Contrarian thinking is not the same as doing the opposite of what everybody else is doing. It is the harder thing: thinking on your own, and being able to say why. And it is not an alternative to value, growth, momentum or index investing. Every one of those styles contains a contrarian decision, and the style only works if you make that decision well.
Start with the question that decides everything
The most common mistake is to ask “what is the crowd doing?” The question that actually decides your outcome is different: what does my return depend on?
Almost every holding’s return comes from two things, and they multiply. The business has to earn more, and the market has to keep paying a similar price for those earnings. If the business earns 40% more while the market’s valuation falls by 30%, you have made nothing — the company did its job and you still lost. That arithmetic is the first thing to check before deciding a position is contrarian, or merely uncomfortable.
Where the contrarian decision hides in each style
Every style asks you to go against something. What changes is which something, and what the payoff looks like.
- Value. The crowd is pricing a business for disappointment. Your bet is that the market has been too pessimistic, and the return comes from a re-rating once sentiment improves. That has two phases, and they feel nothing alike. First the hated phase, when the business is doing its job and the share price is not, and the holding is the worst thing in your portfolio. Then the re-rating, when the market changes its mind and the price catches up with a business that never broke. The margin of safety is what lets you sit through the first phase — and if you cannot survive it, you cannot own it, whatever the balance sheet says.
- Growth. The crowd is pricing a business for ordinary growth. Your bet is that earnings compound faster than the price implies. The payoff is earnings growth, because the multiple is already generous — which means very little margin of safety, and the contrarian decision is when to sell, not whether to buy. Holding a good business through a falling multiple is a position, not a gift, and the exit has to be decided before the multiple starts to fall rather than after.
- Momentum and index. The crowd is trading the trend. Your bet is that the trend has more room, or that your exit is disciplined. Either way you need to know in advance what “the trend is over” looks like, because a trend-following style will always have a long, painful stretch, and the exit has to be written down before that stretch rather than during it.
So the contrarian call is never simply “be different”. It is “be different about this specific thing, and know exactly what you are betting against”.
The same style, both outcomes
Here is the part that usually gets left out of a “be contrarian” argument. The same strategy produced the spectacular winners and the total flops. Value did not simply work; it worked on some names and failed on others, for long stretches, and nothing in the label told you which was which. Growth did not simply work either. The table below works through eight of them — two that compounded for years, two that fell from a peak multiple and did not come back, two that were hated and then re-rated, and two that stayed cheap for the better part of a decade.
| Style | What happened | The lesson | |
|---|---|---|---|
| Asian Paints | Growth | P/E near 100× at the early-2022 peak; earnings grew about 43% over the next four years, but the price fell about 30% as the multiple roughly halved | Holding a business through a halving of its multiple is a position, not a gift |
| Trent | Growth | P/E reached roughly 230× at the 2024 peak; profits roughly doubled over the next two years and the price still fell about 53% from its high | At 200×+, growth has to be extraordinary, not merely good |
| Hindustan Unilever | Growth | Peaked in early 2000 above 60×; sales grew about 6% a year and profits about 6% over the next decade, and the stock went nowhere | Paying 60× for 6% growth is not a style, it is a hope |
| Wipro | Growth | Fell from roughly ₹194 at the dot-com peak to about ₹28 within three months | A multiple can move faster than any business can |
| Apar Industries | Value | Traded at 8× in early 2020 and kept falling while the market rose; profits then grew about 4× and the price re-rated with them | The hated phase is where the money is made, not where it is lost |
| KEI Industries | Value | The same script: under 9× in early 2020 after a heavy fall, then a re-rating to over 40× as earnings tripled | The script repeats; the company does not have to |
| ITC | Value | Sat around 20× while peers traded at 50–80× and went nowhere for years; a ten-year return of 8% against the Sensex’s 160% | Cheap is not the same as mispriced |
| Coal India | Value | Bottomed near 4× in early 2020 and was still roughly where it started twelve years later | A value trap can stay a trap for a decade |
Scroll the table sideways for more →
What the price actually did
The examples above are the essay’s. Here is a different cut, computed the same way for every name: the fall, and how long it took. Two things stand out. The first is that a fall of this size is not an event, it is a decade — most of these bottoms came years after the peak, not weeks. The second is that the wait afterwards is usually longer than the fall itself. A holder who bought the top spent more time waiting than losing.
- Sony-95.1%to the bottom12.8 yrto recover24.7 yr
- Nokia-94.8%to the bottom4.7 yrto recoverNot recovered — still 77% below the peak, 19 years on
- Wipro-92.8%to the bottom2.8 yrto recover11.2 yr
- Cisco-84.9%to the bottom2.4 yrto recover25.2 yr
- Alibaba-79.3%to the bottom2.0 yrto recoverNot recovered — still 65% below the peak, 6 years on
- Trent-71.0%to the bottom1.5 yrto recoverNot recovered — still 43% below the peak, 2 years on
- Wells Fargo-66.9%to the bottom1.5 yrto recover5.7 yr
- Asian Paints-36.4%to the bottom2.2 yrto recoverNot recovered — still 32% below the peak, 3 years on
| Peak | Bottom | Fall | To the bottom | After that | |
|---|---|---|---|---|---|
| Sony (International) | Jan 2000 | Oct 2012 | -95.1% | 12.8 years | recovered in 24.7 years |
| Nokia (International) | Oct 2007 | Jun 2012 | -94.8% | 4.7 years | Not recovered — still 77% below the peak, 19 years on |
| Wipro (India) | Mar 2006 | Jan 2009 | -92.8% | 2.8 years | recovered in 11.2 years |
| Cisco (International) | Apr 2000 | Sep 2002 | -84.9% | 2.4 years | recovered in 25.2 years |
| Alibaba (International) | Oct 2020 | Oct 2022 | -79.3% | 2.0 years | Not recovered — still 65% below the peak, 6 years on |
| Trent (India) | Aug 2024 | Feb 2026 | -71.0% | 1.5 years | Not recovered — still 43% below the peak, 2 years on |
| Wells Fargo (International) | Jun 2007 | Dec 2008 | -66.9% | 1.5 years | recovered in 5.7 years |
| Asian Paints (India) | Nov 2023 | Feb 2026 | -36.4% | 2.2 years | Not recovered — still 32% below the peak, 3 years on |
Scroll the table sideways for more →
The one piece of arithmetic worth memorising
Return = earnings growth × change in the multiple. Multiplication, not addition — which is why one large negative can swallow a lot of positive, and why a large multiple expansion can rescue a mediocre business. Here are a set of combinations. All of them are arithmetic, not opinions about any company.
Five combinations of the same two drivers
Each bar is what you would have earned. The line under it is the arithmetic, so you can check the working rather than take it on trust.
- Business grows 40%, the market pays 30% less−2%
earnings +40% × multiple −30%
- Business grows 40%, the market pays 25% more+75%
earnings +40% × multiple +25%
- Business grows 10%, the market pays 20% less−12%
earnings +10% × multiple −20%
- Business doubles, the market halves the multiple0%
earnings +100% × multiple −50%
- Business grows 25%, the market pays 40% less−25%
earnings +25% × multiple −40%
View the numbers as a table
| Scenario | Earnings | Multiple | Net return |
|---|---|---|---|
| Business grows 40%, the market pays 30% less | +40% | −30% | −2% |
| Business grows 40%, the market pays 25% more | +40% | +25% | +75% |
| Business grows 10%, the market pays 20% less | +10% | −20% | −12% |
| Business doubles, the market halves the multiple | +100% | −50% | +0% |
| Business grows 25%, the market pays 40% less | +25% | −40% | −25% |
Illustration only. These are made-up combinations of two percentages, not market history and not a forecast of any company's returns. The identity behind them — return = earnings growth × change in the multiple — is arithmetic; what you plug in is a judgement, and the judgement is where the risk sits.
What the price already assumes
Second-level thinking is easier to do than it looks, because the price has already made a forecast and you can read it. Take a stock at 60 times earnings, assume the market ends up paying 20 times in five years — a normal multiple for a mature business — and ask what earnings growth is left for you if the price is still going to deliver 12% a year.
The answer is about 40% a year. The same business at 30 times needs about 22% a year. At 20 times the multiple does no work at all, and the earnings have to deliver the whole 12%.
That is what a high multiple is. Not a promise of better growth, but a demand for extraordinary growth, with a built-in assumption that the business gets worse if it does not arrive. It is also the clearest statement of what a contrarian position actually is: you are betting that the market’s forecast is too pessimistic, and now you can name the forecast.
What each starting price demands of the business
Earnings growth a year, over five years, for the price to still deliver 12% a year if the market ends up paying 20 times earnings. The lower you buy, the less the price asks of the business.
- 60× today, 20× in five years39.5% a year
- 40× today, 20× in five years28.7% a year
- 30× today, 20× in five years21.5% a year
- 20× today, 20× in five years12.0% a year
- 15× today, 20× in five years5.7% a year
View the numbers as a table
| Starting P/E | Exit P/E in 5 years | Earnings growth needed a year |
|---|---|---|
| 60× | 20× | 39.5% |
| 40× | 20× | 28.7% |
| 30× | 20× | 21.5% |
| 20× | 20× | 12.0% |
| 15× | 20× | 5.7% |
Arithmetic, not a forecast of any company. The assumptions are stated on every bar: the starting multiple, a 20× exit in five years, and a 12% a year target. Change any of them and the answer changes — which is the point, because a price is a forecast and you can read it.
Second-level thinking: ask what the consensus already assumes
The most portable version of this comes from Annie Duke’s Thinking in Bets. First-level thinking is the obvious analysis: is this a good business, are earnings growing, is the balance sheet sound. Second-level thinking asks a harder question — what is already priced in, and what would have to be true for the market to be wrong?
Second-level thinking is not about being cleverer than everybody else. It is about being specific. “The stock is down 40%, so it is cheap” is first-level. “The price implies earnings fall another 30% from here, and the contract that the whole story rests on is not up for renewal for two more years” is second-level — because it states what has to happen, and you can go and check it.
Five second-level questions, before you call a position contrarian
What is already in the price?
Write down the earnings level or growth rate the current price implies. If you cannot, you are not thinking second-level — you are reacting to a price move.
What would have to be true for the market to be wrong?
State it as a fact that can be checked, not as a feeling. 'Growth slows' is not checkable; 'the contract that the story rests on is not up for renewal for two years' is.
What would have to be true for you to be wrong?
If you cannot answer that, you are not taking a position. You are taking a hope — and a hope cannot be sized.
How long can I be wrong?
'Long enough for this to become my worst holding' is an answer. 'I will hold it for a while' is not. A contrarian call that breaks your plan is not a contrarian call.
What is my exit — as a condition, not a price?
The thing that would prove you wrong, or the thing that would make the opportunity gone. If you cannot name the condition, you have already decided never to sell.
Why it usually goes wrong
- You were early. The most common outcome of a correct contrarian call is that it stays painful for a long time. A position can be right and still cost you money for two years, and you have to be able to live with that before you take it.
- You were late, not early. “Everyone owns this” is not by itself a reason to sell. It might be, if the price already reflects it. The question is what the price says, not how many people hold it.
- Being contrarian became the crowd. This is the trap nobody warns you about. Once “buy the dip” is in every newsletter, the contrarian trade is no longer contrarian — it is the new consensus, and it is priced in.
- Cheapness is not mispricing. Look again at the table. The value failures did not turn out to be expensive. They stayed cheap — for years — while the business did nothing wrong and the market never changed its mind. A value trap looks exactly like a bargain for as long as you can hold your breath.
- Your analysis was wrong. Agreement with the crowd does not make you right, and disagreement does not make you wrong. Only the analysis and the judgement do.
What a retail investor’s version looks like
If you are not picking stocks, the contrarian decision does not disappear — it moves. It becomes a question about deploying money and about when to stop adding to it. In a market where most people have stopped looking at equities because the headlines are bad, sticking to your plan is the contrarian act. The most common way retail investors lose money in a drawdown is not picking the wrong company; it is selling the right one at the bottom, because it stopped feeling like a good idea.
Which is why a written plan matters more than a good opinion. The decision to sell a falling holding, or to keep buying it, should be made when you are calm and the arithmetic is in front of you — not while the position is shouting at you.
Educational content, not financial advice. Nothing here is a recommendation to buy or sell any security. The worked numbers in the charts are arithmetic from the two identities above, not market history. The company examples are worked through in the essay cited in the sources.