What a sunk cost is
A sunk cost is money, time or effort that has already been spent and cannot be recovered whatever you do next. The ticket is bought, the hours are played, the stake is lost. Because these costs are the same for every option still open, they carry no information about which option is better.
The sunk cost fallacy is the habit of letting them influence the decision anyway. People continue with something because of how much they have already put in, not because of what continuing will bring from this point forward.
A non-gambling example first
Suppose someone paid a non-refundable 30 units for a ticket to an outdoor event. On the day it is raining, they feel unwell, and they would honestly prefer to stay home. Two things are true: the 30 units are gone either way, and going will make the evening worse than staying in. Choosing to go “so the money isn’t wasted” adds a miserable evening to the loss. The money was lost when the ticket was bought. The only real question is which choice, from here, is better.
The same shape appears in finishing a dull game because many hours have already gone into it, holding a failing purchase because of what it cost, or continuing a project that no longer makes sense.
How it shows up in games of chance
In gambling, sunk cost thinking often sounds like “I’ve already put so much in” or “I can’t stop now, I’d lose everything I’ve spent”. Each version treats the earlier losses as a reason to continue.
But the next play does not know about the earlier ones. Its odds are the same whether the last hour was kind or unkind. If a game has negative expectation, then each additional stake has an expected loss, however much has gone before. The earlier losses cannot be reduced by continuing; they can only be added to.
A small worked example
All numbers here are made up. A person has lost 200 units and feels a strong pull to continue. They have 50 units left in their spending limit for the evening and think: “If I double up twice, I’ll be back to even.”
Suppose each play is an even-money coin flip that is fair, and the plan is to stake everything each time until the pool reaches 200 or hits zero:
- 50 becomes 100 if the first play wins (probability 1/2).
- 100 becomes 200 if the second also wins (probability 1/2).
- Chance of reaching 200: 1/2 × 1/2 = 1/4, or 25%.
- Chance of losing the extra 50 as well: 75%.
And that is in a fair game, which real games are not. If each win has probability 0.45 instead, two wins in a row has probability 0.45 × 0.45 = 0.2025, about 20%, so the chance of losing the extra 50 rises to roughly 80%. The earlier 200 is not a reason to take that bet. The decision should rest on what the new 50 units face, which is an average loss and a likely further one. This pattern of staking more to recover what is gone is the usual story behind chasing losses.
Why the pull feels so strong
Several ordinary tendencies feed it:
- Loss aversion. Losses tend to feel larger than equal gains, so people try hard to avoid “locking in” a loss, even though the loss has already happened.
- A wish to be consistent. Stopping can feel like admitting a mistake, and continuing postpones that feeling.
- Mental accounts. A session or a month may be treated as one account that must be balanced before it can be closed.
- Hope of a turnaround. A streak of bad luck can seem like it must reverse, which confuses a run of results with something owed. This links to the way risk perception is shaped by recent experience.
Questions that separate past from future
A few practical questions can help move attention to the future costs and benefits:
- If I had not spent anything so far, would I choose to start this now?
- What does the next stake cost, and what is the average result of it?
- Would I give the same advice to a friend who had not been through the earlier loss?
- What else could the time and money do?
In terms of informed decisions, the useful framing is simple: list the options available now, compare their future costs and benefits, and ignore what has already gone. Thinking in terms of the expected value of each option from this moment is a clean way to do it, and so is sketching one of the simple decision trees used for uncertain options.
Sunk cost and the size of a pool
The idea ties closely to how a fixed pool is meant to work. The pool is decided in advance, and money lost from it is gone, not owed back by future plays. Topping up after a loss, borrowing, or extending a session to “make it back” all treat the sunk amount as an obligation. A limit set beforehand protects against this by ending the decision before the feeling arrives.
Not every investment of effort is a trap. Sometimes continuing is right because of the future payoff, such as finishing a course that will be useful. The test is whether the future benefits justify the future costs by themselves.
If it is hard to stop
Feeling that you must recover losses is a recognised warning sign. Anyone who is gambling to win money back, hiding losses or borrowing to play can contact a local problem gambling support service. Outcomes in games of chance are uncertain, and there is no play that is owed to you.