Dear L/S Monkeys,
Unless you’ve been living under a rock, you all probably know that the AI is coming in fast.
In light of this, how do you see this impacting your roles in the Long Short space, any reason for it to impact certain industry sectors over others?
Is it net-net making your lives easier in terms of dats processing? Is it accurate?
How are some of the MF HFs using this to gain an edge?
Thanks a lot — other recruiting related advice would also be much appreciated! :)
Ah, the AI conundrum! It's like a barrel of monkeys, but instead of monkeys, it's algorithms and data models all the way down. Here's the scoop based on the most helpful WSO content:
Impact on Long/Short Strategies:
Accuracy and Efficiency:
Use by Mega Funds (MF HFs):
Recruiting Advice:
Remember, while AI might be the shiny new toy in the finance playground, the ability to interpret and strategize around the data it provides is where the true value lies. Keep swinging through the vines of knowledge, and you'll stay ahead in the game!
Sources: Work at top AM -- AI will virtually kill of entry-level roles in AM over the next 5yrs, Will robots replace your consulting or financial career?, Q&A: London L/S + event-driven analyst, https://www.wallstreetoasis.com/forum/hedge-fund/machine-learning-taking-over-hf-research-analyst-roles-in-near-future?customgpt=1, As a sector specialist, what to do when you sector is out of favor?
The irony
💀
MF HF is a new one
Quite a few funds are integrating AI LLMs
Interesting — have you seen by any chance which ones?
Until the AI conundrum, there was a lot of advice being thrown around that you need to have a programming foundation to be successful at a fund, or atleast gain an edge.
With that out of the way, would you say that this statement has been rendered obsolete.
Thanks!
Programming has never been a necessity for edge in long/short lol, don't know where you're getting that from. Generally speaking, if you need big data tools your team will have a dedicated data analyst to build it out and help you use it. Your main goal as an analyst is to know the narratives and drivers of your names inside out. Most of the big MMs will have these.
Point72 and Citadel have a proprietary form chatgpt they use; its not game-changing or anything. Just a quality of life tool ime.
also the term MF HF isn't a thing
In a different area of finance > no question about it > ai will change things. However, I don't think ai is at the level to give presentations to execs bc it lacks contextual understanding. Also, if you are in corporate finance or government finance, you need to understand laws very well. And AI doesn't do that well with regard to legal understanding, particularly because it lacks the ability to understand context. Certainly won't replace auditing or anything like that either going forward. And it won't replace investment banking partners/a lot of sell-side management roles because those are more salesmen roles that require human touch. Work on the sell-side will def be streamlined at lower levels though. So key is def to get into a leadership position if you are on sell-side/in corporate/in government. I def believe that ai will narrow the field of investment strategies that are purely human-driven. There was also a study showing that if you have a remote job/job that could be done completely virtually, your job is more amenable to ai risk. Google is having ai layoffs right now along with some other tech companies -- they are still retaining software engineers, but ai is increasing efficiency in terms of coding development. So I do actually see a potential shrinking of the buy side by ai but not 100% replacement. And the shrinking starts at the bottom. For comparison, pretty much 100% of the work that paralegals do -- the lowest on the totem pole at big law firms -- will be automated by ai. That's probably like 70% of the work in buy-side finance, 30% in sell-side finance, 30% in auditing, 50-60% in corporate/government finance. Time will tell though -- never thought I would see ai as developed as it has become in my lifetime -- could be worse or much better -- but I think my estimates are relatively conservative. (could be wrong though) Really it will change a lot of things not just for finance but for our economy overall. Waymo is already out in certain cities
The post was about L/S, lol
Solid color — your take makes a lot of sense. Realistically, I can imagine this making the field more competitive and purely returns oriented as opposed to scraping data. With that said, it should be an interesting landscape to navigate and I wonder how it will impact priorities in recruiting.
I am probably the exception here but I basically think LLMs will be able to do everything a junior analyst does but better faster and cheaper in 5 years. By this I mean updating models and a lot of the basic analytical work. Whether the industry adopts the tools at traditional long short funds is a different story. I actually think people are pretty slow to adopt innovation and most people will share the attitudes mentioned here for longer than is economically optimal for them to do so. I also think same holds for what most banking analysts and associates do.
LLMs popular today don’t exhibit agentic behavior / executive function or use techniques like RAG by default nor do they have vision enabled by default. (These aren’t really “language” models - other input forms work just fine with LLMs.) They also are still somewhat bad at math. Solutions for all these issues exist and when they are combined imo it will feel sudden and you’ll probably feel like how a lot of graphic designers feel now. If you think they can’t update an excel file or read research you’re in for a surprise.
Intelligence has always been a rare and important commodity but this hasn’t always been the case. I do think this will rapidly change.
that said even at an extremely slow rate of adoption relative to what’s optimal I still think it will be fast. The companies selling tools to financial folks all have clown / almost unusable products that are thin wrappers around the simple tools available to the public but they’re all growing like 1000% per year at $10k per seat, which imo is demonstration of demand and value people get out of even garbagr versions of these tools that are 1% of what will soon become standard.
If you’ve used the cutting edge of these tools for the last few years you’ll agree with me. All the elements are essentially there but it can be hard to connect the dots unless you’ve spent enough time trying to do so. The problems you guys assume exist are all solvable and it’s a matter of folks just putting them together.
I agree that LLMs will do almost everything better than an analyst. However, this doesn't mean the jobs will go away or pay less. Once a firm has a certain org structure and way of operating, it's very difficult to change that. The work quants do is even more commoditized, but the jobs didn't go anywhere.
I would say though that analysts will have even less leverage over the firm, and have to accept whatever pay the firm gives them. This happened in quant, the total comp did not fall, but became discretionary in virtually all the firms.
None of the problems with LLMs (namely that they, by design, hallucinate answers) are easily solvable and the junior analyst role isn't going anywhere. The value add of an analyst in terms of thinking/types of analysis they can do/types of information they can get you is pretty substantial and really difficult to replace with AI. Yes AI can maybe be used to update models in the future (isn't really great at it now), but even thinking through the main question an analyst gets "tell me if this a good stock" is an analytical exercise that an LLM isn't capable of at this point, and likely structurally will not be capable of for a long time. LLM's ~could~ augment the jr. analyst toolset but that likely just pushes analysts into higher level/higher value add tasks, which has happened anyway with tech innovation for decades.
Exactly. This post is spot on. Analysts do a lot of drudge work. Updating models, writing research notes from scratch. I did start trialing Daloopa for auto-model updates, and it works very well. I also built my own conference call summarizer/chat bot, so that when I get jammed with earnings, I don't have to stay up late anymore adding cc notes to cells. I just copy/paste. Plus it helps get up to speed on new names.
Interesting answer - thanks for sharing. I have tested some of these chat-gpt wrapper like products and was pretty underwhelmed.
Would be great if there was something that was good enough so you can simply ask questions like "it looks like gross margins expanded by 400bps over the last 5 years, what were the main drivers?", and it can incorporate not only all filings but also all transcripts, conferences, notes, sell side research, and ideally, the operating model. Is this pretty much what you mean here? I have tried a few things here and nothing seems to really do it yet. Would also be amazing if it can find similar examples across the history of stocks and do some pattern recognition here... I also remember seeing that demo video where they ask the gen ai to auto-project the financial model and it basically fills it in? Then you can spot check + tune it up, and 90% of grunt work is handled really quickly, leaving you to focus on higher level issues.
In an optimal world where all of these solutions are as successful and strong as we are making them out to be, it seems like the junior analyst will still exist, but their job will be to leverage these tools and present findings + insights. Net/net it may reduce the number of junior analysts anyone needs?
I think you’re crazy if you think LLMs will be able to update complex models in 5 years… they can barely do anything as is in modeling.
You’d need a huge set of data to train on
I'm with commenter above, but I think people get caught up in the replacement fear and not in the volume opportunity, everyone can just do more work with the help of these tools. We will be in a transition period for a long time where they are still tools to use. You can just cover more names at once if your day to day gets faster
100% to both of you — I see this as a huge net benefit and enable working with more volumes since data management should pretty much be taken care of. Quite frankly, it may also allow you to explore new statistics/analysis methods never thought possible.
Buy-side will never be the same — this could be the equivalent of the digital stock exchange taking over literal trading pits.
100% to both of you — I see this as a huge net benefit and enable working with more volumes since data management should pretty much be taken care of. Quite frankly, it may also allow you to explore new statistics/analysis methods never thought possible.
Buy-side will never be the same — this could be the equivalent of the digital stock exchange taking over literal trading pits.
Eligendi excepturi sint tempora possimus. Est quae dolorum illum consequatur minus similique. Nostrum consequatur nesciunt ut amet eos quasi. Sunt dolor provident rerum molestiae quae. Voluptas ratione esse vitae modi delectus quia.
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...