Revenge trading, FOMO and the answer that comes right after a wrong one

Trading FOMO — the fear of missing out — and revenge trading are two of the most discussed experiences in trading, and two of the least measured. Almost everything written about them is description: you should not chase, you should not double down after a loss, you should stay disciplined. All true, all unfalsifiable, and none of it tells you whether you do it.

This piece is about the other half. What does the research actually say the mechanism is, what a famous case shows about who is vulnerable, and — the part that is ours — whether any of it can be turned into a number.

What revenge trading, FOMO and tilt mean here#

Trading FOMO — fear of missing out — is the pressure to act because an opportunity looks like it is disappearing without you. Revenge trading is making the next decision under the influence of the last one, usually a loss. Tilt, as we use the word in this piece, is narrower than either: it is whether the quality of your next decision changes after you get one wrong.

That third definition is ours, not the literature's, and the distinction matters. The first two describe a feeling. The third describes something that can be counted.

What the research says FOMO actually does#

The closest thing to a formal account of trading FOMO is not about emotion at all. It is about attention.

Brad Barber and Terrance Odean studied which stocks individual investors buy, and their finding was that individuals "are net buyers of attention-grabbing stocks" — the ones in the news, the ones with unusual volume, the ones that just made an extreme one-day move.

The mechanism they propose is the useful part. There are thousands of stocks you could buy and only a handful you could ever evaluate, so attention decides which ones enter the running at all. Selling does not have this problem, because you can only sell what you already own. Their summary of it is one sentence: preferences determine choices after attention has determined the choice set.

That reframes FOMO usefully. It is not that the feeling makes you choose badly. It is that the feeling changes what is on the menu, and then you choose normally from a menu that something else wrote. By the time the decision feels like yours, the interesting part has already happened.

Which also explains why "just be disciplined" is such weak advice. Discipline operates on the choice. The distortion happened before the choice.

The cautionary case is not the one you expect#

The usual figure in this story is a famous trader who blew up. We would rather use Isaac Newton, for a specific reason: he was right first.

Newton was, per the archival work of Andrew Odlyzko, among the minority who judged early in the South Sea mania of 1720 that it would end badly, and he liquidated at a large profit — around £20,000, against a net worth shortly before the bubble of somewhat over £30,000. He had called it, and he had banked it.

Then the bubble kept inflating. Odlyzko's account is that Newton "yielded to the prevailing groupthink and jumped back in, almost at the very top". By mid-1721 his net worth was down to about £20,000. The popular record, as summarised on Wikipedia's page for the South Sea Company, has him holding nearly £22,000 of the stock in 1722 and losing at least £10,000.

The line everyone quotes — that he could calculate the motions of the heavenly bodies but not the madness of people — reaches us second-hand, reported decades later by Joseph Spence, and Odlyzko himself writes that Newton "supposedly said" it. Treat it as folklore that happens to be apt.

What is not folklore is the shape of the mistake, and it is the shape that matters here:

1. He analysed correctly.
2. He acted on his analysis and profited.
3. He then watched everyone else go on growing richer without him.
4. He re-entered near the top.

Step 3 is the whole story. Nothing about his reasoning failed. What failed was sitting still while the choice set was being rewritten around him by other people's returns. If the most formidable analytical mind of his century could not sit still through that, "be disciplined" is not a plan.

Can revenge trading be measured?#

Here is where we stop quoting other people and describe our own.

SwipeTA shows you a real historical intraday chart frozen at a decision point and asks which way price went next. It is practice rather than trading — no real money, no broker, no orders. But because every call is recorded, some of the behaviour people describe in trading has a definition here.

Tilt has a definition. It is the accuracy of the answer that comes immediately after a wrong one, set against the accuracy of the answer that comes immediately after a right one. That is what our statistics service computes, and it is deliberately narrow: not "how do you feel after a loss", but "what happened to the very next decision".

That statistic has a hard prerequisite that is worth stating because it is a genuine engineering constraint rather than a design preference. Computing it requires knowing the order the answers were given in. Our attempts table originally recorded which questions you had answered, not the sequence; an ordering column had to be added later, and everything recorded before it exists cannot be repaired. There is no way to reconstruct "what came right after what" from data that never stored it. Tilt, for us, starts from the day we started keeping the order.

Commitment has a definition too. In the rating mode, a call is not just a direction: you also commit how much of your standing rests on it. So the record contains a natural split — your accuracy on the answers where you committed more than the base amount, against everything else. That is one of the few places where the phrase "I was sure about that one" becomes checkable rather than remembered.

Hesitation has a definition. Time to answer is recorded, so the record can separate the calls you made in under a second from the ones you sat on, and score each group separately.

None of these are emotions. They are traces that certain emotions tend to leave.

How many answers does it take to detect a trading bias?#

Having built all of that, the app then refuses to show you most of it for a while.

The question bank is a deliberate 50/50 split between up and down, so a direction bias is measurable against a known baseline. But "measurable" needs a sample. Detecting a 65/35 lean with 80% power takes about 85 decided answers. A 60/40 lean takes about 194. A 55/45 lean — which is still a real lean — takes 783.

So the direction-bias panel stays locked until 85 decided answers, and shows a countdown instead of a number. That threshold is not a product instinct dressed up as statistics; it is where the statistics land, and the constant in the code is 85 for that reason.

This is the trade we are least sure about, so it is worth being explicit. A number shown at answer 20 would be more engaging and would be noise. A feature whose entire value is that its read of you is trustworthy cannot afford to be wrong early, because the first thing a reader does with "you lean long" is believe it. We would rather show a locked panel than a confident number we cannot stand behind.

What this does not claim#

It would be easy to end by saying that practising here will make you less prone to FOMO in a real market. We are not going to, because we have no evidence for it and neither does anyone else.

No study we know of shows that noticing a behavioural pattern in a simulation changes that behaviour where money is involved. The transfer question is open, and we said the same thing in our piece on what the research says about trader intuition. SwipeTA is a training game and a simulation: it holds no money, nothing in it is a prediction about a real market, and it does not provide investment advice.

What it can do is narrower and, we think, still worth something: it keeps a record you did not write from memory. Whether your next call after a wrong one is worse than your others is a fact about you that either shows up in the data or does not — and unlike the story you tell yourself afterwards, it cannot be quietly edited once the outcome is known.

Newton did not lack discipline, and he did not lack intelligence. What he lacked, standing there in 1720 watching everyone else grow richer, was a record of what he had already been right about. The rest of the parameters behind our own numbers are on the methodology page.