1. Writing the thesis as an explicit bet
Level 6: From reading to decision
The previous five levels have assembled the instrument: a price is a forecast, that forecast breaks down into value drivers, there is a mechanism for walking the road backwards, the market pays for the surprise and not for the figure, and the driver that explains the most value is how many years the advantage survives. With that you know how to read what a price takes for granted. What is missing is the one thing that turns that reading into money or into a loss: deciding.
And deciding starts with a gesture that looks like paperwork and is not. Writing the thesis down. Beforehand.
"I like the company" says absolutely nothing
That the business looks good to you, that management is convincing, that the product sells. All of it is true and, precisely because it is true, it is already inside the price: the people who bought before you believed it too.
Rappaport and Mauboussin sort the problem out with an idea worth repeating to yourself: you do not invest in companies, you invest in the difference between what a company will do and what its price already takes for granted it will do. An extraordinary company whose price discounts the extraordinary is a run-of-the-mill investment; a mediocre one whose price discounts disaster can be an excellent one. The quality of a business and the quality of an investment are different things.
Something uncomfortable follows from that: a thesis is necessarily comparative and necessarily a disagreement. If you agree with the market, you do not have a thesis; you have a description.
The thesis in five lines
- Line 1. What the price assumes. Translated into the value drivers from level 2, with figures and a horizon: what sales growth, what margin, what incremental investment per euro of new sales, what cost of capital, and for how many years all of that holds.
- Line 2. What you believe. The same quantities, your version. With a number and a deadline. "Better than people think" is not a line 2.
- Line 3. Where the difference lies. One single driver. The one that decides it. If you cannot point to one, you do not have a thesis.
- Line 4. Why you and not the market. What you see, know or can bear that whoever sets that price does not see, does not know or cannot bear.
- Line 5. How it gets settled. What observable event, at roughly what date, will turn your belief into price.
Five lines. It fits on half a sheet of paper. If it does not fit, there is more than one thesis stuffed inside it.
Line three is what separates a thesis from an opinion
Disagreeing on one driver is a bet. Disagreeing on all five is almost always an assembly error: a cost of capital put in wrong, forgotten debt, a share count that ignores dilution, a commitment that does not appear on the balance sheet. The leading hypothesis is not that the market is asleep.
There is a worse and more frequent case: disagreeing on two drivers in opposite directions. You believe in less growth and in a higher margin. It may be a magnificent thesis, but as written it says nothing, because the net effect depends on arithmetic you have not done yet. A thesis that points up and down at the same time cannot be refuted, and what cannot be refuted is no use for deciding.
Line four is the most uncomfortable, which is why hardly anyone writes it
There are three legitimate answers and a great many that are not.
- Horizon. You can wait four years. A good part of the money setting that price is judged every quarter and cannot. That asymmetry is probably the only structural advantage a private individual has.
- Interpretation. This is what Howard Marks calls second-level thinking: everybody knows the fact, but you read a different consequence out of it and can say which one.
- Arithmetic the aggregate hides. The market extrapolates the consolidated figure; you have separated out what sits underneath, and that breakdown is already published.
These are not a line 4: having read the annual accounts end to end, management inspiring confidence in you, or having put in a great many hours. Effort is not information. And if you cannot find a line 4, the likely explanation is not that none exists: it is that you have no edge in that idea.
Why writing it beforehand changes the outcome
Because memory does not store decisions, it reconstructs stories. If you win, you will remember a crisp thesis that came good. If you lose, you will remember that you had a bad feeling about it all along. Neither version existed when you bought, and without the original you cannot do the one thing that produces learning: separate the quality of the decision from the quality of the outcome. Annie Duke insists that a good decision can turn out badly and a terrible one can turn out well, and that judging by the scoreboard is the shortcut that stops you improving.
Philip Tetlock found, across twenty years of scored forecasts, that the people who improve are the ones who state specific, dated, checkable claims and then score themselves. Those who speak in generalities are never wrong and never learn.
There is an immediate effect too: writing it down is how you find out you do not have one. Lines 3 and 4 come out blank far more often than anyone expects, and that half-finished sheet of paper has prevented more losses than any indicator.
Worked example (illustrative figures, not from any real company)
Price 100. The reverse reading says: sales growing 9% a year for seven years, operating margin steady at 18%, thirty cents of incremental investment for every euro of new sales.
- Line 1: that.
- Line 2: growth of 6%, not 9%; a margin of 22% within three years.
- Line 3: the margin. That is where it is decided, and you accept that on growth the market is ahead of you.
- Line 4: that 18% blends two businesses. The higher-margin one accounts for 30% of sales today and accounted for 12% four years ago. The company publishes the mix every quarter; the market extrapolates the aggregate.
- Line 5: at the pace of the last eight quarters, the consolidated margin passes 20% within about six more and stops being something anyone can ignore.
That is no longer a fondness for something. It is a bet with a name, a number and a date. And a bet with a date can be lost, which is exactly what the next module needs.