The kill switch: market psychology and your own mind
The answer
A kill switch is a sell rule you write while nothing is happening, in observable terms, with a date on it.
It exists because the mind that has to sell is not the mind that bought. Every known bias in investing arrives at the same moment: when the position is large, the news is loud, and there is no time. The switch moves the decision to a moment when none of those things is true.
Why this costs you money
There is one specific behaviour that costs individual investors more than any other, and it has been measured.
People sell their winners and keep their losers. Terrance Odean, in a 1998 paper in the Journal of Finance using the records of thousands of individual accounts, found that investors realised gains at a significantly higher rate than losses, and that the winners they sold went on to outperform the losers they kept.
The reason is that a loss is not a loss until you sell. While you hold, it is a temporary situation with a story attached. The moment you sell, it becomes a number that describes your judgement.
So people wait for break-even. Not for a price at which the company is fairly valued, and not for a price where the money could be better used. For the exact price they happened to pay, which is a fact about their own past and about nothing else in the world.
Meanwhile the money sits in the worst position they own, doing nothing, for years, while the position they should have kept was sold at a small profit to produce the feeling of having been right.
The cost is not one bad trade. It is a portfolio that gradually fills with mistakes, because the mistakes never leave, and empties of successes, because the successes are the only ones easy to sell.
How it works
Market cycles have a shape, and so do the feelings that go with them. The sequence is roughly the same every time.
| Phase | What prices are doing | What it feels like | The decision it produces |
|---|---|---|---|
| Optimism | rising steadily | reasonable, justified | buying a sensible amount |
| Greed | rising fast | urgent, obvious | buying more than planned |
| Denial | first serious fall | temporary, a chance to add | refusing to reduce |
| Panic | falling fast | permanent, dangerous | selling something |
| Capitulation | falling, low volume | hopeless, finished | selling everything |
| Despair | flat, near the low | never again | holding cash through the recovery |
| Hope | rising off the low | suspicious, a trap | staying out |
Read the last column. Every decision in it is the opposite of the correct one, and every one of them feels correct at the time. That is what makes this different from a knowledge problem.
Five biases produce that column.
Loss aversion. A loss of a given size feels considerably more powerful than a gain of the same size. The finding comes from the work of Daniel Kahneman and Amos Tversky, published in 1979. This is what makes holding a loser feel safer than selling it, though the money at risk is the same either way.
Anchoring. Your judgement attaches to the first number you saw, usually your purchase price or the stock's highest price. Neither influences the future. The market does not know what you paid.
Recency. Recent events feel more likely to continue than they are. After 3 good years, risk feels low and future returns feel high. Both beliefs are most wrong at exactly the moment they are strongest.
Herding. When you do not know what to do, other people's behaviour becomes evidence. In a market this is circular, because they are watching you. A crowd carries information about a crowd, not about a company.
The fear of missing out. The damage is not that you buy. It is that you buy without a sell condition, because the reason for buying was that something already rose, and "it rose" contains no instruction about when to stop.
The defence is not to feel differently. You will not. The defence is to make the decision at a moment when the biases are not active, and to write it so nothing is left to interpret later.
There are only 3 legitimate reasons to sell, and every kill switch should name one of them.
- The reason for owning it is no longer true. The thesis broke. This is the most important one and the hardest to write, which is why it must be written in advance.
- The size is wrong. The position has grown past its limit and must be trimmed. This is not a judgement about the company at all.
- The money has a better job. Something else is clearly more attractive, or you need the money for the purpose in your plan.
"It fell 20%" is not on the list, unless a fall of 20% is the declared sell condition of a momentum position. A price fall is not information about a business. It is the thing you were trying to make a decision about.
What it tells you, and what it does not
A written sell rule tells you what to do on the day the news arrives, which is the only day it matters. It converts the worst decision you will make all year into an instruction you follow.
It does not tell you that you will follow it. A rule is only as strong as the habit of reading it. This is why the review date in your plan matters: the rule must be in front of you at a scheduled moment, not stored somewhere you visit when you are already frightened.
It also does not remove emotion. Nothing does. You will feel exactly the same things in the next fall as you did in the last one. The rule works by making the feeling irrelevant to the action, not by reducing the feeling.
And a rule can be wrong. A sell condition written 2 years ago can be triggered by an event that turns out not to matter, and you will sell something you should have kept. That happens. Over many decisions the written rule still wins, because its errors are random and the errors it prevents are systematic.
The decision rule
For each holding, write one line and date it:
"I will sell [holding] if [observable event], regardless of what the price is doing. Written on [date]."
Then apply the 3 tests:
- Observable. Could a stranger read the event and agree it had happened?
If not, rewrite it.
- Not a price move. Does it name a business fact or a size breach? If it
names only a price fall, ask which style that belongs to and whether that is really how you bought.
- Dated. Without a date it is not a rule. It is an opinion you had once.
Try this now
Five minutes, on one holding. Start with the one you are least comfortable about, because that is the one where the rule is missing.
- Pick a holding you have owned for more than 6 months.
- Write the sentence: "I own this because \_\_\_." One line, from the first article in this cluster.
- Now write: "I will sell it if \_\_\_." The blank must be an event another person could confirm: a specific fall in a number in the results, a change in management, the loss of a licence or a large customer, a debt level crossing a figure you name, or the position exceeding a stated percentage of your portfolio.
- Write today's date at the end of the line.
- Put the line somewhere you will see it at your next scheduled review. The notes field in your broker app works. A single text file works.
What you should see. The sentence in step 3 is much harder to write than the one in step 2. Most people can say why they own something. Very few have ever said what would end it. That difficulty is the finding, not an obstacle.
For at least one holding you will not be able to write step 3 at all, because the honest answer is "if it goes down enough that I give up". Write that down too. It tells you the position was never sized or reasoned.
And for one holding you will write the rule and immediately notice the event has already happened. That is the most valuable 5 minutes in this cluster.
Repeat for 2 more holdings this week. Do not attempt all of them today.
Three real cases
1. Yes Bank, March 2020 (India) — waiting for break-even The Reserve Bank of India placed Yes Bank under a moratorium on 5 March 2020, limiting withdrawals to ₹50,000 per depositor. The Union Cabinet approved a reconstruction scheme on 13 March 2020, under which State Bank of India led a group of investors putting in ₹120 billion. Additional Tier 1 bonds issued by the bank were written down entirely, a decision later challenged in court. Many individual shareholders had bought during the fall that preceded this, at prices they described as cheap compared with the bank's earlier highs. The earlier high was an anchor. It contained no information about the bank's capital position, which was the only fact that mattered, and which was observable in the filings before the moratorium.
2. The Nasdaq Composite, 10 March 2000 to 9 October 2002 (United States) — the full emotional cycle, dated The Nasdaq Composite peaked at 5,048.62 on 10 March 2000 and reached a low of about 1,114 on 9 October 2002, a fall of roughly 78%. The whole sequence in the table above is visible in that period: buying accelerating into the top, denial through the first fall in spring 2000, repeated rallies that produced hope, and a final low reached when almost nobody was interested. Individual companies with real businesses fell 80% or more and took many years to recover. The information required to be cautious was public throughout. What was missing was any rule that could act on it while prices were still rising.
3. GameStop, 28 January 2021 (United States) — the fear of missing out, on one day GameStop shares began January 2021 around $17 and reached an intraday high of $483 on 28 January 2021, with pre-market trading above $500 the same morning. The largest volume of retail buying occurred close to the peak, after the story had been reported everywhere. Several brokers restricted purchases the same day, citing collateral requirements. The price fell over 80% from the peak within days. Almost nobody who bought on 28 January had a sell condition, because the reason for buying was that the price had risen, and a rising price does not tell you when to stop.
The question that resolves it
A novice looks at a falling holding and asks: will it come back?
An expert asks: would I buy this today at this price, knowing what I now know?
The second question removes the purchase price from the decision, which is the single most powerful move available in investor psychology. If the answer is no, you are holding it for a reason that has nothing to do with the company.
What would make this wrong
If these biases were not real, then individual investors would sell winners and losers at similar rates. They do not, and the pattern appears in account data across countries and decades.
The honest limits.
Written rules can be over-applied. An investor with a sell condition on every holding, checked weekly, will trade far too much and pay for it. The rule exists to be consulted at scheduled reviews and when its named event happens.
A sell rule can also be a way of avoiding thought. "I will sell if the price falls 15%" is easy to write and almost always wrong for a long-term holding, because it converts a normal fluctuation into a permanent loss. The difficulty of writing a good rule is not a flaw in the method. It is the method.
Biases are not always errors. Loss aversion is a reasonable response for somebody whose money cannot be replaced, and refusing to sell into a panic has been correct far more often than not. The problem is a feeling arriving at a moment when it cannot be examined.
And one limit applies to this whole cluster. Knowing about a bias does not reduce it. Every one of these effects has been demonstrated in people who had just been taught about them. That is why the defence is procedural rather than educational.
In India
Three features of the Indian market interact with these biases.
Circuit limits. A stock at its lower circuit cannot be sold, because there are no buyers at that price. An investor who intended to act on a rule may find the decision has been removed. This is a strong argument for writing rules that trigger on business events rather than on price falls, because business events are visible before the price band closes.
The absence of market makers in cash equities. No firm is required to quote a price in an ordinary listed Indian company. In a panic, small companies can go days with almost no trading. The plan must assume that the exit from a small position may not exist during the week you want it.
A derivatives-weighted retail market with published outcomes. SEBI's studies of individual traders in the equity futures and options segment report that a large majority made net losses, and that a large majority of loss-making traders continued to trade afterwards. The September 2024 study reported that more than 75% of loss-making traders continued in the segment. That is the behavioural finding inside the financial one, and it is the clearest published evidence anywhere that the problem is not information.
Indian investors also face a particular version of herding, because tips travel through messaging groups with no record and no accountability. A written, dated rule is the only thing in the account that has a timestamp.
In the United States
The American market produces the same biases with different equipment.
Continuous coverage. Financial news channels, social media and brokerage applications designed to be checked often all increase the number of occasions on which a decision can be made. More decisions is worse, because the average decision made in reaction to a screen is below the average decision made on a schedule.
Automatic contribution as a defence. The 401(k) payroll deduction is the most effective behavioural tool in either market, because the money is invested before the investor sees it. Contributions continued through March 2020 for millions of people who would not have chosen to invest that month.
Company stock and identity. Employees holding stock in their own employer face loss aversion and anchoring on a position they also feel loyal to. That combination resists a rule, which is why the rule must be written before the position becomes large.
The Odean 1998 study above used United States brokerage data.
Where they differ, and what that tells you
The biases are identical. Human beings are not different in the 2 countries. The equipment that turns a bias into a loss is different.
In the United States, the most common route from bias to loss runs through frequent decisions in a taxable account and through concentration in employer stock. The defence that works is automation: payroll deduction, target-date funds, scheduled rebalancing.
In India, the most common route runs through leveraged derivative positions with short expiries, where a bias has hours rather than months to produce a loss, and through small companies that cannot be exited when everybody wants out at once. The defence that works is a stated instrument limit and a stated size limit, because automation cannot help with a position that should never have been taken.
What that tells you is which sentence to add to your own rule. An American investor's most valuable extra line is usually about automation: "the monthly contribution continues regardless." An Indian investor's is usually about instruments and size: "no leveraged position, and no position above this percentage, regardless of conviction."
Both are one sentence. Both must be written before they are needed.
Carry this
- Write the sell rule before you need it, name an observable event, and date it.
- There are 3 legitimate reasons to sell: the reason broke, the size is wrong, the money has a better job. A price fall is not one of them.
- Your purchase price is a fact about your past. The market has never heard of it.
- Ask "would I buy this today at this price?" That question removes the anchor.
- Knowing about a bias does not reduce it. Only a written procedure does.