"The Dog and the Frisbee": Why Simple Rules Outperform Complex Models in Financial Uncertainty
A 2012 paper by Andrew G Haldane, Executive Director of the Bank of England, and Vasileios Madouros argues that simple heuristics often outperform complex mathematical models when navigating financial crises. The authors use a dog catching a frisbee as an analogy: despite a chaotic physical environment, the dog succeeds using only the simple rule of keeping its gaze angle constant, whereas human regulators attempting increasingly complex market prediction models often fail.
In their 2012 paper "The Dog and the Frisbee" (https://www.bankofengland.co.uk/-/media/boe/files/paper/2012/the-dog-and-the-frisbee.pdf ↗), Andrew G Haldane, Executive Director of the Bank of England, and Vasileios Madouros present a counterintuitive argument: simple heuristic rules are often more effective than precise mathematical models when addressing complex financial crises. Using the example of a dog catching a frisbee, they note that while the physical environment is extremely complex, the dog can succeed simply by following the rule to "keep the line-of-sight angle constant." In contrast, human regulators attempt to build increasingly complex models to predict markets, often with adverse results.
The core of this theory lies in distinguishing between "risk" and "uncertainty." Mainstream economics, based on the Arrow-Debreu and Merton-Markowitz frameworks, assumes that future states have known probability distributions, allowing for hedging through calculation. However, the real world is characterized by Herbert Simon's concept of "bounded rationality," where information collection and processing costs are high and cognitive capacity is limited. In such an uncertain environment, complex models overfitted to historical data tend to mistake noise for signal, leading to fragile predictions; conversely, simple rules that ignore some information prove more robust.
For investors, this perspective challenges the traditional reliance on massive datasets and complex algorithms in active management. Citing experimental evidence, the paper shows that in sports forecasting, medical diagnosis, and even stock picking, simple passive strategies or single-indicator heuristics often outperform complex active analysis. For instance, the "recognition heuristic"—choosing winning teams solely based on name recognition—outperformed FIFA and ATP ranking systems based on extensive historical data. This suggests that in investing, reducing information overload and adopting more robust simple rules may yield steadier long-term returns than pursuing extreme precision with complex models.