Anyone who has watched a stock price slide knows the uncomfortable feeling: a tightening in the chest, the urge to check the screen one more time, the quiet hope that things will bounce back. Traders face this moment constantly, and the decisions they make in those seconds can shape portfolios and careers. But what actually goes on inside a person’s head during a losing streak, and do the mental shortcuts we rely on shift from hour to hour?
A research team at the University of Mons in Belgium set out to track exactly that. Their qualitative investigation, published in the Journal of Next-Generation Research 5.0, followed eight university students through a three-day simulated trading experience and mapped how six well-known mental biases appeared, shifted, or faded as the participants wrestled with a market perceived as unfavorable.
The question driving the research
Traditional finance assumed investors were rational calculators. Over the past few decades, behavioral finance has pushed back, showing that emotions and mental shortcuts shape trading decisions in measurable ways. Yet most of that research captures a single moment in time. Professor Alain Finet and his colleagues Kevin Kristoforidis and Julie Laznicka wanted to know how biases change during an active trading period, not just whether they exist.
The team focused on six biases familiar to behavioral finance researchers. Overconfidence refers to people overestimating their skill or knowledge. Anchoring describes the habit of clinging to an initial reference number, such as the price paid for a stock. Availability bias is the tendency to lean on information that is easy to recall. Representativeness bias involves judging situations by how closely they resemble a familiar pattern. Herd behavior describes following the crowd. Prospect theory, developed by Daniel Kahneman and Amos Tversky, captures how people feel losses more intensely than equivalent gains, a phenomenon known as loss aversion.
How the experiment was set up
In January 2025, the researchers recruited eight management students, seven men and one woman, from the University of Mons. Each person received a virtual portfolio of 100,000 euros and traded shares on the CAC40, the main French stock index, through an online platform called ABC Bourse. Trading ran for twelve one-hour sessions spread across three consecutive days. Participants could see how their peers were performing in real time, which added a layer of social pressure similar to a real trading floor.
During those three days, the CAC40 drifted slightly downward. The loss was modest in objective terms, but participants reported perceiving the market as difficult and discouraging. News about DeepSeek, a Chinese artificial intelligence company challenging American rivals, dominated headlines on the first two days. An underwhelming earnings report from LVMH and an anticipated Federal Reserve announcement shaped the third day.
Once trading ended, a single researcher conducted individual semi-structured interviews with each participant. The conversations lasted between 33 and 59 minutes and generated transcripts averaging about 6,300 words per person. The three authors then read the transcripts repeatedly, coded specific passages line by line, grouped the codes into themes tied to each bias, and tracked how those themes shifted from day one through day three. The team deliberately avoided using artificial intelligence tools, arguing that current technology cannot fully capture emotional reasoning.
What the interviews revealed
Loss aversion emerged as the strongest and most consistent thread across the group. Even though the money was virtual, participants described the experience of losing as painful, frustrating, and at times demoralizing. One participant said losses “hurt a little, even if it’s not my money.” Gains, by contrast, produced moderate satisfaction rather than excitement. Several participants refused to sell losing positions, holding on in the hope of a rebound. Others did the opposite, using strict automatic stop-losses to cut losses quickly. A few responded to mounting losses by taking bigger risks in an attempt to recover, a pattern the researchers described as a “what’s done is done” attitude.
Availability bias showed up in nearly every interview. Participants leaned heavily on information that was easy to grab: price charts, familiar company names such as L’Oréal or LVMH, short news items, and popular technical indicators. Fundamental analysis, which involves studying balance sheets and financial ratios, was largely rejected as too slow or too complex for the pace of the simulation.
Overconfidence turned out to be the most variable bias. Some participants started with low confidence because they had little experience. For others, early confidence eroded as losses piled up, sometimes ending in resignation. A few maintained their confidence by interpreting past experiences positively or by sticking to rules they trusted.
Representativeness bias appeared in stable ways. Participants with prior experience in cryptocurrency trading tried to apply those same tactics to stocks. Others reached for academic frameworks they had learned in class, treating well-known indicators or models as reliable guides simply because they were well-known.
Anchoring bias was less common but showed up in specific forms, such as participants setting fixed gain or loss limits like plus or minus 20 percent, or holding onto the expectation that prices would return to their purchase point.
Herd behavior took a surprising shape. Rather than copying the trades of peers, participants used the group as a source of emotional comfort. Several said that knowing others were also losing money helped them feel less alone. As one participant put it, the group became a kind of reservoir for shared disappointment.
What this means for people who trade or advise traders
For financial educators and advisors, the findings suggest that teaching about biases in the abstract may miss the point. The way a bias shows up on day one can look very different by day three, especially when markets move against the trader. Training that helps people notice these shifts in real time, such as journaling decisions or using pre-set rules, may be more useful than one-off lessons.
The study also points to the emotional weight of losses even when no real money is on the line. For firms that use trading simulators to train new hires, this is a reminder that simulated environments can still produce genuine emotional responses worth discussing openly.
Some caveats deserve attention. The sample included only eight students, and qualitative research of this kind is designed to produce depth of insight rather than statistical generalization. The participants were mostly male, reflecting broader patterns in finance research but limiting what the work can say about gender differences. The market during the experiment trended slightly downward, and the authors note that biases might behave quite differently in a rising or highly volatile market. They suggest future research could examine bull markets and explore how biases feed into one another over time.
