Interview, Other
Why knowing the future doesn't always help predict markets | The Economist
- Market price movements are driven primarily by expectations relative to actual news rather than the news itself, meaning positive announcements may fail to boost stocks if results only marginally exceed forecasts.
- Artificial intelligence models are projected to slightly outperform humans in predicting market direction but remain equally ineffective at determining correct bet sizing, with some models potentially losing money in experimental settings.
- Expert macro traders are expected to significantly outperform both lay participants and AI systems, potentially doubling their capital by adjusting bet sizing based on confidence levels, including the strategy of abstaining from trading on low-confidence days.
- The critical differentiator among experts, AI, and lay participants is the degree to which investment sizing is scaled according to confidence, with experts predicted to massively increase leverage when highly confident.
- Market participants may not care about the accuracy of predictions regarding future earnings or economic growth if the goal is profitability, as it can be rational to bet on a direction even if the market's future prediction based on that direction is incorrect.
- Investors are expected to fail to adequately address the fundamental question of optimal bet sizing, which is identified as the most crucial factor in taking advantage of confidence to generate profits.
- The experiment involving the average person, who had access to future headlines, projected an average final capital of approximately $51.