Learn the Trees First, Then You Can See Which Way the Wind Blows — Notes on Statementdog on How an Institutional Analyst Researches an Industry

Contents

1. How does someone who never studied medicine come to understand a whole healthcare industry?
2. Every sub-sector plays by different rules — how do you find the key point?
3. If you know an industry inside out, can you beat the index?
4. Where does AI get stuck in healthcare?
5. So how do I use any of this myself?
6. Where to look yourself
7. The One Thing to Take Away

All things being both caused and causing… I hold it impossible to know the parts without knowing the whole, or to know the whole without knowing the parts.

—— Pascal, Pensées (1670; translated here by the author)

Statementdog episode 574, released on 1 November 2026, features Vincent, former executive director in asset management at JPMorgan and co-founder of Manbao Pro. He says that after twenty years of looking at companies he can tell a good business model from a bad one, but when it comes to when the wind picks up — when the tide comes in — he usually finds out after the fact, or only once it has already arrived and he has stepped back to look at the whole. He lays out the framework he teaches: company value is compounding, (1+R) to the power of N, where R is the return on invested capital, N is how long that return holds up, plus how management allocates capital. He also draws his own boundary around it — the framework gets you to a share price, and the excess return after that comes from how your understanding of the industry maps onto that price.

How does someone who never studied medicine come to understand a whole healthcare industry?

Short answer: find the one storyline the whole industry shares in that period, work along it, and pick up the technical knowledge later.

Vincent was assigned to healthcare. He started as an intern at a hedge fund in his third year of university; his boss had come out of a US fund, spent time at SAC (today’s Point72), and returned to Asia to cover the technology supply chain. In that era the money came from an information gap — mailing DigiTimes to the US was enough. He went from passive components (the small capacitors and resistors inside phones and cars) to Chinese property, education, cement and consumption upgrade. After he joined JPMorgan the firm wanted to turn analysts into sector specialists. Healthcare was hot at the time and nobody wanted it: learning a new industry is tiring, and then having to talk to portfolio managers about it is more tiring. He says he does not think he is much smarter than anyone else, but he has an edge in how fast he learns, and he was willing, so he took it.

Chinese healthcare at the time had a storyline sitting right there: import substitution — domestic products displacing imported ones. The method was as plain as a checklist. List every company, see whether its product is cheaper than the foreign one and whether the quality has closed the gap — his bar was “it doesn’t have to be better, as good is enough” — then go visit hospitals and ask how big the remaining gap feels. As long as the gap is narrowing, the story holds. Financing was not an issue, the government would back them, and hospital tenders would generally go through. His example was Shandong Weigao, a small company making drainage bags, needles and tubing, up against international players like Baxter.

Working that storyline required no pharmacology at all. The order was: first a criterion you can actually execute, then the expertise grows out of it.

Every sub-sector plays by different rules — how do you find the key point?

Short answer: write down the rules of the game first, then ask what decision the leader got right back then.

Healthcare has the advantage that the rules are written down. A new drug goes through phase one, two and three, needs approval, manufacturing carries its own requirements, and a biologic gets its plant inspected. Lots of detail, but every step is knowable. His order is: lock down the rules, then look at who is already the leader and what they got right — picking the right product, or playing differently from everyone else. Then look at the cycle: when a patent expires, generics arrive, so a new product has to be found to carry the growth. The competitive landscape keeps shifting, and the shifts are where the opportunities are.

His course at National Taiwan University is built that way, one sub-sector a week: drugs, biotech, vaccines, cell and gene therapy, devices, equipment, hospitals. Each has a different business model, each gets a representative company, and the discussion covers why that share price did well and why it later did not.

The most interesting part is the companies that resist classification. Yunnan Baiyao went to the consumer analyst at JPMorgan because it mostly sells toothpaste; Beijing Tong Ren Tang stayed with him. He once worked through Ejiao: a supplement plus an investment good, with a price increase of 25% a year, the same story as Moutai, which is why people who bought Ejiao often bought Moutai too. The product is supposed to be cooked down and consumed within a year or two, yet plenty of distributors and buyers stockpiled it and resold it. He put the stockpiling share of demand at roughly 30%, with 70 to 80% actually consumed. He says products with an investment attribute all land around that ratio.

He also unpacked why supplements are hard to make money on. A drug can run an 80% gross margin with selling costs around 30%, and stacking drug after drug spreads that cost, giving you operating leverage. Supplement customers are scattered, marketing runs at 40% or more, and adding products does not push it down, because every one of them needs channel incentives and advertising. No approval risk, and a path to profit that looks nothing like a drug’s.

If you know an industry inside out, can you beat the index?

Short answer: he says it is hard, and the reason sits in the industry structure rather than in how good your research is.

Some people can invest in technology alone; biotech cannot work that way. Biotech stocks are tied tightly to funding conditions. A product goes from idea to market through several rounds of financing, and one bad market in the middle ends it. Patents have a clock, so a two or three year delay lets someone else catch up and your numbers change entirely. Whether biotech survives correlates with market liquidity — good market, easy financing, higher odds of staying alive.

The other layer is how much of it is investable. Healthcare’s share of GDP rises with demographics, but much of the increase sits outside the stock market, in hospitals and health services. He says adding up the market caps of the big pharma names draws a relatively flat line that moves up and down, which looks different from the way the whole technology sector curves up and to the right.

He also offered a decomposition: return splits into stock-specific, sector, and market. An analyst owns the first two — add in the sector when it is good, buy less when it is not, and pick the better names within it. A portfolio manager owns all three. The equity-versus-cash split goes back to the investor.

One of the hosts added a line I want to keep: the familiar version is that you find out who was swimming naked when the tide goes out, but when the tide is coming in you do not even know whether you have shorts on, so you still have to get on the boat.

Where does AI get stuck in healthcare?

Short answer: between handing you more options and picking the right one.

Vincent says everyone understands AI can hand you plenty of drug candidates, or help resynthesise. But you still go through every stage of the process and only one succeeds in the end, and whether AI can find that one has not been demonstrated. He draws the parallel with research reports: you can have AI write a report on every listed company in Taiwan, and you still do not know which one makes money. The difference is feedback speed — a report answers immediately, while a drug takes four or five years from phase one and costs hundreds of millions of dollars.

He says the one running furthest ahead is a Hong Kong-listed AI drug discovery company with a compound in late-stage trials; the market is watching whether it can push into phase three and win approval, or license straight to big pharma. One or two cases coming through would change confidence in the whole area. Until then these companies are mostly selling shovels, and selling shovels requires somebody to spend money first proving there is gold. The hosts put it more bluntly: the shovel seller has not struck gold either, and still tells you to go dig.

The wearables section is the one I found most useful. People want AI to infer the things you cannot measure from the few you can, but claiming you can predict means clearing the FDA — clinical trials, consistency, variance kept low. So most wearables can only say “I am able to”, with no clinical support behind the sentence, and they end up as consumer products. He says the real problem is that the loop never closes: once you know a risk, you need to be able to prevent it and intervene. You detect a hypertension risk — how does the physician prescribe against it? How high is that high? With those links missing, the business model does not hold.

He contrasts it with CGM, continuous glucose monitoring, a device worn on the body that reads glucose continuously, which does measure something real. Early versions needed regular calibration, newer ones only calibrate once, and current ones are calibration-free. For patch or optical approaches, that first reference point still means a finger prick.

So how do I use any of this myself?

Start with the pain. You finish a company’s earnings call, it talks about next quarter’s new products and how the pipeline has been adjusted, every company talks that way, and then you notice you have ten sets of notes and cannot say where the industry is heading.

Vincent admits this is his own situation. He points to Chinese healthcare: regulatory standards were raised and the poor-quality players were cleared out. Talk to a company and all you get is next quarter’s pipeline and which new products are launching. Only after talking to many of them and thinking it over did he notice they were all saying the same thing — the competitive landscape had improved, and the days when anyone could walk in were over. That was the moment to move the valuation. He says it goes like this every time: you see the trees and miss the forest, and while you are watching how handsome one tree looks, you do not notice the wind outside has changed direction.

So the method has two halves. The first is hearing enough detail, and there is no shortcut there. The second is stepping back deliberately, putting many sets of notes side by side, and looking for the phrasing everyone uses and nobody flags as the point. Stepping back is something you have to schedule; it will not happen on its own while you read note number ten.

The second half connects to the framework he teaches. Company value compounds: (1+R) to the power of N. R is the return on invested capital, N is how long that return holds up, plus how management allocates capital — dividends, buybacks, capex, acquisitions — with every one of those moves feeding back into R and N. He points out that R and N often trade against each other: earn a lot in the near term and N may shorten, because competition and opportunity cost follow. All three are dynamic, so what the framework gives you is a set of questions to answer, with the answers still yours to fill in. He also says it is one tool in a toolbox, and dropping it in does not make you invincible.

Last is his view on entering the field, and this part carries outside investing. He calls investing an apprenticeship industry — the first boss you work for, the first person you learn from, largely sets your style afterwards. The hedge fund where he started ran a little over 300 million dollars and later reached about a billion, so he rode a rocket. But if he were choosing today, he says he would first look at how strong the team is, what the style looks like (how much leverage, how concentrated), and whether a good year actually gets shared. Failing that, he would go to a brand name. The price of a brand name is having to convince yourself why you are staying during the good years, and envy runs strong in this business — every time someone leaves it is another multi-billion fund — so you have to make peace with yourself.

He added a time marker too: Taiwan has been at an inflection point these past few years, and an inflection point does not stay one. If he had hit this environment when he was young, he says, the first thing he would think about is this — suppose it holds for two or three years and then goes away, do I have the ability to stand on my own, and is my money enough.

Where to look yourself

• Statementdog episode 574, “An institutional analyst’s method for industry research, ft. Manbao Pro co-founder Vincent”, 1 November 2026
• The US FDA’s classification guidance for digital health and wearables, for when a measurement feature needs a clinical pathway
• Recent earnings calls and investor decks from Eli Lilly and Novo Nordisk, against how metabolic disease portfolios get framed
• A decade of drug review reform documents from China’s NMPA, matching the “standards raised, landscape improved” point in the episode
• To practise the R and N split, start from any company’s investor deck and annual report: find return on invested capital, then ask what is holding its durability up

The One Thing to Take Away

One idea: once you have heard enough detail, you have to schedule a separate act of stepping back, because it does not happen by itself. Vincent talked to dozens of companies and each one only discussed its own next-quarter pipeline. The conclusion that the competitive landscape had improved was in none of those meetings; it grew once he put all of them side by side. Collecting and generalising use two different brains, and the second one needs a slot in your calendar.

Here is something I have tried, and it works outside work too. Pick one thing you have been dealing with intensively — a job change, a family member’s illness, your child’s situation, some question you have been chasing for half a year — and write down every claim you have heard about it, one per line. Copy first, do not analyse. Leave it a day, then do one thing: find the sentence that showed up more than three times and got treated as background every time. I have done this a few times, and what turns up often has nothing to do with the storyline I thought I was in, and it is usually where the thing is actually going.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.