Essay
Every Investment Is a Hypothesis
Five questions I ask before I commit to anything
I spent the better part of a decade on experiments that did not work.
That is not a confession. That is molecular neuroscience. It is most science, honestly. You design something carefully, you run it, and the cells do not care what you were hoping for. So you do it again. The job was never to be right on the first attempt. The job was to be wrong in a way that taught you something, and to be wrong before it got expensive.
When I left research and moved into real estate and private investing, people assumed the useful thing I carried over was analysis. Comfort with numbers. Spreadsheets.
It was not. Plenty of people can build a model, and a model will tell you whatever you ask it to tell you.
What actually transferred was a habit of mind. Treat every claim as a hypothesis, including your own.
Every investment is a hypothesis about the future. It is a claim that certain things will happen, in a certain order, within a certain window, and that you will be compensated for the chance that they do not. Most people skip past the claim and go straight to the part where they are right.
Richard Feynman put the whole problem in one line: “The first principle is that you must not fool yourself, and you are the easiest person to fool.”
Here are the five questions I ask before I commit capital to anything.
- What is the hypothesis?
- What has to be true?
- What would prove me wrong?
- What else could explain this?
- What do I do when it breaks?
1. What is the hypothesis?
State it in one sentence, out loud, in plain language.
Not “this looks like a good opportunity.” That is a feeling. A hypothesis sounds more like this: this property will produce meaningfully more net income three years from now than it does today, because rents in this submarket have room to move and because the current owner has been running it loosely.
Now you have something you can test. You can go find out whether rents have room to move. You can look at the expense lines and decide whether “loosely” is real or wishful. A feeling gives you nothing to check. A claim gives you a list.
If you cannot say the sentence, you are not ready to write the check. That one discipline will talk you out of a surprising number of deals, and it costs you nothing but a minute of honesty.
2. What has to be true?
Every projection is a stack of assumptions wearing a suit.
Rent growth. Expense ratio. Timeline. Financing. Occupancy while the work is happening. What the property is worth at the end, and what interest rates look like on the day somebody else has to buy it.
Write them down as a list. Then ask which ones carry the weight. It is usually two or three, not twenty. In a lot of real estate deals the thesis rests on the exit assumption and the timeline, and most of the rest is rounding.
The reason to do this in writing is that assumptions hide inside models. A number sitting in a cell looks like a fact. It is not a fact. It is somebody’s opinion about the future, formatted.
3. What would prove me wrong?
This is the question almost nobody asks, and it is the most valuable one on the list.
In research, you answer it with a control. And a control is not there to make your result look good. A control exists to take your result away from you. You build the thing most likely to destroy your finding, and you run it on purpose, because if the finding cannot survive that, you would much rather learn it at the bench than in print.
A useful moment in my research career was the one where a control took an exciting result away from me.
I was studying a protein called UNC-13 and what it does at the synapse, the junction where one nerve cell passes its signal to the next. Delete the gene, and the signal did not pass. That was striking, because published work at the time suggested that cells missing syntaxin, the protein everyone considered essential, could still get a signal through. If that held, UNC-13 was even more fundamental than the famous one.
Then I ran the control. I recorded from the syntaxin mutants myself and saw the same apparent signal the earlier work had reported. It was not a signal. It was an artifact of the salt solution used in the assay. Their finding did not hold, and my exciting interpretation went with it.
The UNC-13 result itself was real. Deleting the gene did block the signal. What disappeared was the claim I had built on top of it, that UNC-13 was somehow more fundamental than the protein everyone had spent years on. The control did exactly what a control is for. It argued with the story I wanted the data to tell.
Most investors never build a control. They decide, and then they collect reasons.
So write your disconfirming evidence down before you commit, while you can still think about it clearly. What would you have to see in the first year to conclude the thesis was wrong? Which number, which event, which silence from the sponsor? Define it in advance and it becomes information. Define it afterward and it becomes an excuse.
4. What else could explain this?
Confounders. In science, the alternative explanation you failed to consider is the one that ends your paper.
In investing, the most common confounder is the calendar.
When someone shows me a track record, the question is not “are these results good?” The question is: did this person produce these results, or did the years produce them while this person was standing there? A rising market makes a lot of people look skilled. Cheap debt makes even more.
So I ask about the deal that went sideways. Anyone who has been doing this for real has one. What happened, what did they do about it, when did they tell their investors, and what did they change afterward. That answer tells me more than any set of numbers, because it separates a process from a period.
An operator who cannot name a hard one is either new or not being straight with me. Neither is disqualifying by itself. Both change what I am willing to do.
There is a lot more to say about reading an operator. It deserves its own piece, and it will get one.
5. What do I do when it breaks?
Not if. When something in the plan does not go the way it was drawn.
The question was never whether reality will depart from the model. It will. The question is which assumptions move, by how much, and what you do when they do.
Decide while you are calm what you will do when you are not.
That means knowing before you commit where you sit in the capital stack, what happens if the business plan takes two years longer than it should, what a capital call would mean for your situation specifically, and what you would need to see before you would consider putting another dollar in. Write it down.
Your judgment under stress is worse than your judgment right now. You will not believe that while you are calm, which is precisely why it has to be on paper before you need it.
I invested as a limited partner in someone else’s apartment syndication. Before I funded it, I had already decided what I would do if the deal got into trouble and asked for more money. Capital calls are a normal tool, and sometimes they are exactly what carries a good deal through a rough stretch. My rule was narrower than that. If the problems looked structural rather than temporary, I was not going to add to the position.
The call came. I read it as structural. I declined.
The deal did not survive, and the additional capital would have gone with it.
That is precommitment, and it is the whole point. I want to be careful about how I tell the story, because the rule is not what made me right. Plenty of deals do get through a rough patch on a capital call, and the investors who funded it are glad they did. What the rule actually did was let me make the decision from something I had written down while I was calm, instead of from the feeling of watching an investment I liked ask me for money on a deadline.
This is also the question that keeps the other four honest. It is easy to talk about downside in the abstract. It is a different thing to have written the sentence “if this happens, here is what I do.”
What this does not do
I want to be careful here, because this is the point where frameworks usually get oversold.
None of this guarantees you will be right. The future stays uncooperative. What it does is help you find out you are wrong earlier, limit what being wrong costs, and learn something from it instead of just losing money.
Good scientists do not become confident by eliminating uncertainty. They become comfortable acting in spite of it, because they trust the process they used to decide. The same is true of good investors. The ones I respect are not the ones with the best predictions. They are the ones with the best questions.
A word about the neuroscience, since people ask. My training was on the molecular side: cells, synapses, the machinery sitting underneath behavior rather than the behavior itself. So I am not going to hand you a tidy story about your amygdala and the stock market.
But spend years watching very small changes at a synapse produce very large changes in an animal and you come away with a bias. It turns out to be a useful one. Look for the mechanism. Distrust the explanation that only describes the surface. Ask what is actually driving the thing in front of you, then ask what would prove you wrong.
Where this goes
Each of these five questions is bigger than the section I just gave it. Over the next few months I am going to take them one at a time. What disconfirming evidence actually looks like inside a private deal. How to read a track record for skill instead of timing. How to write your own break-glass rules before the glass needs breaking.
You will notice that none of the five questions is “what does it pay.”
That is deliberate. And it is the subject of the next one.
Tim Fergestad, Ph.D. is a scientist and private-market investor. He writes about investing and decisions under uncertainty at TimFergestad.com, and is the founder of Oak Street Assets.
Nothing here is investment, legal, or tax advice. It is how I think, offered in case it is useful to how you think.