Emotional investing, and the rules that survive it

The defence is not calm

The reliable defence against an emotional investing decision is not a calmer investor. It is a decision already made — automated, written down, or reduced to an arithmetic check — at a moment when making it cost nothing.

Almost everything written on this subject asks the reader to feel differently. Stay disciplined. Don’t react. Think long term. As instructions these are fine; as designs they are hopeless, because they ask for the one input that is guaranteed to be unavailable at the moment they are needed.

The word usually reached for here is panic, and it is the wrong word, because it names the person instead of the process. Nothing about the person changed between Tuesday and Thursday. What changed is the conditions the decision is made under — and conditions are a thing you can design around, in a way that character is not.

So this article is about the design. What stress does to a decision, why the same portfolio produces different actions depending on how you arrived at it, and the three structural fixes — automation, a written policy, a rebalancing rule — that turn out to be one fix wearing three hats.

What stress does to the decision

Start with the concrete version. It is 09:40 IST, the index opened 4% lower, the portfolio screen is open, and a decision is being made about money that is not needed for 18 years.

Two things are different about that moment, and neither is a character flaw.

The first is attention narrowing. The psychologist J. A. Easterbrook proposed in 1959, in Psychological Review, that as emotional arousal rises, the range of cues a person actually uses to make a decision shrinks. He was describing performance on laboratory tasks, not portfolios, so applying it to a screen full of red is an inference rather than a demonstrated result — but the inference is a small one, and it describes the experience accurately. The 18-year horizon, the contribution schedule, the reason the allocation was chosen: all of it is still true and none of it is in view.

The second follows from the first, and it is worth saying that it follows rather than being separately demonstrated. Narrowing does not drop cues at random: what survives is what is present — a price, a day’s move, the last three months — and an 18-year horizon is not present anywhere on the screen. It is a fact about the goal, carried in memory, which makes it exactly the sort of cue that goes first. So the decision ends up weighed over the horizon the screen is showing, not the horizon the money actually has.

Put the two together and the shape of the problem is clear. A long-horizon decision is being taken with a narrowed field of view and a shortened horizon. That is not a bad decision made by a bad decider. It is a decision being made in the wrong units.

The same portfolio, arrived at two ways

Two people each hold ₹18 lakh in the same equity scheme, with the same goal and the same date. One arrived there from ₹22 lakh. The other arrived from ₹14 lakh.

Their holdings are identical. Everything that will happen next is identical. The decisions are not identical, and nothing in the portfolio accounts for the difference — it lies entirely in the direction each of them arrived from.

The mechanism is an asymmetry Daniel Kahneman and Amos Tversky set out in Econometrica in 1979: the curve mapping an outcome to the weight it carries in a decision is steeper on the loss side than on the gain side, so a loss counts for more than a gain of the same size. They named the framework prospect theory. The popular version attaches a multiple to it — losses hurt about twice as much. Treat that number with suspicion; the magnitude is contested in the literature, and the effect needs no coefficient to explain what is happening on the screen. The asymmetry itself is a subject in its own right; what it does here is set the reference point.

What it explains is that the reference point does the work. The portfolio is not being judged against the goal it exists for. It is being judged against the highest number it ever showed, which is a figure with no economic meaning whatever — it is simply the most recent thing that felt like the truth.

The same asymmetry shows up in what people actually trade. Hersh Shefrin and Meir Statman named it in 1985 — the disposition effect, the tendency to realise gains readily and hold on to losses — and Terrance Odean documented it in 1998 in the accounts of a US discount brokerage, finding that investors realised their winners at a higher rate than their losers. One market, one dataset, one period: it is an empirical finding with a scope, not a law of nature. But it is the same shape, and it is why the question “shall I book this?” arrives so much more readily than “shall I close that?”

Where the decision gets made

Every technique in the rest of this article does one thing: it moves a decision out of the moment it arises and into a moment when it is cheap. Here is what that shift buys, and what each shift costs, because none of them is free.

Read the last column rather than the third. Every one of these rules is worse than perfect judgement, and the whole case for them rests on perfect judgement not being on the menu at the moment it would be needed.

Three techniques, one mechanism

The three standard answers are usually presented as a list of unrelated good habits. They are not unrelated. They are one idea implemented three ways, and seeing that is what lets you build a fourth.

• Automation moves the decision earlier in time — it is taken once, months before the conditions that would corrupt it exist.
• A written policy moves it into a different state — same person, different conditions, and the version who wrote it can bind the version who reads it.
• A rule with a threshold moves it out of judgement entirely — into arithmetic, where the answer does not depend on who is asking.

So the generator is: for any decision you keep getting wrong, ask which of those three moves is available. The target is not better judgement under pressure but fewer decisions that require it.

And there is a reason this beats simply trying harder, which is mechanical rather than motivational. The occasions on which discretion actually gets exercised are not a random sample of occasions. They cluster in the weeks when the market is moving most — which is to say the weeks that move the portfolio’s value furthest, in whichever direction. A policy that is better on an ordinary Tuesday and gets overridden in a bad March has been evaluated on the wrong days.

Automating the contribution

The strongest of the three is also the dullest, and it is strongest for a structural reason: it is the only one that removes the decision rather than constraining it. A standing instruction debits a fixed amount on a fixed date without asking anyone anything.

A systematic investment plan — a SIP — is usually explained through rupee-cost averaging, which is a real but secondary benefit. The primary one is structural: the default flips, so that inaction becomes the thing that continues and skipping requires an act. The act is demanded at exactly the moment you are trying to make acting expensive.

That defaults do heavy lifting is one of the better-evidenced findings in the field. Brigitte Madrian and Dennis Shea studied a single US firm that switched its retirement plan to automatic enrolment (Quarterly Journal of Economics, 2001) and found participation rose sharply when the default flipped, with nothing else changed. Richard Thaler and Shlomo Benartzi’s Save More Tomorrow programme (Journal of Political Economy, 2004) used the same lever forward in time, letting people pre-commit future pay rises to higher contributions. Both are US retirement plans, not Indian mutual funds, so what transfers is the mechanism rather than any figure: whatever happens when no decision is made is what mostly happens.

There is a related finding worth knowing, because it cuts against a habit that feels like diligence. Shlomo Benartzi and Richard Thaler’s work on myopic loss aversion (QJE, 1995), and the laboratory experiment reported by Thaler, Tversky, Kahneman and Schwartz (QJE, 1997), point the same way: participants who saw the outcomes of their investments more often allocated less to the risky asset. Checking more frequently is not the same as knowing more — it changes the frequency at which the loss asymmetry gets applied.

The trade-off, stated plainly, because automation is not free either. The same insulation that stops fear stopping the SIP also stops information stopping it. A mandate set five years ago keeps buying an allocation that may no longer match the goal, and nothing in the system will ever raise its hand about that. Which is why the review has to be a scheduled event with a date, not a response to a feeling — and why the honest treatment of whether to stop a SIP in a fall starts by separating a cashflow problem from a fear problem, since automation is the right answer to only one of them.

The policy written while it is boring

An investment policy statement is a page that says what you will do, written at a time when nothing is happening. Institutions have had them for decades. For an individual the version that works is roughly one page and takes an evening.

What earns its place on it:

• The target asset allocation, and the band around it that triggers action.
• The contribution — amount, date, and what happens to it when income rises.
• What would make you exit a holding, stated as a fact about the holding rather than about its price.
• What you will do at a 30% drawdown — a fall from peak to trough — including the case where you do nothing, written down as a choice.
• The review date, and only that date.

The specific mistake, and it is nearly universal on a first attempt: writing a policy that specifies a feeling. “I will stay calm and not react to short-term noise” is not a policy, because it has no trigger, no action and nothing that can be observed to have happened. A policy specifies an action, a trigger and a date, and if a sentence has none of the three it is a mood board.

The second failure mode deserves a name, because it kills more policies than anything else. A document containing the clause unless conditions are exceptional has already voided itself: every crisis satisfies its own exception, and the clause is read for the first time on precisely the day it lets you through. Call it the escape hatch. If you need room to deviate — and sometimes you genuinely do — the room should cost something rather than be conditional on something. A rule that says any deviation requires a written reason dated at least 48 hours earlier does not prohibit anything. It simply prices the deviation in the one currency that the moment is short of.

Rebalancing rules decided in advance

Rebalancing — selling what has grown past its target share and buying what has fallen below it — is the decision where a rule most obviously beats a judgement, because it asks you to sell the thing that is working and buy the thing that is not.

Two rules do the job, and both convert the question into arithmetic.

• A calendar rule — check on a stated date each year and correct back to target.
• A band rule — act when a holding drifts more than a stated number of percentage points from its target, and otherwise leave it alone.

Both remove the part that stress corrupts, which is not the arithmetic but the timing: neither rule contains a step at which anyone decides whether now feels right. The mechanics of each, and the tax and cost consequences of the choice between them, are a subject in their own right — what matters here is only that the choice gets made in advance.

The cost is real and worth stating. A band rule will sometimes trade when sitting still would have been better, and it books gains and their tax on a schedule that has nothing to do with whether booking them was wise. The rule is not claiming to be optimal. It is claiming to be executable in the third week of a bad March, which the optimal policy is not.

Finding it in your own record

Whether any of this describes you is not a question anyone can answer forward, and an article that tells you how you will behave in the next fall has overstepped what it can know. Backwards, though, it is answerable from data you already have.

Export every buy and sell you have made, with dates, and put the index level or the scheme NAV beside each one. Then ask three questions of the list.

1. Do the deviations from the plan cluster, or are they spread evenly through the years?
2. What was happening in the market in the two weeks before each one?
3. How long after each exit did the money go back in — and at what level?

The tell is not that you deviated but whether the deviations cluster. A handful of changes spread across 10 years is a portfolio being managed. Four of them inside one quarter, all in the same direction, is the market managing the portfolio, and the record says so without anyone having to make a claim about your temperament.

This is also the honest reason a third party sometimes helps, and it is not that they know more: an adviser sits outside the moment, which is a structural position rather than a skill, and it is the same position a written policy occupies for free. What that is worth, and what it costs, is a question with an arithmetic answer.

What structure does not fix

Three limits, because a technique presented without them is being sold rather than explained.

A rule can be wrong, and structure makes a wrong rule durable too. The same machinery that stops a good plan being abandoned in a fall will keep a badly sized contribution or a mismatched allocation running for years without ever printing a complaint. Automation is a multiplier on the quality of the decision it automates.

The pressure does not go away. Rules do not remove the feeling; they remove its access to the account, which is a smaller claim than most articles on this subject make and the only one that survives contact with an actual bad week.

And none of it addresses a genuine cashflow problem. Someone whose income has stopped is not experiencing a behavioural difficulty and does not need a policy document — they need liquidity, which is the entire job of an emergency fund. The reason a buffer belongs before any of this is that it stops a market decision from being forced by a household one, and a forced decision is not improved by having a rule about it.

Testing the rule you actually wrote

A policy written on a calm evening makes a claim about a bad year, and that claim is checkable against history rather than argued about.

The useful figure is not the return. It is the worst peak-to-trough fall along the path and how long it lasted, because that is what the rule would have had to survive. FNOTrader’s Mutual Funds app runs a contribution schedule against the published NAV history from AMFI, the mutual fund industry body — around 34 million NAV rows — reporting the internal rate of return for cashflows on irregular dates (XIRR) alongside maximum drawdown, and rolling returns across every available start date rather than the one that happens to end today.

Run the instalment you have written down through the worst window your scheme has on record. If the path is one the policy would not have survived, the policy needs changing on the calm evening rather than during the fall.

Past performance is a record of what happened, not an indication of what will. FNOTrader is not a SEBI-registered investment adviser and this is not investment advice.

Common questions

Continue reading

More in Behavioural Finance · App: Mutual Funds · Definitions: glossary · Free tools: calculators · All: every article