How much modeling do experienced HF analysts/PMs actually do?

Bump

Credit PM here so a little different. We don't really rely on street models for anything, not there is a ton of overlap between publishing and hy/distressed credits anyway. Over the years my time spent modeling has declined and my analysts take care of most of it, but there are a handful of situations I still own models for that either 1) pre date analyst hires since I don't really like handing off risk unless someone actually left, 2) are ideas i had over the weekend while shitting or something and just banged out the work, or 3) more involved modeling stuff in restructurings that have less to do with earnings and more like new debt / equity considerations we get and mapping that our across funds which is really more an ops exercise 

Bump

People need to stop fucking obsessing over 'modeling'.  Apologies for the language but it's literally just a small part of the overall job yet gets talked about constantly as you're imagining HF work to be non-stop Excel work, which is literally basic enough that college freshman can do it.

This is akin to asking how much time do you spend on the phone?  How much time do you spend researching companies?

Point is, it's a small piece of a large set of skills you need for this job and the amount of time one spends on anything depends on the firm.  Responses here will literally give you insight into nothing because you'll get absolutely none to almost every day.

And I guarantee the latter is performing worse.

I do a lot of it? Like at the end of the day a stock price is earnings times multiple. Sure you can have a thought directionally that revisions are going up just through qualitative views without a model, but when do you know that the revision you’re expecting is priced in or not? Plenty of times stocks run a bunch into a print, someone without a model might say “oh the stock has pre traded my revision” but then the revision is bigger than you think and you miss the juicy part of the catalyst. The model is what keeps you grounded in whether something has gone too far or still has room to run. 

A lot. Not sure what this question is trying to get at but if I have a view or some information about a company I want to sensitise it to my model so I can see how it filters through into profits. 

Cover biotech so trading decision are less model dependent than in other sectors but I'm mostly thoroughly reviewing the assumptions in sell side models and adapting them as I see fit.  I use the modeling to assess up/down into a given catalyst and then determine my buy/sell price targets at any given time.

I'm rarely building a model from scratch as (as you pointed out) its a huge time sink.  I don't think this would be the case if I covered another sector where valuations were more "DCF grounded."

I would take a “seat weighted average” here of people’s opinions if you are looking for a job. The jobs that hire the most will require the most modeling from young people. It is among the most important skills for getting hired by firms doing a lot of hiring, which are primarily multimanager firms that do model in a lot of detail. This is not a universal answer but I do think it is fairly clear this is crucial as a young person. 

Is this sort of dumb? Maybe. I mean modeling is in some sense commoditized. You can hire an Indian or more recently an SaaS artificial Indian (AI) to do this via canalyst. But at the end of the day people want a lot of control over this since it’s essentially the key work artifact for most roles. When you get in the weeds and develop experience you’ll naturally learn that the details do matter and there’s definitely a better and worse way to do analysis but without being in the weeds yourself you will not see this. This is far from a universal opinion but it IS a general opinion at most places that are hiring. To the extent it’s not, those people that disagree also have less need to hire you to begin with. 

imo its truly a mix dependent on your style. ive worked with analysts who have very detailed models with a ton of drivers, full balance sheet, tons of detail. ive also worked with analysts who just have the very basics, which you learn over time what is important and what is a waste of time. some people like to rely on sell side more, some dont look at it at all until they have done a good amount of work to not bias their view. but the majority of my day is not spent modeling, its listening to calls, reading research, talking to experts, talking to my PM, writing emails, etc... what is more important than the actual model output itself is the thought that went into the model and the insights you make from it that defend your thesis. 

building your own models consumes time up front but you will be 500% quicker updating your own model, maintaining it and linking it up with other aspects of your process as opposed to stumbling through somebody else's model every time you open it up

honest opinion outside of work is modelling is useless for alpha generation. but you need modelling to validate yourself and to justify getting paid in front of your PMs

I always thought reasoning and industry knowledge are more important than the modeling aspect of the job, not to downplay the value of it. Modeling has its role but not critical to alpha generation especially in today's day and age.

1. We build our own models. Sell-side models are literally 500 rows with so many useless line items and impossible to even track. We build a full Income Statement but build abbreviated Cash Flow and Balance Sheet line items. There are maybe 2-3 line items (at most) in the model and probably spend most of my time thinking about those drivers. Everything else is just noise and your job as an Analyst is to focus on the most important drivers and forget everything else 
2. It only takes maybe 2-3 days to build a model from scratch for a new idea. If I'm exploring new ideas, I will do back-of-the-envelope math and run it by the PM before deciding if it is worth building out a detailed model
3. The model is really a tool to hold myself accountable when the company reports so I do the work and not take shortcuts. I spend most of my time reading, updating the team with findings, etc. vs building models
4. If I inherited a messy model, I would rebuild it myself. You should standardize your model template so it is easy to update quarterly
5. Models are also our "memos." If I ever wrote a word memo longer than 2 pages my firm would take me behind the dumpster and shoot me for wasting their time. Models are pretty straight forward and should portray whatever your qualitative thesis is

It’s a key skill for landing the job though how much modeling matters day to day depends entirely on the PM. Some barely use it and others require it for nearly everything.

Build it. Otherwise how do you have any idea what your delta vs consensus is...whether you look out a quarter or 18-24mo, otherwises you are just putting a thumb in the air

Also w/ the model if keeps you honest as a guardrail. Why are you so bulled up on a stock if your nbs can't support it? And vice versa. Forces you to spend more time looking at the setup a clean way, and frankly forces you to update your views in a way that drives into math for what a stock is worth

Build your own models. Takes longer, but sell side models are useless with way too many line items and often broken links. Takes me usually half a day to a day to build a model. I used to do 3 statements, realize that was a huge waste of time. Just build income statement and revenue drivers. Keep it simple stupid -- that's the guiding principle

I think I have a model built for all but 1 company in my coverage. It's enormously also helpful when stack ranking opportunities. You may want your PM to add to 3 opps, but he's gonna come back and say there's room for only 1. How do you pick? Narratives and vibes? That's where the model comes in. Force a reassessment of your thesis and translate that into the model, go from there

Depends. When I know a company cold I just look for drivers / bridges to the main numbers that will make a difference. RPO, revenue growth, margins/costs, something event driven. Usually not all of these move the needle (especially outside mega caps), so the goal is to have a good view on a few of the main metrics.

For example, for a certain company RPO may be all that really moves the needle and where there is risk or upside. In that event I will just seek to use alt data, channel checks, expert calls, calls with private businesses, conduct surveys, looking at / speaking to public competitors/value chain parties that may report off cycle or in a foreign country etc to see how that trends.

Sometimes though several numbers really matter and move the needle. coreweave is maybe the prime example off the top of my head because need to have view on backlog, top line, KPIs related to power, pp&e and depreciation and capex split into WIP vs depreciable, detailed debt schedules and interest costs, cash burn, etc. 

finally the point of a model isn’t always precision. It’s to better understand drivers of variance vs historical performance, guidance, and consensus. Building a model for a new company hopefully makes you sharp on it, and lets you figure out what questions to ask all the relevant people on calls/conferences. The model is more useful as a tool to help you build a framework on a company, than it is a decisive trading tool.

The more I do this job, the more useless a model feels from my seat (concentrated long-only). It clearly matters a lot for a MM pod, but my experience has been the opposite. At a SM, you’ll never have enough resources to beat the pods at calling quarters and building the most accurate model. I think the only way to have any reasonable edge is to buy something out of favor and be willing to hold for a longer period of time until flows start coming back or have so much conviction in a company / theme that you size up the position. That can come from high level modeling or just fundamental research (ie if I build conviction through 100 expert calls or modeling line by line SG&A so be it)

An interesting thing to consider is the Kenny G & the sellside model usage discovery...pretty illustrative on many levels and appropriately so IMHO