Regret: Anticipating It, and What a Past Trade Does to the Next

Regret is the cost of learning that a different choice would have turned out better. People expect it, so regret acts on a choice long before it is felt. Anticipated regret pushes a decision towards whatever will be easiest to justify later, and the easiest thing to justify is not always the option with the better odds.

Regret rests on one property that separates it from every other cost in this subject. Regret needs two terms, never one. There is the outcome that actually arrived, and there is the outcome the other choice would have produced, and regret is the distance between them. However poor the result was, take away the second term and the regret goes with it. The two term property is stranger and more useful than it first looks. Regret can be raised or lowered without touching the odds, the amounts or the outcome, purely by changing what a person will find out afterwards. Nothing else in this part of behavioural finance carries a lever like that.

What is regret, as a cost rather than a mood?

Ordinary speech treats regret as a feeling, somewhere between disappointment and embarrassment. A feeling arriving after everything is settled cannot change anything, so the everyday reading is not wrong, only useless for decisions. Decision theory uses the word more narrowly. Regret is a cost paid at the moment a person learns what the other road produced. Everybody knows in advance that they may have to pay it, so the cost is already sitting inside the decision.

Unusually, two economists arrived at that formulation independently in the same year. Bell, writing in Operations Research in 1982, and Loomes and Sugden, writing in the Economic Journal in 1982, both proposed that a person choosing between options is not scoring each option on its own outcome alone. Each option is scored on its outcome plus a term for how that outcome will compare with what the rejected option turned out to do. Once the comparison term is inside the score, a choice can be perfectly sensible on the outcomes alone and still lose to a rival option that will be easier to live with.

The effect shows outside money first. A traveller at a railway station faces two ticket queues, picks the left one, and spends four minutes watching it crawl while the right one clears. In a second version a wall hides the other queue. Same wait, same train. The second version costs the wait; the first costs the wait plus something extra, and the extra is not imaginary.

Why does regret need a comparison to exist at all?

A poor outcome on its own produces disappointment. Regret needs a second term beside it: a counterfactualThe version of events that would have followed a choice nobody actually made., a clear picture of what the rejected option went on to do. Without that picture the arithmetic has nothing to subtract, and a bad outcome stays a bad outcome rather than becoming a mistake.

The two term requirement is why regret is not simply a bigger version of loss. A loss is measured against nothing but itself. Regret is measured against a forgone outcomeWhat the option that was not taken would have produced., so its size depends on something outside the decision entirely: whether anybody ever finds out. Regret is the only such cost a person can shrink or enlarge without changing a single fact about the choice itself.

Take a household choosing between two schools. The child does reasonably well. If nobody there ever learns how the other school fared, the outcome is simply the outcome. If a neighbour's child went to the other one and the results come up at every gathering for eleven years, the same outcome now carries a commentary. Nothing about the education changed, only the availability of the second term.

What is Anticipated Regret, and how does it change a choice before anything happens?

Anticipated regret is regret that has not happened yet, and may never happen, doing work on a decision today. A person weighing two options runs a short private film of each one going wrong, notices that one film is much harder to watch, and moves towards the other option. No outcome has occurred and no comparison has been made. The cost is entirely expected, and it is already deciding things.

Being expected in advance is why regret earns a place in a serious account of choice rather than a footnote about feelings. A cost that only arrives afterwards is a matter for consolation, not for analysis. A cost forecast in advance sits in the same column as dealing charges and tax, and leaving it out is not tough mindedness but an incomplete sum. Anticipated regret is a forecast, and like every forecast it can be wrong in both directions. A forecast deserves examining rather than either obeying or dismissing.

Watch how it works on Meera Sundaram at Palash Advisory Services Private Limited on 12 October. She is looking at Suvarna Chemicals Limited, standing at Rs 4,60,000/- against a cost of Rs 4,00,000/-. Two films run. In the first she sells, books Rs 60,000/-, and the holding then climbs without her. In the second she holds, and the Rs 60,000/- that was sitting there evaporates. A gain that was available and then vanished is the sort of thing a person tells themselves about for years, so the second film is the one she cannot watch. So she sells. Notice what has not entered that reasoning anywhere: any view about what Suvarna Chemicals Limited will do next.

Why does a decision actively taken hurt more than one that was merely allowed to happen?

Hold two situations side by side. In the first, a person switches out of a holding and it then does well without them. In the second, a person considered switching into that same holding, did not, and it then did well without them. The money involved can be made identical. The regret is not. Regret attaches far more sharply to something a person did than to something the same person merely failed to do, and the gap has nothing to do with the size of the cost.

Behind the gap is a difference in material: an action regretRegret for something a person did, which is felt more sharply than regret for something they left undone. has everything a reproach needs and an omissionSomething a person did not do, as opposed to something they did. has almost none of it. The act has a date and a moment when something was decided. The act leaves a line in a record. An act departs from how things stood, so it fits in one sentence beginning with the word I. An omission has no date, no moment and no line, and describing it takes a paragraph about what somebody was thinking over several weeks. The material for a reproach is simply not there in the same quantity.

The same asymmetry sits outside money everywhere. A driver who leaves the usual route for a shortcut and hits a jam thinks about that turning all evening. A driver who stayed on the usual route and sat in the same jam is annoyed and forgets it by dinner. Same lost hour. One of them made a turn.

What is Trading Regret, and which decisions does it attach to?

Trading regret is regret that lands on a decision to buy or to sell. Trading regret does not spread itself evenly over a holding. The concentration falls on the decisions that have a date and a price beside them, and largely leaves alone the far larger number of moments when somebody did nothing in particular.

Lump the four shapes together and the loudest one gets mistaken for the biggest one, so they are worth separating. There is selling something that then rose. There is buying something that then fell. There is holding something that then fell further. And there is not buying something that then rose. The first two are acts with a moment attached, the second two are stretches of time in which nothing was decided, and regret attaches to the first pair far more sharply than to the second, whatever the money says.

The concentration matters for a plain reason. If regret is louder on acts, then anything done to reduce it will look like doing less, and doing less is not automatically the same as deciding better. Naming the four shapes is what stops the loudness of a feeling being read as a measurement of a cost.

What did the pair of 12 October actually cost, and which one stings?

The Palash decision log carries one pair of decisions taken on the same afternoon, and between them they isolate the effect almost perfectly. On 12 October Meera Sundaram sold Suvarna Chemicals Limited whole at Rs 4,60,000/-, against a cost of Rs 4,00,000/-, so Rs 60,000/- was realisedTurned into cash by selling, so a gain or a loss stops being on paper and becomes final.. On the same afternoon she kept Kesari Logistics Limited, then standing at Rs 1,95,000/- against a cost of Rs 3,00,000/-, saying she would sell it when it got back to Rs 3,00,000/-. One act, one thing left undone.

On 31 March the log was read back. Suvarna Chemicals Limited had risen 8.0 per cent since the sale, so the holding she no longer had would have stood at Rs 4,96,800/-, and Rs 36,800/- had been forgone. Kesari Logistics Limited had fallen a further 20.0 per cent, from Rs 1,95,000/- to Rs 1,56,000/-, so Rs 39,000/- more had gone. The whole point sits in how close the two land, so work both sums yourself first.

Rs 36,800/- against Rs 39,000/-. The two costs are Rs 2,200/- apart. The gap is 5.6 per cent of the larger cost, and the larger cost is the holding she left alone. The decision that cost more produced almost no regret, and the decision that cost less is the one she still talks about. Cost and regret have come apart about as cleanly as a single case can show. One pair of decisions cannot show how often the ranking reverses. It can only show that the money and the sting are two different quantities.

What happens to regret when the road not taken can never be checked?

Everything above depends on one thing that is easy to miss: on 31 March somebody read the log back. Had nobody ever looked up what Suvarna Chemicals Limited did after the sale, the Rs 36,800/- would not exist as a felt cost at all. The Rs 36,800/- would still be a fact. Nobody would know it, and an unknown comparison costs nothing. Regret is manufactured by feedbackFinding out afterwards what the option that was turned down actually went on to do., not by the outcome. Two people with identical results can therefore carry completely different costs.

Since people know this about themselves, the chance of finding out starts changing choices in advance. Offer somebody two options and tell them they will never learn how the rejected one turned out, and the unusual option becomes noticeably easier to take. Tell them the result will be posted on a board next month, and the ordinary option starts looking better without a single number about it having changed. The slider below holds both options completely fixed and moves nothing except how likely it is that the comparison ever becomes available.

What does a past trade do to the next one?

A decision does not end when it settles. A settled decision leaves behind a revised estimate of how bad the comparison feels for that kind of act. After 31 March, Meera Sundaram has learned something specific: selling a holding that then rises produces a number she has to look at. The next time a similar sale is on the table, the expected cost of that comparison is already higher than it was on 12 October, for reasons with nothing to do with the holding in front of her.

The carry over attaches to the kind of act, not to the situation. It survives a change in every fact that ought to matter. The carry over is worth being exact about. It is easy to call it learning and leave it there. Some of it is learning. If the reason for the sale was weak, noticing that afterwards is genuine improvement. But the raised cost applies just as strongly when the reasoning was sound and the outcome simply went the other way, and in that case nothing has been learned about anything. The estimate moved because a comparison was painful, not because evidence arrived.

Watch it outside money. A cook who once changed a recipe for a large gathering, and had it fall flat, hesitates over the next change even when the dish and the guests are entirely different. The hesitation is real and it is doing real work. None of it is evidence about this dish.

When do the defensible choice and the best choice come apart?

Here is the sharp edge of the whole subject. Anticipating regret does not push towards the safe option. The pull is towards the defensible choiceThe option that is easiest to justify to somebody afterwards, which is not always the one with the better odds., meaning whichever option will be easiest to explain if it goes wrong. Most of the time the defensible choice and the sensible choice are the same thing. The overlap is why the pull rarely causes trouble and why it is so hard to spot when it does.

The two separate at one particular moment: when the option with the better odds is an unusual one. Everybody in the room would have done the same, so an ordinary choice that fails gets an ordinary reaction. An unusual choice that fails gets a very different reaction, and the reaction is not about the odds. The four possible reactions to a decision are sorted by whether the choice was ordinary and whether it worked. The odds at the time enter none of the four, and their absence is precisely what makes the pull towards defensibility so difficult to argue with.

Somebody deciding for another person feels the pull twice over. The explaining is not hypothetical: a real conversation is coming, and it will be about the result. Do not file this as a problem of professionals, though. A person deciding alone still has that conversation, with a partner or with themselves at two in the morning, and it goes the same way.

What can be written down before the outcome that still counts afterwards?

The correction is a record made in advance. Not a forecast and not a promise, but a written statement of what would make this decision a sound one whatever it goes on to do. Writing it takes about ninety seconds, and it changes what is available to say later. Nothing else about the decision was ever wrong.

Think about why that matters. When a decision is questioned after a poor result, the person is reaching for a defence, and if nothing was written down the only defence available is the result itself. The result is the very thing being complained about, so defending with it is a losing position by construction. Judging a decision by how it turned out has a paper behind it: Baron and Hershey, in the Journal of Personality and Social Psychology in 1988, showed the same decision being rated differently depending only on what followed it. A record written before the outcome is the only defence that does not depend on the outcome, and a defence of that kind is what resolve can never supply.

The Palash decision log shows how unusual the written standard is in practice. Across the 240 logged decisions, a written reason appears on 84 of them, or 35.0 per cent. From 4 November, 20 of the 60 investors adopted a written checklist, and over the next four quarters they recorded a reason on 34 of their 41 decisions, being 82.9 per cent, against 19 of 63, being 30.2 per cent, for the other 40. The 82.9 per cent against 30.2 per cent measures process and nothing else. A group of 60 people over eight quarters is far too small to show a difference in returns, so no such difference can be read out of the figures.

Where is anticipating regret the correct thing to do?

Half the value of the idea is lost when regret is presented as a flaw to be stamped out. Anticipating regret is correct whenever the regret is a genuine future cost that the person will actually bear. If somebody will spend two years unable to look at a statement without a particular thought arriving, that is not a bias. The discomfort is a cost, it lands on a real person, and a calculation that leaves it out has understated what the option costs.

The test is short. The question is whether the discomfort is something the decider will carry, or only the awkwardness of having done something unusual in front of people who will have forgotten by March. The first belongs in the sum. The second does not, and putting it there buys a worse option with real money. Anticipated regret is a distortion only when it selects a worse option purely because that option is easier to explain, and it is accurate accounting in every other case.

There is a second place where it is plainly correct. Where a poor outcome would force a change in how somebody lives, anticipating it is not squeamishness. Meera Sundaram holds a reserve of Rs 1,10,000/- against a monthly outgo of Rs 55,000/-, or two months of it. A decision that could put that reserve at risk carries a consequence no odds calculation describes on its own.

How does an adviser, or somebody deciding alone, put this to use?

Devika Rao at Palash Advisory Services Private Limited does not sit in front of a client and say the words anticipated regret. She separates two questions clients arrive with fused together: what is the better option here, and what will this feel like in eight months. Both are real and both belong in the conversation. The damage happens when the second answers the first without being asked to.

The practical move is small. Before the decision is taken, she writes down the standard by which it should be judged, in the client's own words, and reads it back. The note gives her something to open on 31 March that is neither a valuation nor an apology. The written standard turns the later conversation from a discussion about the result into a question about whether the standard was met, and those are different conversations. A lender does the same from the other side: a file recording why an advance was made and what would have counted as a warning can be reviewed after a default without collapsing into hindsight.

Nobody else is in the room to ask the awkward question, so a person deciding alone gets the same benefit from the same ninety seconds, and arguably more. The expected valueThe average of the possible outcomes, each weighted by how likely it is to happen. of an option can be worked out on paper afterwards. The reasoning the decider had at the time cannot, unless it was written down at the time.

Not every decision earns ninety seconds of writing, and pretending otherwise is how a good habit dies in three weeks. Two things decide it: how much the decision matters, and how visible the forgone outcome will be. One that barely matters and that nobody will check needs nothing. One that matters and whose alternative will sit on a screen every morning is where the note pays for itself.

What does regret not explain on its own?

Regret explains why a cost that has not happened can change a decision, and why two decisions with almost the same price tag leave completely different marks. Regret does not explain how much a loss hurts relative to a gain of the same size, a separate measurement on a separate curve. Nor does it explain why a purchase cost gets held in mind as the point everything is measured from. And it says nothing about what any holding was worth.

One more distinction matters. The claim is not that people decide in order to avoid regret, as a conscious aim. The claim is narrower and better supported: the expected comparison is one of the inputs, it can be large, and it varies with something that has no business varying a decision, namely how likely it is that anybody finds out.

Meera Sundaram, Devika Rao, Palash Advisory Services Private Limited, the Palash decision log, Suvarna Chemicals Limited and Kesari Logistics Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.