MM HF: non-consensus on 1st level vs 2nd level
Hi all. I have a case with a MM HF (C / Baly / MLP / P72) and am a little confused on the precise nature of how one "disagrees" with consensus in a stock pitch, both on the case and in the job. Do you:
1a.) Take consensus cash flows, value them, and form an opinion on whether the market value is fair or not (kind of a level 1 approach)
1b.) Make up your own cash flows with little reference to consensus, value them, and form an opinion on whether the market value is fair or not (also a somewhat level 1 approach)
2.) Take the key drivers of consensus cash flows (e.g.: SMB penetration for an IT Services company) and come up with your own thesis on whether it's likely to be better / worse than the sell-side opinion (based on industry conversations, alt data, convos w. mgmt etc.). Then once you have your guess on the likelihood of bull/base/bear, run it through a valuation model (which should reproduce market price in a 30/40/30% distribution - i.e.: you assume the market is "right" about valuation, but potentially wrong about a key driver) and from there see what the fair value is. You then know the exact catalyst of the thesis - whenever the driver is proven to be right / wrong.
I suspect it's 2 (assume the market is right about price in the base case but maybe wrong about likelihood / magnitude of drivers), but idk how much of 1a or 1b to mix in. Is it fair for one to mix in a little more of the SM 1a/1b approach (I like the stock, think it looks cheap, and there are no specific catalysts other than continued strong prints)? Also, does this balance between approaches differ between a case and the job? On the job you know the names a lot better, and presumably have better access to the data that makes forming non-consensus numerical views on drivers much easier. TIA!
Your job is not to disagree with consensus lmfao
Fair enough, but presumably one does need to occasionally hold non-consensus views to find alpha?
Why 30/40/30, not 20/40/40 or 35/35/30 or 40/30/30?
Just an example - thought it should be something somewhat normally distributed (25/50/25% should work too)
Got a little confused at some of this but I'll take a stab - also interested to see what others may chime in with. My advice is don't get too heady with the investment philosophy of it all / fitting it to an exact investment style or structure of what you believe it needs to be like. The job is to monetize a stock that is under your coverage as best as possible.
1a) not sure what "Value consensus cash flows" mean. A DCF? A multiple? If so you will end up at consensus right?
1b) how do you determine if market valuation on your own cash flows is fair?
2) What is your valuation model?
Today's stock price reflects a lot of different things - intersection of macro backdrop, with sector backdrop, with recent news flow, and finally the expectations of future results all reinforcing the LT (which is what a multiple really is btw); all of these views / trajectories aggregated back into today's price
Generally, an opportunity arises from probability distribution leaning one way too heavily on that range of future outcomes - or sometimes an entire outcome is totally un-recognized (rarely true and is still that probability distribution anyways I guess).
Your job is to decompose the fundamental value pitch of the company, isolate out the key drivers around value creation / future outcomes (this is the core debate of the stock), and sensitize relative to current price implied expectations for the stock today / consensus estimates.
"Consensus" is both reflected by the range of analyst estimates (so yes sell side might be $5 in EPS or CFPS, but that is driven by xyz unit sales at xyz margins), and the market price today (this is decomposed into a growth rate + discount rate + ROIC).
So a lot of what you are playing at, especially in MMHF world, is proving or disproving different tails on the probability distribution. So if next quarter that stock that trades at in-line multiple beats EPS by 2%, but it was because they are taking market share faster in their key product category, and it reinforces the view that they can be a durable 7% grower or whatever, that can still be a monetizable event.
So practical implementation:
- Model out the business and identify / isolate out key value drivers; your goal is to thoughtfully quantify the critical drivers here. You should incorporate how management views things, how consensus views things (basically mgmt.'s view), and where you view things with your primary research
- Compare vs. both explicit sell side numbers / mgmt. guidance, and the price implied numbers (reverse DCF, embedded in multiple, multiple historical range + peers, "normalized earnings" maybe)
- What is the reasoning for that gap - is it diligenceable, with how much conviction, etc.. Catalyst/timing also important here (this is the thesis)
- What are the range of valuation outcomes based on what you are modeling (gets more art than science here at times)
- When you assign probabilities to the different outcomes, try to use some real reasoning here, not just random numbers. Mauboussin calls it propensity and frequency of events or whatever, but you get the idea.
Thank you for this - this is very helpful. A couple responses:
1a.) Yep, fair enough
1b.) Well maybe you are projecting much lower cash flows than the market, even in the base case
2.) Anything. I had in mind a probability-weighted multiple, but DCF would work too
A couple follow-up Qs:
A: Okay, so essentially one assumes the price is fairly valued based on some distribution of outcomes, which we back-solve from the market price given our model and assumptions? We then compare this to mgmt and our own diligence to form a view on if the market-implied distribution is legit, and make a bet if it's sufficiently far from far from what we think the real distribution is? Is this a fair representation?
B: Presumably if we have the opinion that the stock isn't fairly valued, we will follow a different process based on finding fair value
C: Should one be doing the key drivers variation from mgmt/consensus & probability distributions modeling in a case, or a more straightforward valuation where we take consensus as fact?
D: Are there any resources you know of to practice this sort of thing? Other than physically doing it from 10Ks, which I am doing, but tough to tell if I'm doing it right (as I have no access to mgmt, SS reports, alt data, or worked examples like this on my personal computer)
Thanks again for your time :)
You are either a student or working in IB I am guessing (if doing a MMHF case study). As such, you have access to SS research just need to dig deeper. Every school has sell side access I think - I can still access it from my school through a platform in the library website and I graduated a long time ago. You likely have more tools at your disposal just need to get creative and dig deep. Getting consensus numbers is easy but if you need them I can send the bloomberg snapshot if you DM me.
It sounds like you are confusing the valuation stuff with the variance to future results. I had trouble following a bit, but maybe I am making things too confusing. Just a simple version...
- Model out the company. See where you fall vs. the explicit sell side numbers. Your goal is to be independent and thoughtful; SS and mgmt. still relevant
- Compare today's multiple with its historical range. Expanded? Lower? Why? (the answer is related to where market thinks numbers are heading or the premium it deserves given the business model); You may want to do a quick reverse DCF to get a better sense of the implied rates in the business at today's price. In fact at your stage it probably makes sense to as it will help for learning.
- Now sensitize both; what are the potential ranges you can model and with how much certainty. What are the valuation ranges?
2025 2026 2025 2026 Rev $10 $12.00 Price Today $250 $250 Consensus Growth 20% Multiple 25x 21x 2025 2026 Rev $10 $12.50 Price Target 25x $313 Variant Growth 25% vs. '25 consensus 31x $313Your price target ranges are future metric X current multiple (for simplicity sake). Multiple expansion occurs as company re-rates to higher growth / ROIC expectations. A 20% grower beats expectations and people start to see it as a 25% grower going forward. I did a little example below.
More so than specific price targets, you are taking the over and under on outcomes and relating them back to plausible price ranges. You can't predict the future with any accuracy, but you can weigh risk/reward and make good bets
Here is a really simple example...
If your variant metric is revenue in 2026, why do you take the 2025 multiple (25x) rather than the 2026 multiple (21x) to arrive at your target price?
Non repudiandae sint enim sequi. Placeat expedita iste blanditiis. Reprehenderit iure natus quam animi ut et omnis.
Natus sunt neque quaerat. Tempora et corrupti aperiam architecto aut nostrum. Minima aut velit ut voluptatibus. Quae eum ut alias.
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...