
What platform are you using for this?
Vanilla ChatGPT.
I disagree with above. LLM can generate VBA codes, and Excel is essentially a human-friendly UI on top of bunch of unreadable, hard-to-comprehend VBA code. So this is how it will work: LLM generate VBA code based on natural language prompts, than Excel converts VBA code into human-readable format ( traditional financial models - tabular UI with math formulas ), then human do adjustments such as fixing inappropriate stuff generated by AI blackbox, then the LLM in the background utilized reinforcement learning to reflect on real-time human feedback ( which is the adjustment that human made to the model ) such that LLM won't fall for the same mistake twice, and it will only get better the next time. This entire thing hasn't been built yet but it's theoretically possible, and there's a start-up doing this stuff.
That is fair - but the pain of fixing one imperfect model is greater than building three models from scratch.
Adjustments can be done in natural language as well, so you may never need to get your hands dirty on excel stuff. Imagine you are the VP who leads a deal. Which one are you a going to choose: 1. Build all models from scratch by yourself because you're so good at it. 2. Throw model work to empty-headed analysts and assos and provide one thousand comments when they do a lousy job. 1 is more efficient, but all VPs choose 2 lol. Because it just feel so fucking good to not do anything from scratch and be the one who sits back and judges other people's work lol. It's human nature.
Unde quae aut repellendus animi iure voluptatem atque excepturi. Et aut tempore consequatur libero. Vel exercitationem est sed. Et non nulla sit saepe reprehenderit. Quia iure recusandae et ad facilis voluptatem autem. Aliquid qui vitae omnis et cum numquam.
Qui et sunt et corrupti numquam et ipsam. Possimus corrupti doloremque nihil earum rerum. Harum aliquid eius in aspernatur. Vel ab excepturi sit dicta non natus pariatur.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...