Trauma & Impulse Notes
Gambling behaviour

Why the Wins Stay in Memory and the Losses Fade

Ask someone how they have done over the past year and you get a story. Ask the bank and you get a column of figures. The two rarely agree, and they disagree in a predictable direction: the remembered version is kinder.

Two records of the same year

This is not dishonesty. Memory was never built to keep ledgers. It keeps what was vivid, and it keeps what was retold, and neither of those is the same as keeping what happened most often.

How availability does the arithmetic

When a question is hard to answer properly, people quietly answer an easier one instead: how readily do examples come to mind? Amos Tversky and Daniel Kahneman described this shortcut as the availability heuristic, and most of the time it serves well enough, because common things really are easier to recall than rare ones. It fails whenever something is memorable for a reason other than frequency.

A large win is exactly that. It arrives with a jolt of feeling, it has a shape you can describe — the hand, the number, the moment — and it gets retold over dinner. A hundred small losses have none of those properties. They are indistinguishable from one another, they happen in the background of an unremarkable evening, and nobody repeats them to anyone. So when the mind asks how often this tends to go well, the wins are already standing at the front of the queue, while the losses, which outnumber them, cannot be counted at all because they were never filed as separate events.

Three other things memory does with a gambling record

It keeps score for the system rather than against it. Anyone playing with a method — a favoured number, a staking pattern, a time of day — is well placed to notice the occasions when it worked and to find a reason to set aside the occasions when it did not. Confirmation bias requires no deliberate effort. It is simply what happens when evidence that fits gets filed and evidence that does not gets explained.

It promotes near-misses. A round that stopped one symbol short is remembered as a round that nearly paid, which files it closer to a win than to the loss it actually was. The feeling behind that is covered separately in the note on the near-miss effect.

It hears mostly from the winners. The gambling stories in circulation are the ones worth telling. People who lose steadily tend to say nothing at all, so the pool of anecdotes you have absorbed since childhood has had its failures quietly removed before you ever saw it. Survivorship bias is the general name for reasoning from a sample that only the survivors entered.

Why the sum is worse than it feels

Take a purely illustrative case: twenty short sessions, three of them ending ahead by a noticeable amount, seventeen ending slightly down. The three are episodes with details attached. The seventeen collapse into a single vague impression of ordinary evenings. Asked afterwards how the twenty went, the honest answer from memory is "mixed, about level" — and it can be the honest answer even when the arithmetic says otherwise. The figures here are an illustration of the shape of the error, not a measurement of anything.

The only correction that works is written

You cannot out-remember a biased memory by trying harder to be fair. The bias operates on what was stored, so the correction has to happen before storage, outside your head.

  1. Log every session, including the short dull ones. Date, amount taken in, amount left with. The dull ones are the whole point, because those are the entries memory will not keep.
  2. Write it at the time, not later. A figure reconstructed at the end of the month has already been through the process described above.
  3. Add up months, not sessions. A single evening tells you nothing either way. A running total across twenty or thirty sessions is the first number that means something.
  4. Guess before you look. Write down what you think the total is, then total it. The gap between the two is the most useful thing the log will ever tell you, and it is worth repeating every few months.
  5. Count the time as well as the money. Hours are spent even on evenings that end level, and they are the part no statement records.

What the figure is for

A written total is not a scolding device, and it is not there to prove that any particular evening was a mistake. It exists so that decisions about money and time rest on a record instead of on a highlights reel. Deciding the limits in advance, as set out in the note on money, time and stopping, is much easier when you know what the last three months actually came to.

If the log is being avoided, rounded off or quietly abandoned, that reluctance is itself worth noticing, and it matters more than any single entry in it. See the diagnostic criteria and help and resources.

Last updated 2026-11-20 · Trauma & Impulse Notes editors