How long until models are universally, among other things, generated via machine learning? Interested to see what ML does to the future of IB with it becoming prominent in the space more and more every year. Eventually they’ll be able to consolidate.
Need some of my BB guys / ladies to weigh and tell me if it’s prevalent on the street yet. I know it is in ER and within HFs but curious if it’s in IB yet
| +67 | Any tips for sleeping well? | 36 | 4h |
| +50 | Wealthy Parents / Jaded | 13 | 12h |
| +40 | Background check issue IB Full Time | 21 | 11h |
| +36 | Venezuela Debt Restructuring - LAZ vs CVP | 12 | 3h |
| +34 | How are hours rn for SA (interns) | 21 | 1h |
| +28 | Q dealflow slowdown | 22 | 2d |
| +25 | HL vs WB | 19 | 15h |
| +24 | Investment Banking in Mexico | 6 | 2h |
| +23 | UBS GIG, LevFin, FSG Interns working Sundays and 85 hour weeks their first week? | 11 | 16h |
| +22 | MM bottom bucket bonus | 9 | 13h |
Career Resources
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.
I came from a stats background and have very mixed feelings about machine learning.
ML is a black box for the most part. It is difficult if not impossible to explain why ML models make the choices they do.
ML is incredibly useful for fields like advertising, marketing, etc where not much money is riding on model outputs. Finance is tougher. ML will fly in the HF space where people are much more risk tolerable, but when it comes to M&A or capital markets offerings, its much harder to trust a model that no one can explain.
TLDR is that ML is very cool, but indivual decisions made by models are difficult to justify with any kind of risk framework.
Interesting - I’m in a very heavy MA environment (like most on this forum) and have really thought about diving into ML to just make modeling simply less a component of my workload. Guess this could also be useful for prospecting and such. Just trying to find automation in my day to day lol
I definitely understand the drive to automate/be more efficient. My last job was managing a top of the house style credit team at a commercial bank.
We had a quant who could build machine learning models that kicked the ever loving shit out of FICO. The problem was that we couldn't really explain why it was rejecting certain borrowers. Which is a huge regulatory no no. Also just generally a risk management no no.
Even if the models were legally useable, if they ever broke it would be hard for us to know until it was way too late and losses would have been enormous.
At the end of the day, our quant wound up using machine learning to explore variables and tune their weights in house built models. So, still useful, but our final models were never ML.
There's definitely a line we need to walk between automation/efficiency and transparency. And its hard. Lol. Just my 2 cents.
What if they ran a model on the side with actual bankers and tested results. I’d imagine if it had accuracy similar to it and could drive the M&A processing fees down enough they might be interested.
Ehhh maybe. You might be missing my point a bit though. When multi billion dollar transactions are on the line, people trust what they can understand, and people are willing to pay a lot to understand. "Because the model said so" isn't good enough anymore. Especially post GFC.
This is all assuming regulators would even permit ML models to be used to make decisions, which historically has not been true. You can't even legally use a ML model to determine whether or not John Doe of Springfield Iowa should qualify for a $5000 credit card.
ML is most useful when regulations and transparency are not a concern. I dont think M&A is that place.
Quia quae temporibus voluptatem dolorem exercitationem ab. Porro eligendi minus fuga exercitationem ad non aut accusamus. Laboriosam fugiat et eos at vel vel ad.
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...