12 COMMON BUY-SIDE MODELING APPROACHES (that you won't see in sell-side models): Why are buy-siders so obsessed with their models? The financial model really is the backbone of buy-side investing - it shows you business momentum, NPV of incremental changes, R/R potential, and..
...critically, revision potential vs. street (among so many other things). Hedge funds LOVE EPS upside! I asked a pod PM who guest spoke at my class one day whether he would ever own a stock that he didn't expect had EPS upside? "Never" - was his answer. So how do buyside...
..analysts model to assess EPS upside potential? MULTI-APPROACH REVENUE BUILD: My first step in building a model is always to assess the entirety of what is disclosed data ways (in Ks/Qs, press releases & investor decks). Dump all of this into your model. Forecast revenue...
...in many different ways. Forecast by geography (i.e. Asia will grow faster than Europe, most of the time). Forecast by a bottom up product build. Forecast by organic growth w/ an estimate of inorganic growth. Back into or estimate what you don't have. Does company report...
...units and revenue, but not price? Well, divide revenue by units to impute price, then forecast by units and price prospectively. Be creative! YOUR MODEL IS USUALLY ONLY AS GOOD AS YOUR REVENUE FORECAST. So Step 1 is get revenue correct, Step 2 figure out how that rev flows...
...through the P&L. So in my models, often you will see 2-3 different guesses on forward revenue based on varying approaches, then I will weigh which is likely to be most accurate.
REVENUE DATA CORRELATIONS: Again peak to the primacy of revenue. What data has been correlated to the revenue of your company in the past, and what can give you real-time views on that? Alternative data is clearly it's own tweet threat. But there is plenty of open access data
..that can help you forecast accurately. BLS data can be helpful in assessing business momentum shifts. When i traded gun stocks back in the day, background checks and google trends ("buy a gun") were critical. Drop that data INTO your model, assess historical correlations, and..
..use that real-time data to derive more accurate revenue correlations. REVENUE COMP & SEASONALITY: I'm consistently shocked by how easy or tough comps trip people up. Under revenue, add a % 2 year (this years Q / 2 years ago, as a growth rate) and % 3 year growth figure...
..this normalizes for tough/easy comps and is helpful in considering the true underlying momentum of the biz. I will also look at 5-10 year average seasonality, i.e. what has been Q3 to Q4 sequential growth in the past, and what do i forecast this year? Is it out of line?
DETAILED COST BUILD: Lots of sell-side will do a simple revenue forecast then slap on a % margin, or model by bps. This approach won't consistently get you to the right EPS answer. Think critically about each item - which is variable, which is fixed? For example, is mgmt..
...committed to hold R&D at 5% of sales? Well, model it that way! Is SG&A mostly fixed, with a 3-5% annual inflation factor? Model it that way. If you model the cost structure in a real-world fashion, as your revenue assumptions change, the revenue will automatically flow through
...to EPS in a more accurate fashion. A detailed approach to each cost item will also help you understand the financial situation of the company. For example, DKNG in '21 did $1.3bn of revenue, $500m of Gross Profit and $981 of sales & marketing for the year. Looking at this...
...broken down, you can quickly assess that DKNG is in a very serious dilemma unless then can meaningfully improve GM's or meaningfully reduce sales & marketing. Rather than being enamored by the secular growth in gaming, this modeling approach can sharpen your focus around what.
...DKNG really needs to focus on to drive positive FCF. (OTHER 8 APPROACHES THIS AFTERNOON)