How Large‑Scale Human Behavior Shapes Stock Market Reactions to Earnings Reports
- Derin Goktepe
- Nov 10, 2025
- 3 min read
Overview
Financial markets are often described as rational systems, yet real price movements frequently tell a different story. In my published research, The Impact of Large‑Scale Behavior on the Stock Market, I analyzed earnings reactions for the top twenty companies in the S&P 500 to see whether stock prices respond logically to new information or whether human psychology plays a larger role. The findings showed that only 40 percent of earnings reactions were rational, while the remaining 60 percent showed signs of irrational behavior. As the paper notes, “price reactions were not always rational, and concepts such as negativity bias and herding behavior often affected stock prices negatively.”
This post summarizes the research and explains the behavioral patterns that appeared most often.
Why I Studied Behavioral Economics in Earnings Reactions
Traditional finance assumes that markets efficiently incorporate new information. Behavioral economics suggests something different. Investors often react emotionally, especially during earnings season when uncertainty is high and trading volume spikes. Earnings reports are ideal for studying this because they introduce clear new information and often lead to large price moves. My goal was to test whether concepts like negativity bias and herding could explain these reactions more accurately than the Efficient Market Hypothesis.
Methodology
To keep the analysis consistent, I focused on the top twenty companies in the S&P 500 and examined all of their earnings reports from 2023. For each company, I compared earnings per share and revenue to analyst expectations and then measured the next day’s price reaction. Each reaction was classified as rational, slightly irrational, irrational, or highly irrational. The dataset was limited to mega‑cap companies because they are less likely to move for random or illiquid reasons and provide cleaner behavioral signals.
Key Findings
Rational reactions were the minority
Only eight of the twenty earnings reactions were rational. The remaining twelve showed some degree of irrationality. This means most price movements did not align with what the Efficient Market Hypothesis would predict.
Negativity bias appeared repeatedly
Investors consistently focused more on negative details than on overwhelmingly positive earnings. Nvidia is a clear example. The company reported results that strongly exceeded expectations, yet the stock fell 2.5 percent the next day. A minor comment about export restrictions overshadowed the positive news, which is a classic example of negativity bias.
Herding behavior amplified declines
In several cases, small initial drops led to larger declines as more investors followed the early movement. Broadcom slightly beat expectations, yet the stock fell 5.5 percent. The most likely explanation is that early selling created fear and triggered more selling.
Tech companies were affected the most
Although tech companies made up 40 percent of the dataset, they accounted for 75 percent of the highly irrational reactions. This suggests that tech stocks may be more sensitive to sentiment, guidance, and fear of future uncertainty.
Why These Findings Matter
The results show that earnings reactions cannot be understood through financial metrics alone. Investors respond to tone, guidance, macroeconomic fears, and emotional cues. Behavioral economics helps explain these patterns and offers a more realistic view of how markets behave. Understanding these tendencies can help investors interpret earnings reactions more accurately and avoid emotional decision‑making.
What I Learned as a Researcher
Conducting this study showed me that markets are not purely rational systems. Human behavior plays a measurable role, even in the largest and most widely followed companies. Behavioral economics is not just a theoretical framework. It is a practical tool for understanding real‑world price movements.
Full Research Paper
Readers who want to explore the complete analysis can access the published paper here: