Introduction: The Pervasive Influence of Cognitive Biases in Trading
Trading systems are not created in a vacuum; they are products of human decision-making, and thus are susceptible to a wide range of cognitive biases. These mental shortcuts, while often efficient in daily life, can become systematic pitfalls in the complex and uncertain environment of financial markets. The study notes on which this article is based identify several key preferences and biases that influence three critical stages: the development of a trading system, its testing, and its live operation. Understanding these biases is the first step toward building more objective and resilient trading approaches. This article examines each bias in detail, explaining how it distorts perception, how it can lead to flawed conclusions, and what conditions make it more likely to occur. It does not offer trading recommendations but rather provides an educational overview of the psychological factors that can undermine even well-intentioned system design. By recognizing these tendencies, traders and system developers can at least be aware of the traps that await them. The biases discussed include representation, reliability, lottery, small numbers, conservatism, randomness, and the need to understand, as well as biases in testing such as degrees of freedom, hindsight, and lack of self-protection, and operational biases like the gambler's fallacy and profit/loss asymmetry. Each will be explored in the context of how it can lead to suboptimal decisions and how awareness can help in conceptualizing a more robust trading system.
Representation and Reliability Biases in System Development
The representation bias refers to the tendency to treat a symbol or a model as the thing it represents. For example, a trader might come to view a daily bar chart as the market itself, or believe that a favorite indicator is a direct reflection of market reality. However, as the notes emphasize, any representation is merely a shortcut for conveying a large amount of information, and it can distort that information. A bar chart compresses price action into four data points—open, high, low, close—losing the nuances of intraday movements. An indicator like RSI is a mathematical transformation, not the market. When traders forget this, they may react to the representation rather than to the underlying market dynamics. The reliability bias compounds this issue: people tend to assume that the data they use is accurate. Yet historical market data and real-time feeds are frequently riddled with errors—missing ticks, bad prints, or incorrect timestamps. If a system is built on unreliable data, its signals and backtest results are suspect. For instance, a backtest that uses a price spike from a data error might show a profitable trade that never actually occurred. The combination of representation and reliability biases can lead to a false sense of precision and control. Traders may develop systems that perform well on flawed historical data but fail in live markets. Awareness of these biases involves questioning the fidelity of data sources and remembering that all models are simplifications. It also means verifying data quality and understanding the limitations of any chart or indicator used.
