Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice

Battle Count: 113

While not directly applicable to trading, the insights into human decision-making under risk and over time could inform models of market behavior and investor psychology.

Requires custom implementation of a language model and generation of synthetic datasets, but overall approach is well-described and manageable for researchers with ML experience.

The paper provides detailed information on model architecture, training data generation, and evaluation procedures, enhancing reproducibility.