Your sister is a chartered accountant. You send her the annual report of a company you have spent three weeks on and are about to put ₹6 lakh into. She reads it over a weekend and rings on Sunday evening to say she would not touch it. You ask why. She says the receivables look wrong to her and the growth is being bought rather than earned. You explain the receivables — there is a genuine reason, you have checked it. She listens, and says she still would not touch it. The conversation ends politely and nothing has changed for either of you, and the ₹6 lakh goes in on Tuesday.
Something valuable was available in that call and it was not collected. A competent person with no stake and access to the same documents reached a different conclusion — which is among the rarest and most useful inputs an individual investor ever gets, and rarer still because most people never seek it. The waste is not caused by stubbornness. It is caused by the conversation having no structure, so it defaults to the only structure available, which is a mild argument.
A second opinion is not useful because the second doctor is better. It is useful because the two of them looked at the same films and, if they disagree, the disagreement localises to something specific — a shadow one reads as scarring and the other as something else. A patient who only hears "he says operate, she says wait" has learned nothing. A patient who gets them to name the shadow has learned exactly where the uncertainty is.
A disagreement about an investment works the same way. "He likes it, she does not" is worthless. The value is entirely in locating the one fact or judgement on which the two views actually diverge — and getting there requires a question that nobody asks by default.
Four disagreements that look identical in conversation
Before anything can be resolved, the disagreement has to be sorted, because the four kinds require completely different handling and a conversation that mixes them goes nowhere for hours. Almost every unproductive money argument is two people working on different rows of this table without knowing it.
| Kind of disagreement | What it sounds like | How it is settled |
|---|---|---|
| About a fact | "Their debt has gone up." "No it has not." | By opening the document. This is the easiest kind and it is astonishing how often an hour is spent on it — settle facts first, before any discussion of what they mean |
| About what the fact implies | "Receivables up 40% means the earnings are not real." "Receivables up 40% is what happens when you win a government order book." | Not by evidence alone. Both readings are consistent with the number, so the argument is about which mechanism is operating — and that is answerable by looking for a third fact that only one mechanism would produce |
| About probability and weight | "The regulatory risk is serious." "The regulatory risk is overdone." | Only by making the numbers explicit. Two people who both say "unlikely" and mean 5% and 30% will argue indefinitely. Say the numbers aloud and the gap usually turns out to be smaller, or usefully larger, than the argument suggested |
| About goals and risk capacity | "I would not put six lakh into one midcap." "It is 4% of my portfolio." | Not by evidence at all, ever. This is a disagreement about whose money it is and what it is for. It is settled structurally — by limits and separate pots — and treating it as an argument about the company is how households spend years disagreeing about the wrong thing |
Finding the crux
Within the first three rows, the productive move is to locate the crux: the one fact or judgement such that, if it were reversed, you would change your conclusion. Both sides state theirs. If either cannot name one, that is the finding — a position with no crux is not a view about the company, it is a preference, and no amount of discussion will move it.
- 1Write down what would change your mind — before they reply
Two or three specific, checkable things. Doing this first is the most reliable defence available against confirmation bias, because once the rebuttal is under way every argument will be assessed by whether it supports the conclusion you have already reached. Written in advance, the list is a commitment you can be held to by your own handwriting.
- 2State their case until they agree you have stated it
Steelmanning is not politeness. It is a test of whether you have understood the objection at all, and it fails far more often than people expect. "So your view is that the growth is funded by extending credit to weaker customers, and the receivable ageing is where that would show?" If they say no, that is not a setback — you have just learned that you were about to rebut something they never said.
- 3Ask directly for their crux, and give yours
"What would have to be true for you to change your mind?" and "Here is what would change mine." Two sentences. The disagreement usually collapses from a whole company to a single line item, and that line item is then something one of you can go and check.
- 4Convert the crux into a dated, checkable statement
Falsifiability is what turns a disagreement into information. "The receivable days will not come down" is an opinion. "Receivable days will still be above 90 in the next two annual results" can be wrong, and in eighteen months one of you will have been. Write both statements down with the date. Without that the disagreement leaves no trace, and neither of you ever finds out which of you reads accounts better.
- 5Decide separately from resolving
The conversation does not have to end in agreement, and it usually should not. What it should end in is a smaller, sharper open question and a decision made in full knowledge of it — including, quite often, a smaller position than you originally intended, which is what genuine unresolved uncertainty is supposed to produce.
Whose disagreement counts
Not all disagreement is information, and treating it all as equally weighty is its own failure. The weight of a dissent depends on whether the person is inside their circle of competence, whether they have looked at the same material, and whether their incentives are aligned with yours.
- They have read the same documents, not a headline about them
- They can state your case back to you accurately
- They name a crux, and it is checkable
- They have no position, no commission and nothing to prove
- They work in or near the industry and know something about how it operates
- "That sector never does well" — a view about a category, not about the company
- A confident opinion from somebody who has not opened the report
- Competence in another field being applied here as authority
- Someone who is short of the stock, sells a competing product, or earns a commission on the alternative
- "I just do not have a good feeling about it" — which is honest, and is not analysis
The row that evidence cannot settle
The fourth row of the table deserves its own treatment, because it is the one that damages households. When the disagreement is about how much risk the family should carry, no amount of analysis resolves it, and continuing to argue the company is a way of not having the real conversation. The resolution is structural rather than rhetorical. The household mechanics of it — whose money, which pots, what has to be decided jointly — are the subject of their own lessons in this track; what belongs here is only the recognition, because a conversation that never reaches this row will run for years on the accounts.
- Separate the two questions explicitly. "Is this a good company?" and "should this household have ₹6 lakh in one midcap?" are different, and a person can be right about the second while wrong about the first.
- Give the more cautious view authority over size rather than over selection. A workable household rule: the more risk-averse person sets the ceiling on how much goes into any single decision, and the person doing the work chooses what goes in it. Both constraints are then respected without either view having to be defeated.
- Agree the limits in a calm week, in writing, and in rupees. Limits negotiated during a disagreement are negotiated by whoever is more forceful that evening.
- Keep a genuinely separate pot if that is what it takes. An amount each person may deploy on their own judgement, whose size neither has to justify, removes the majority of these arguments at a cost that is usually small against the peace it buys.
- Treat persistent silence as a red flag, not as consent. A household where one person stopped objecting two years ago has not reached agreement, and you will find that out at the worst possible moment.
The Sunday phone call
A capable friend who has read the same annual report tells you the growth is being bought with receivables and that he would not invest. You have a reason for the receivables that you find convincing. He is unmoved. You intend to buy on Tuesday, at 4% of your portfolio.
Your spouse objects to a ₹6 lakh position in a single midcap. You demonstrate, with the accounts, that the business is sound. The objection remains. What is most likely happening?
Module checkpoint: taking a decision apart
5 questions. Answers are revealed once you submit all of them.
1.You lose money on a holding after a fact emerged that had been sitting in the previous year’s annual report, which you did not read past the highlights. Which bin does this belong in?
2.Which of these two commitments should receive more deliberation: ₹15 lakh into a liquid index fund, or ₹3 lakh of booking money on an under-construction property?
3.A person says they "booked profit" in one holding and that another is "only a notional loss". What is the inconsistency?
4.A rights issue closes in four days and you have not looked at the company in two years. Which fact most affects what you can still do?
5.A capable friend who has read the same annual report disagrees with you and will not be moved by your explanation. What is the most useful next step?
Behen CA hai, wahi annual report padhi, aur bol rahi hai ki main haath nahi lagaungi. Aap apna jawab dete ho, woh maanti nahi, phone rakh jaata hai — aur mangalwar ko paisa lag jaata hai. Sabse kaam ki cheez wahin chhoot gayi. Ek sawaal poochh lo: "aisa kya nikle toh tumhara mann badal jaaye?" — aur apna bhi likh do. Bahes poori company ki thi, ab ek line pe aa gayi, aur woh line document mein dekhi ja sakti hai. Aur agar shaq mita nahi paate, toh position chhoti kar lo — size isi kaam ke liye hota hai.
- Sort the disagreement first: about a fact, about what it implies, about probability, or about risk capacity.
- Only the first three are settled by evidence; the fourth is settled by structure, and treating it as an argument about the company wastes years.
- Write down what would change your mind before hearing the rebuttal, and state their case until they agree you have stated it.
- Locate the crux and convert it into a dated, falsifiable statement — that is what turns a disagreement into information.
- An objection you cannot dismiss and cannot verify in time is exactly what a smaller position size is for.
Mark it done to track your progress through the curriculum.
Common questions
Short, direct answers to what people ask about this topic.
- crux meaning in a disagreement
- A crux is the one fact or judgement such that, if it were reversed, you would change your conclusion. Once both sides name theirs, an argument about a whole company usually collapses to a single line item that one of you can go and check. If a person cannot name a crux at all, that is the finding — a position with no crux is a preference rather than a view, and no amount of discussion will move it.
- stating the other side’s argument in its strongest form before replying is called
- Steelmanning — the opposite of attacking a straw man. It is not politeness but a test of whether you have understood the objection at all, and it fails far more often than people expect. State their case back until they agree you have stated it; if they say no, you have just learned that you were about to rebut something they never said.
- whose disagreement about a stock is actually worth listening to
- Someone who has read the same documents rather than a headline about them, can state your case back accurately, names a crux that is checkable, and has no position, commission or point to prove. A credential is not a conclusion: a chartered accountant is authoritative on how a transaction has been accounted for and is not thereby authoritative on whether the business will do well. Confidence from outside a person’s circle of competence, or from a forum where everybody already agrees, is not evidence about the company.
- we both said the risk was unlikely and still disagreed — why
- Because “unlikely” is not a number, and two people who privately mean 5% and 30% can argue indefinitely without discovering that they are describing different things. Saying the probabilities aloud usually shows the gap is either much smaller than the argument suggested or usefully larger. This only settles disagreements about weight and probability — disagreements about a fact are settled by opening the document, and should be settled first.
- how do I record a disagreement so I find out later who was right
- Convert each side’s crux into a dated statement that could turn out to be false. “Receivable days will not come down” is an opinion; “receivable days will still be above 90 in the next two annual results” can be wrong, and in eighteen months one of you will have been. Write both statements down with the date, because without that the disagreement leaves no trace and neither of you ever finds out who reads accounts better.