What does a model look like at a L/S hedge fund? What is the diligence process like?
Can someone please describe in painstaking, excruciating detail what modeling looks like at a top L/S SM? I am in PE right now and we are only able to make these absurdly detailed models because we get access to every single piece of information down to the G/L level and are able to create models with dozens of spreadsheets and call dozens if not 50+ customers to project account level details in many cases.
I have a hard time understanding how it is possible in a HF role to create models with builds behind even business unit revenue level line items without having more detailed information than is typically presented in 10-Ks or earnings calls. How is this possible? How do you build these models and what is the diligence process? I simply do not understand one bit.
This is a good question for Cathie Wood (Cathie: "where my dart board at?")
Look OP here I am going to continue my rant and say I was trawling through some tech 10-Ks and there's just nothing to go from. Nothing. And to know that the average L/S does not hire market consultants or FDD guys also blows my mind. Who figures these things out for you? PE in a nutshell is, get the workbooks from sellers and send to your Big4 of choice to get them to nail down all the nitty gritty accounting stuff (not really relevant for HF I guess?), get market work done by Bain (why does nobody do this in concentrated L/S, this is so crucial), and get your Big4 of choice to do all the cuts of financial data that you want (BUT WHERE DO YOU GET THIS DATA IN THE PUBLIC MARKETS?). I'm not talking about alt data, I'm talking about actual company financial information. What if I want to take my own build of Net and Gross retention but I can't cut it how I'd like because I don't have the underlying data? What if, instead of "Revenue Line A" I actually want to build it up from units * price and get that information? What if I want to toggle # of sales reps because I want to jockey management to hire more reps and can show the ROI on that would be accretive? None of this is possible without comprehensive information, yet still L/S funds SOMEHOW have to model. HOW can this be possible? What is going on inside these funds? Surely they are not just throwing a dart at a market and riding tech beta all the way home?!
Because modeling is pretty stupid/pointless and the only reason you do all that is to justify to lps the fees you charge. Suprised you think so highly of the pe diligence process, everyone knows its a joke.
Hf guys can simply just trade off "tips" and the actual underling financials don't matter...only the direction of the stock.
Every PE analyst / associate has had the experience of building a hugely detailed model at maximum granularity building revenue, COGS, etc by geography & product. Principal / partner looks at the output and says the following:
The idea that "the model" helps you make good investments and have greater certainty in their outcome is a fallacy. PE alpha is driven by leverage, illiquidity, a bit of smart sector allocation and a lot of spotting frauds / pumped-up numbers that don't get caught in today's credulous public markets.
Will being able to model out net and gross retention to a tee for the next few quarters REALLY change your thesis on the company? If ROI from marginal sales reps is 1.67 instead of 1.72, does it even matter? Will building out a hyper-granular AR collection schedule move the needle on whether you invest or not?
It turns out most companies' long term success can be boiled down to a few key factors. That's the beautiful thing about public markets: all you have to do is have a different view on one aspect of a company than the market. You don't have to care about anything else; if the market thinks net retention will 110% and you think it will be 115% and you're right then you will outperform over time.
Inclined to agree. Whenever I have built >5 tab models looking to project individual accounts / similar level of detail, I have sought company's guidance and a bunch of other stuff mentioned in your comment. The thing is, for anyone looking to own it in the same manner as a PE and maintain it as a PortCo, they have the access to build these. Even then, I feel the need to build this to this level of detail is optional / very case dependent.
The PE modelling process is just an extension of the 10-20 tab sell-side operating models: often longer monitoring time frames on PortCos, more granular cuts (instead of forecasting just by SKU based on mgmt estimates you may even call up customers who buy the goods and try and balance the supply with their demand pipelines etc) and more inter-dependence between advisors. You don't often get the CDD advisors' data feed into the sell-side model in IB, barring directional alignment, but there may be more overlap while in PE.
To be honest, not the biggest fan of PE models. At that point modelling is far removed from "valuation" in the sense of "structuring" an efficient transaction (i.e., a la HF looking for some creative bet) and much more what a F100 FP&A person may do when preparing their budget after engaging some advisors or by extending the remit of their auditors (give or take).
It's also funny how life comes full circle: top grads flock to IB shunning FP&A / CFO track roles in corporates so that they can exit to PE. Once in PE they get to work on basically forecasting the general ledger for PortCos or to be PortCos. I digress.
Knowing that this is the kind of brain that some PEs are working with makes me glad that I decided to go into public market investing...
I've done private markets in the past and know all the reports you are talking about. What do you think PE diligence is trying to accomplish with all those fancy 3rd party studies?
It is trying to guess what EBITDA is and what multiple it should trade at... exactly the same thing public investors are doing through channel checks, alt data, and, yes, good-old fashioned SEC filings review. Of course, there are a handful of true operational PE shops that occasionally makes something out of nothing (i.e. KPS); but that is the work of operating advisors who are ex-operators, not you as the PE deal monkey.
First off, PE advisors tell you what they already know you are inclined to believe - when is the last time Bain or KPMG told your partner not to do the deal already deep in a process? Second, how does having historical data on 100 variable inputs make you a better forecaster of future outputs? Third, all of the other sophisticated bidders have access to the same information - how do you gain any variant perspective working with the exact same data that you KNOW your competitors are also reviewing?
The real reason you do these studies is because PE deals move on a tight timeframe and thus you must "create" expertise under that deadline that you don't currently have. Public investors can diligence a name for years before they pull the trigger. They are also done to help your partners feel safer and cover their asses in case something goes wrong and "pitch" it to the committee.
Do you wonder how Warren Buffett or John Malone ever did deals on the back of a napkin without reviewing every SKU going back 20 years divided by geo or employee packages line by line? Or do you think Jeff Bezos was able to run Amazon only because he knows how many forklifts (and which models) were in each of his warehouses (he doesn't...)?
This post is such a classic PE false precision superiority. A better model does not make a better investment. The companies that provide the most information/line items are often the worst investments.
Riding tech beta? PE is all about timing the macro cycle with leverage. It’s way higher beta than L/S. A few 8x EBITDA entries and 15x EBITDA exits with leverage can make a PE fund and that is not skill with a 2000 line model
There is no superiority. I just don't understand how what you describe occurs. How do you construct the model on that opportunity. You need to identify the drivers of value. Constructing the model more than anything is a tool to understand what elements of the business will drive value (e.g., by playing with the drivers you can see what business changes lead to what returns results). Let's say we are talking about a company that does not provide the most information line items. What are you modeling? What are the things you have to model to get comfortable with the investment? Without those provided by IR, how do you get those?
Here is an example - take a drilling rig company. You can see the history of how many rigs operate in the US per month and well as day rates. You can track a company's market share over that period as well. You know the rough cost to operate a rig, the maintenance capex, and SG&A. From there, a very simple model/math will get you where you need to be. It can be as easy as I have 50 rights, getting 25k a day in revenue, with 10k a day in cost and 75mm in SG&A and 1mm a rig per year in maintenance capex. So know you know that 1 rig will get ~9mm in revenue, with ~3.6mm in opex, 1mm in maintenance capex thus 4.4mm in cash flow per year. A company might own 150 rigs and have 50 operating. So you can easily calculate breakeven and incremental cash flow with very simple numbers. Projections can be based on overall market expectations etc. but the point is keeping it relatively simple. From there you can play around with subtle changes to rig count or day rates but you don't need to get too complex. Then you layer in mgmt comments, industry info, what is the competitive advantage of the company (location, IP, brand, mkt share, etc.) things more qualitative in nature to get a view on where it might go.
In contrast, a PE guy might model out the cash flow of every single rig, run sensitivities based on every specific rig contract, factor in wages of every rig worker comp package, try to do something fancy with basin level projections, have the SG&A be extremely detailed, blah blah blah. But how much more are you learning by doing that and what is the likelihood of that being more impactful than just multiplying 50x15kx360 - 75mm, and changing the 50 to 75. You're not really going to learn that much and getting too detailed is kinda a complete swag anyway. Granted a drilling rig company is a more simple business model but the same concept can be carried over to even more complex industries/companies. Not to mention what's the point of having 10+ variables, go forbid something like 50+ variables - all moving in different directions that is entirely outside of your control.
The real question you should be asking is “does a model need all this detail” and the answer is no it doesn’t. Models are a directional tool, not a crystal ball.
Classic pe nerds here.
"How does anyone make money in LS if i cant build a 24 tab model to the ledger level detail and have bain make a 150 page market study to read!"
I honestly think PE is just as pathetic/unimaginative than IB, if not worse. They just think because they call themselves "investors " they somehow arent.
In my opinion models are mainly meant to see the downside of the investment along with seeing how the company views it's growth prospects/financials and those general aspects of a company . The real alpha is generated by talking to management, channel checks, talking to customers, viewing the macro enviroment/industry tailwinds, capital allocation efficiency, alt data, and the rest of the gauntlet. After doing this, you have to make a judgement call with whether the prospects of the company are better than the consensus. Modeling is just doing the due diligence that the company isn't a ponzi scheme albeit some TrashCos have fallen through the cracks.
#1- private equity models are stupid. Way too much granular detail to be useful.
#2- HF models are indeed based off of 10k/q plus earnings reports and calls. Intellectually, they’re a bit lacking but for any given stock only like 3 KPIs matter… so you barely need a model anyway. Any analyst worth their weight in salt can tell you the eps impact based on a few metrics. If I told you the top line miss and bottom line beat, most decent HF monkeys could within 5 minutes tell you everything in between and the likely reasons — without running a model.
In reality, if I were starting/running my own company and wanted a financial analyst to build out a model to understand in a bit more detail, it would be more detail than a HF model (basically just meant for a Michael Scott-type audience) and less detailed than a PE model which is intellectual masterbation.
1. If you cover a given sector/company, it’s very obvious to you. If it’s not very obvious to you, you’re incompetent. For proctor and gamble it’s top line organic growth. For Adobe it second and third derivative revenue growth, i.e q/q accel/decel in growth, margin by which they beat guidance/consensus growth estimate. For Chipotle it’s same store comp growth, accel/decel, and 4-wall margin. For a bank it’s roa/roe, asset growth, and loan loss trends. For an HMO it’s utilization and reimbursement rates and so on…
2. how do you develop a view? That’s the entire game everyone is playing. But the obvious way to accurately get a view isn’t really a ‘view’. A real view is a differentiated take, and reading the WSJ or getting some dog shit commoditized credit card data from Yippit isn’t it either. It’s accumulating disparate datapoints both quantitative and qualitative, and using that often conflicting body of information to assemble a reasonable expectation. Keep in mind the most likely outcome is already priced in. You need to know what the most likely outcome is, so you know what’s priced in. Then you need to try to discover when that most likely outcome is unlikely to happen (which is inherently a divergent view), whether it will be better/worse, and how the stock will react. That’s basically the whole game. Its understanding what people think and why, if they’re likely to be wrong, and how they’ll react when they find out. Within that framework how valuable do you think a model is? Asked an other way, how differentiated is your 6th grade arithmetic?
You do realize they just ask the sell side for the models
The real question you should be asking is "does a model need all this detail" and the answer is no it doesn't. Models are a directional tool, not a crystal ball.
Pizz did u just blatantly copy paste another person's work and pass it off as your own? You must actually work at a hedge fund I'm impressed
Hedge fund models are quite different from P/E models. In P/E, you have granularity because you have access to opco as the buyer. In the HF space, that would be MNPI. Therefore, hedge fund models need to incorporate all known data (10-k/q/investor presentations) but cannot exceed that level of granularity on a per company basis due to the aforementioned restraint. Whereas PE analysis is like solitaire, HF analysis is like poker.
Generally, you'll start with a sell-side model that is a fully fleshed 3-statement model + all schedules. You will isolate critical factors and use your firms' resources to develop an edge on them. For example, say you are analyzing an industrial company that manufacturers and ships components used for cameras. The critical factors here may be 1) firms desire to expand is decreasing operating margins through an increase in SG&A, 2) new orders for buyers (say Apple and Samsung), and 3) input costs.
You develop an edge by:
1) scrapping job posting info and through surveys with HR specialists within that space to see if employment is ramping up faster/slower than expected
2) new orders are increasing by reviewing downstream manufacturing activity from Samsung/Apple (e.g. Foxconn capacity levels)
3) input costs through satellite images on primary sector activity (actual mining of the raw material that will feed into the component) with observations on each bottleneck in the supply chain (refinery activity, transportation, etc.)
You use this information to develop a mosaic view of the company -this gives you incremental insight vs other HF or public equity investors. Remember -MNPI is "against the rules" so you are assuming most participants are basing their buy/sale decisions on the information that exists on the sell side model (all known press releases, filings, etc.). Your additional insights from scrapping, surveys, imagery, etc., gives you an edge from a probability standpoint.
Most MMs are looking to trade the quarter, so the view is primarily driven by whether EPS or a KPI will beat or miss- not whether the stock is going to go up or down. The PM will determine the timing (have a view on the multiple and price action).
To summarize, a HF analyst is primarily focused on whether KPIs will beat or miss in the quarter and year but is limited by MNPI. The longest models I have seen were ~10 sheets or so (cover, one for each statement, 2-3 schedules, ratios, valuations, earnings tracker vs. street & guidance, etc.). That information is all required to establish a view on a stock, but MOST of that (statements, ratios, schedules) will come from a sell side model, so you can focus on critical factors.
Hypothetically speaking -- the edge is derived by developing insights on information that the sell side has not analyzed, and ideally, that most other buy sider analysts have not figured out/done yet/know of.
For example, perhaps everyone already uses glassdoor and indeed data to develop insights on hiring...but you may have a contact who works at a HR company that specializes in recruiting industrial engineers and salespeople. That person gives you color and what they are seeing with hiring trends (ex-MNPI!!!). If you talk to this person weekly, and last week they went from "hiring is continuing at a strong pace" to "hiring dropped off, no one wants to move forward...budgets are being halted", you can quickly look at information across the supply chain and your rough order book analysis to determine what's driving this, and if it'll be material to your company's earnings. A halt in hiring could be either 1) a sign that a company has reached its hiring objectives or 2) that something really bad happened, or something else completely. But the better you get at having sources of edge (mainly: PEOPLE IN OTHER INDUSTRIES WHO CAN GIVE YOU TIDBITS OF WHAT's GOING ON) the faster you will be at taking new information and applying it to your model to see the impact on earnings.
Typos sorry 3 bottles of wine in why am I on WSO