Behavioral finance, risk management

Why losses lead to riskier bets

The script repeats from small forex accounts to Bill Hwang's $20 billion. Behavioral economics has measured one mechanism behind it: people deep in the red become more willing to accept bets they would normally refuse.

I. The pho shop story

Anyone who has spent enough time near markets has heard a version of this story. A friend of a friend trades forex. Starts small, wins. Scales up until, at the peak, he clears a billion dong, roughly forty thousand dollars, in a single day. Buys a car, buys a house, gets married. Then the losing streak begins. The car goes first. Then the house. Then loans from the in-laws. Then the divorce. His aging parents bail him out and set him up with a pho restaurant to start over. A while later the itch returns, and he sells the pho restaurant to fund his way back into the market.

The story sounds invented because it follows the script too closely. But the script does not care about scale. Its biggest versions come with court records and investigation reports:

Who	Peak	Bottom	Mechanism
Jesse Livermore	Made about $100 million shorting the 1929 crash, commonly converted to roughly $1.5 billion today.	Bankrupt in 1934, five years after the peak of his career.	A lifelong loop: broke in 1908, bankrupt in 1914, rich again by 1916. Win with big size, lose with big size.
Nick Leeson (Barings)	The star who generated most of the Singapore branch's profits.	£827 million in losses in 1995, twice the bank's available capital; the 233-year-old Barings sold to ING for £1.	Hid losses in account 88888 and kept doubling the Nikkei position to win them back; the Kobe earthquake ended the game.
Bill Hwang (Archegos)	Around $20 billion in personal wealth; a portfolio levered to roughly $160 billion with bank money.	Lost nearly all of it over two days in late March 2021 (per Bloomberg); lending banks lost over $10 billion, Credit Suisse alone $5.5 billion; an 18-year prison sentence.	Cascading margin calls when ViacomCBS turned; size was never cut while cutting was still possible.

Their earlier results did not prevent the collapse. Once losses appeared, all three maintained or increased their risk instead of reducing their positions. These cases do not prove a psychological mechanism by themselves, but they show that experience and past performance offer little protection once a position has moved beyond control.

At the retail level, the aggregate damage from frequent trading is also large. Barber, Lee, Liu and Odean measured it on complete trading data for the Taiwan market, 1995 to 1999: individual investors' net trading losses equaled 2.2% of GDP per year. A follow-up study on the same market counted fewer than 1% of day traders, about 4,000 out of 450,000 in a typical year, earning reliable profits net of fees.

The sale of the pho restaurant shows how persistent a reference point can be. The business was not failing, but the owner's current standard of living was still being measured against the old peak of "a billion dong a day." The gap from that peak made an income-producing business look like a setback that had to be recovered.

II. Mental capital: the second account

The tendency to raise bets while losing has a studied mechanism. A drawdown reduces the balance shown in a brokerage app while also weakening the ability to judge a new opportunity independently of the existing loss. Traders often call the ability to stay calm and follow a plan mental capital.

1. Why losing makes people take wilder risks

When an account is at break-even, a high-risk trade is easy to see for what it is: it may win a lot, but it may also lose a lot. After a loss, the same trade becomes more attractive because it offers a route back to even. Closing the position means accepting the loss; placing another trade postpones that feeling.

Same risk, different perception

Both branches in the lower example have the same probability. Line weight shows attention, not probability.

Kahneman and Tversky's prospect theory explains this shift through a reference point. People look at the money they have left and compare it with a marker such as their cost basis, the account's peak, or the amount they once had. While they remain below that marker, the chance to regain it can crowd out the more important question: is the new trade worth the risk?

A person who has not accepted a loss is more likely to take a gamble they would normally reject. Thaler and Johnson (1990) called the preference for a gamble that offers a path back to even after a loss the break-even effect. In a trading account, it appears in familiar reasoning: add another order to lower the cost basis, increase size to recover faster, or hold a position because selling would confirm that the original decision was wrong. The old loss is now determining the risk of the new trade.

2. The 10,000-account dataset

Terrance Odean tested this behavior on real data: trading records of 10,000 accounts at a large US discount broker, 1987 to 1993, published in the Journal of Finance in 1998. The result: a stock sitting at a gain was roughly 50% more likely to be sold than a stock sitting at a loss in the same portfolio. Investors cut flowers and watered weeds, exactly as prospect theory predicts. The pattern could not be explained by portfolio rebalancing or trading costs, because Odean controlled for both.

Over the next 252 trading days, the winners investors sold outperformed the losers they kept by about 3.4 percentage points. Sold winners earned an excess return of +2.35%, while held losers returned -1.06% against the CRSP index. Investors sold the group that subsequently performed better and held the group that subsequently performed worse.

The effect reversed in December, when the tax benefit created a clearer reason to realize losses. This detail shows that loss-holding can change when a concrete incentive becomes strong enough to overcome the feeling that "it is not a loss until I sell."

3. The loss-holder's spiral

Put these pieces together and the chain reaction of an underwater account plays out almost on script:

How a drawdown pulls an account lower

This is a possible chain reaction, not an inevitable path for every account or every market cycle.

This spiral explains why veteran traders hate deep drawdowns: the person at its bottom has run out of both cash and composure at exactly the moment both are worth the most.

III. The math of drawdowns

The gain required to break even rises faster than the loss. A 10% loss requires an 11.1% gain. A 50% loss requires a 100% gain.

Loss	Gain needed to break even	Scale of the recovery
-10%	+11%	The gain needed to break even is only slightly larger than the original loss.
-20%	+25%	The remaining capital has to rise by one quarter.
-30%	+43%	The required gain is already 13 percentage points larger than the loss.
-50%	+100%	The account has to double just to return to its starting point.
-70%	+233%	The remaining capital has to more than triple.
-90%	+900%	The remaining capital has to increase tenfold.

This arithmetic explains why traders who survive for decades talk about defense before offense. Paul Tudor Jones, famous for calling the October 1987 crash, put it this way in Jack Schwager's Market Wizards:

Many investors believe success depends on finding a stock or coin that is about to rise sharply. The idea sounds reasonable, so they keep hunting for a winner even when the opportunity is unclear and the risk is already too high.

Suppose VND 10 million is invested in a very risky asset and loses 50%. The remaining capital is VND 5 million, so the next investment has to gain 100% just to return to the starting amount. Finding an asset that doubles is already difficult. Even if the investor finds one, the final result is only a return to VND 10 million.

If the loss is stopped at 20%, VND 8 million remains. The same 100% opportunity would take the account to VND 16 million, 60% above the starting capital. Finding a good investment still matters, but the capital left when that opportunity appears determines how much the investor can make.

IV. Learning how not to lose money

Preventing mistakes also contributes to long-term returns. A period with no gains leaves all the capital available for the next opportunity. A bad decision can take both the capital and the ability to make the next decision calmly.

Learning how not to lose money does not mean avoiding every losing trade. Small losses are a normal cost of investing. The danger is a loss large enough to change the way the next decision is made, then trigger larger positions, a broken plan, or another trade placed only to get back to even.

Many ways to make money, few ways to lose it

Jim Paul and Brendan Moynihan approach the same problem through failure in What I Learned Losing a Million Dollars. Paul read about Peter Lynch, Bernard Baruch, Jim Rogers, Paul Tudor Jones, Richard Dennis, and other successful investors in search of a shared secret. He did not find one. Their strategies contradicted one another: what one avoided was exactly what another did to make money.

Paul shifted the question from how people win to how they lose. Large losses often come back to a few familiar errors: bad analysis, failure to follow a limit set in advance, or allowing the loss to become a matter of ego. People may make money with different methods, but nobody needs to turn one losing trade into a reason to break the entire system.

Amateurs win by making fewer mistakes

In Extraordinary Tennis for the Ordinary Tennis Player, Simon Ramo divides tennis into two games. Professionals often win a point with a shot the opponent cannot return. At the amateur level, points more often end because one player hits the net, sends the ball out, or double-faults. Winners do not need many outstanding shots. They need to make fewer mistakes.

Problems begin when amateurs copy the professional game. Big serves, changes of direction close to the line, and constant trips to the net require technique, fitness, and positioning they do not yet have. Trying to hit a beautiful winner increases the number of unforced errors. At that level, keeping the ball in play offers the better chance of winning.

For an individual investor, the equivalent goal is to stay in the game long enough. There is no need to win every trade, but each loss has to be small enough to leave capital for the next one. Without clear skill or an edge, using margin, calling tops and bottoms, and trading constantly only increases the number of self-inflicted errors.

Cash preserves choice

Long-term investing does not require an account to hold a position at all times. Sitting out may produce regret while the market keeps rising, but cash preserves the right to reject a setup that falls short and buy when a better opportunity appears. That choice is especially valuable after a losing streak, when the urge to break even can override the quality of the next trade.

Two principles in Bernard Baruch's investment philosophy are to keep a substantial share of capital in cash and focus only on the area one understands. Cash provides room to correct a mistake without a margin call forcing a sale. A defined circle of competence keeps the investor from chasing every story attached to a rising price.

A winning streak does not reveal how much came from skill and how much from favorable conditions. Investors cannot control the cycle or the timing of the next opportunity. They can control position size, leverage, cash, and whether to accept a setup that does not meet their standard.

V. The 8-point setup

The spiral often begins when the trade is selected. Suppose a trader's checklist has 10 criteria, covering the company's position, capital flows, the price base, and the invalidation point. A qualifying setup should make it possible to write down the reason for buying, the condition that would prove the thesis wrong, and the maximum loss before placing the order.

The harder decisions involve setups that meet only seven or eight criteria. They are not necessarily bad: the chart looks acceptable, the story sounds reasonable, and only one or two boxes remain unticked. The person holding cash can easily tell themselves "probably fine." If the setup fails, the drawdown and the urge to recover it begin. If it merely goes sideways, the cash reserved for a better opportunity is already tied up.

VI. Why dodging a crash is not simple

Cutting one position when its thesis fails is different from selling an entire portfolio to call a market top. The drawdown table can suggest a simple conclusion: just avoid the crashes. By Ben Carlson's calculation using S&P 500 data from 1990 through mid-2026, $1 held continuously grew to about $40. If the 25 worst days were avoided, it grew to roughly $240.

But miss the 25 best days over the same period and that $1 grows to only $8. The two groups of days often occur close together. Per JPMorgan data over 20 years, 7 of the S&P 500's 10 best days landed within two weeks of the 10 worst days. Selling into a panic therefore increases the risk of missing the rebound.

An investor who put $10,000 into the S&P 500 in 2005 had $71,750 by the end of 2024 if they stayed invested, but only $32,871 after missing the 10 best days. Exiting before the decline and returning before the rebound require two consecutive correct decisions, not one.

Why getting out and back in is difficult

The value bars use the same scale. Best-day and portfolio-return figures are from JPMorgan data summarized by CNBC in 2025.

The difference is the order of operations. The top-caller acts on a forecast of the future; the risk manager acts on the current state of the portfolio. The second approach is precisely the spirit of Nassim Taleb's "panic early": panicking early, while there is still liquidity and still a choice, costs less than panicking late, when many people need to exit at once. For someone planning to remain in the market for more than one cycle, the practical goal is to keep risk low enough to avoid being forced to sell during a bad session.

VII. Five working rules

Five rules, ordered by decision point

Set the first two rules before entering. The other three keep an old loss from controlling the next decision.

Closing

A 50% loss demanding a 100% gain is the arithmetic part. The hard part is behavioral: Odean's data shows that investors tend to hold losers longer than winners, while Thaler and Johnson's experiments show that after a loss, people become more willing to choose a gamble that offers a path back to even. Cutting losses early therefore preserves both capital and the ability to evaluate the next opportunity without the old loss controlling the decision.