The boiled frog theory says that people fail to react to a threat when it builds slowly, and react quickly when the same threat arrives all at once. In crypto, this shows up when a position slides a few percent at a time, warning signs pile up one by one, and each new low starts to feel normal. Finance research backs the core idea. A 2014 study in The Review of Financial Studies found that investors underreact to information that arrives in small, continuous steps compared with the same news delivered in one jump. The frog story itself is biologically wrong. The human pattern it describes is real and measurable.
Where the metaphor comes from, and why the frog part is wrong
The story is rooted in 19th-century physiology. In 1869 Friedrich Goltz found that a frog with its brain removed stayed in slowly heated water, while an intact frog tried to escape when the water reached about 25 °C. In 1872 Heinzmann reported that normal frogs did not try to escape when water was warmed slowly, at under 0.2 °C per minute over 90 minutes.
Modern biologists reject the popular version:
- Douglas Melton of Harvard said in 1995: "If you put a frog in boiling water, it won't jump out. It will die."
- George R. Zug of the National Museum of Natural History: "If a frog had a means of getting out, it certainly would get out."
- Victor H. Hutchison, a zoologist at the University of Oklahoma, said in 2002: "The legend is entirely incorrect!" He described research in which frogs grew more active as water warmed by about 1 °C per minute and jumped out if they could.
So the metaphor is useful for people, not frogs.
What the research says about gradual change
People miss changes that happen slowly
In psychology, change blindness is the failure to notice a change in what you are looking at. Early experiments only produced it when a brief disruption, such as a blank screen, hid the moment of change. Simons, Franconeri and Reimer (2000, Perception) showed that it also happens when a change unfolds gradually in plain view. A 2006 follow-up in Psychologica Belgica had faces change expression gradually. Only 15% of observers noticed the change within a single trial, and gradual changes were detected three times less often than changes with a disruption.
A price chart moving a little each day works the same way. Each day looks almost like the one before.
Markets underreact to continuous information
Zhi Da, Umit Gurun and Mitch Warachka titled their 2014 paper "Frog in the Pan: Continuous Information and Momentum" (The Review of Financial Studies, vol. 27, no. 7). Their hypothesis is that gradual, frequent changes attract less attention than infrequent, dramatic ones. In the published version, six-month momentum fell from 5.94% for stocks whose news arrived continuously to -2.07% for stocks with discrete news, at similar cumulative returns. The continuation did not reverse in the long run. Investors were slow to price in news that came in small doses.
Ben Carlson's 2016 post at A Wealth of Common Sense makes the same link for a general audience. Drawing on Wes Gray and Jack Vogel's Quantitative Momentum, he argues that investors underreact to gradual information and overreact to dramatic moves.
The arithmetic of slow losses
Small losses compound into large ones:
| Pattern | Cumulative loss |
|---|---|
| One 30% drop | 30% |
| Twenty consecutive 5% drops | about 64% (1 − 0.95^20) |
| 2% per week for 26 weeks | about 41% (1 − 0.98^26) |
| One 50% drop | 50% |
A single 30% crash feels more alarming than twenty 5% declines, but the slow path costs more than twice as much.
The biases that keep people in the pot
Noticing a decline late is only half of it. Several well-studied biases then make it harder to act.
Loss aversion
In prospect theory, losses weigh more heavily than equal gains. Tversky and Kahneman's 1992 paper in the Journal of Risk and Uncertainty estimated a loss-aversion coefficient of 2.25: a loss feels roughly twice as heavy as a gain of the same size.
The disposition effect
Shefrin and Statman (1985, Journal of Finance) named the tendency to sell winners too soon and hold losers too long the disposition effect. Terrance Odean's 1998 study of about 10,000 discount-brokerage accounts found a clear preference for selling winners over losers. Rebalancing and trading costs did not explain it, and later returns did not justify it.
Sunk cost
Arkes and Blumer (1985, Organizational Behavior and Human Decision Processes) found that people keep going with something because they have already put money, effort or time into it. In their field study, theatre-goers who had paid more for season tickets attended more plays. In questionnaires, people with a sunk cost rated a project as more likely to succeed.
| Bias | What it does | How it looks in a crypto position |
|---|---|---|
| Gradual-change blindness | The slide goes unnoticed | "It's only down a bit this week" |
| Loss aversion | Taking the loss feels too painful | Waiting to "get back to even" |
| Disposition effect | Winners get sold, losers kept | Trimming a winner to fund a loser |
| Sunk cost | Past spending justifies staying in | "I've put too much in to sell now" |
These biases often travel with FOMO, which pulls people in near a peak, and with rug pulls, where hype is kept up on purpose until the exit.
How much do small drops add up to?
Enter the size of each drop and how many happen in a row to see the total loss, compared with one single drop.
| Size of each drop | Number of drops in a row | Cumulative loss from the starting value |
|---|---|---|
| 5 % | 20 | 64.2 % |
| 2 % | 26 | 40.9 % |
| 30 % | 1 | 30 % |
| 50 % | 1 | 50 % |
- Each drop is measured from the value left after the previous drop, so losses compound and do not simply add up.
- Drop sizes and counts are your own inputs, not figures from any source or a prediction for any asset.
This is compounding arithmetic on your inputs only: it assumes no rebounds between drops, leaves out fees, leverage and liquidation, and cannot tell you whether a decline will continue or what to do about it.
Three crypto cases of slow heat, sudden boil
Terra (May 2022)
According to the SEC's February 2023 complaint against Terraform Labs and Do Kwon, UST was marketed as a "yield-bearing" stablecoin paying up to 20% through the Anchor Protocol. The SEC alleges that investors were misled about UST's stability. In May 2022, UST lost its dollar peg, and UST and its sister tokens fell close to zero. On CoinGecko, LUNA (now Terra Luna Classic) shows an all-time high of $119.18. The high yield was visible for months before the break, and it was the kind of sign that is easy to get used to.
A major exchange failure (November 2022)
A large offshore crypto exchange collapsed in November 2022. Its founder, Samuel Bankman-Fried, was convicted of fraud and conspiracy and sentenced on 28 March 2024 to 25 years in prison, with $11 billion in forfeiture. Announcing the sentence, the U.S. Department of Justice quoted U.S. Attorney Damian Williams, who said he had stolen "over $8 billion of his customers' money". The DOJ also said he defrauded investors of more than $1.7 billion and lenders to Alameda Research of more than $1.3 billion.
The October 2025 liquidation cascade
On 10–11 October 2025, crypto markets recorded what was widely reported as the largest liquidation event on record. CoinGlass-based figures cited by several outlets put it at about $19.16 billion, roughly 87% of it in long positions. Estimates vary between sources, and some argue the true total was higher. Reports tie the sell-off to a U.S. announcement of new 100% tariffs on Chinese imports. Coverage also describes leverage building up for weeks beforehand.
Habits that counter gradual drift
These are general risk-control habits that respond to the research above. They are not recommendations to buy, sell or hold anything.
- Write the exit condition before you enter. Studies of the disposition effect and loss aversion show that decisions made after a loss are biased. A rule written in advance takes the decision away from the moment when bias is strongest.
- Compare against the starting point, not yesterday. Gradual-change blindness works because each day is compared with the one before. A scheduled review against your entry price and original thesis restores the comparison that matters.
- Write your thesis in one sentence. If you can no longer restate why you hold a position without new excuses, your reasons have drifted.
- Ask whether you would buy it today. Sunk cost is about money already spent. The question looks only at what comes next.
- Treat yields you cannot explain as a warning. In the Terra case, the SEC complaint centres on a yield of up to 20% presented as stable.
- Understand liquidation mechanics before using leverage. In October 2025, most of the losses came from leveraged longs closed automatically in a fast move.
Limitations: most of the research cited here studied stock investors and lab participants, not crypto traders specifically.
Frequently asked questions
What is the boiled frog theory in simple terms?
It is a metaphor for failing to react to danger that builds slowly. Investors apply it to positions that lose value in small steps until the total loss is large.
Is the boiling frog story true?
No. Modern biologists such as Douglas Melton, George Zug and Victor Hutchison say real frogs try to escape as water heats. The idea survives because people, not frogs, show the pattern.
Is there evidence that investors behave like the frog?
Yes. Da, Gurun and Warachka (2014) found that investors underreact to information arriving gradually, and Simons and colleagues (2000) showed people miss changes that unfold slowly.
How is it different from loss aversion?
The boiled frog effect is about not noticing a loss as it builds. Loss aversion is about being reluctant to accept a loss once you see it. They often compound each other.
