# Loss Aversion and Learning in Professional Golf Putting

## TL;DR

Professional golfers show loss aversion by prioritizing par putts over birdie attempts and improve performance with immediate feedback, revealing insights into behavioral economics.

## Contribution

New empirical evidence on loss aversion and learning using high-frequency golf putting data from the KPGA Tour.

## Key findings

- Golfers are more likely to succeed at par-saving putts than birdie attempts from the same distance, indicating loss aversion.
- Second putts show higher success rates than first putts, demonstrating within-hole learning from immediate feedback.
- Players adjust their pace strategically to avoid costly errors, showing self-regulation under loss-framed incentives.

## Abstract

This paper provides new field-based evidence on loss aversion and short-run learning using high-frequency performance data from professional golf. Leveraging over 100,000 putts recorded during the 2020 Korea Professional Golfers’ Association (KPGA) Tour, I examine how professional golfers adjust their putting behavior in response to reference-dependent incentives and immediate feedback. The structure of golf creates a natural empirical setting to test behavioral predictions: scoring rules establish salient reference points (e.g., par), while putting decisions are discrete, individually executed, and financially consequential. I find that players are significantly more likely to convert par-saving putts than birdie attempts from equivalent distances, consistent with loss aversion and reference-dependent preferences. Par putts are also executed more aggressively, but players regulate pace to avoid costly three-putt errors, indicating strategic self-regulation under loss-framed incentives. In addition, I document robust evidence of within-hole learning: second putts—taken shortly after the first under near-identical conditions—exhibit substantially higher success rates. These patterns are confirmed in logistic regression models with nonlinear distance controls and player fixed effects. This performance gap persists across scoring frames and aligns with models of reinforcement learning and dynamic belief updating. The findings illustrate how behavioral biases and adaptive learning interact in high-stakes, real-world decisions and highlight the value of professional sports data for testing core theories in behavioral economics.