Rational Long Shot in Dynamic Tournaments with Interim Performance Feedback

We propose a rational model to explain the use of counter-intuitive tactics in dynamic (two-person) tournaments. The model predicts when and why both leaders and trailers sometimes intentionally take “risky” tactics in order to win the final victory in high incentive tournaments in the mutual fund market, professional sports, competition for executive positions and so on. We then conduct a controlled laboratory experiment on this dynamic tournament with interim performance feedback, and find results that coincide with theoretical predictions in the following three ways: First, players follow model prediction 78% of the time. Second, regression analysis shows that 95% of players follow prediction when theory suggests the “safe” (piece-rate) tactic, while when theory suggests the “risky” (long shot) scheme, trailers follow this prediction 74% (81% when it dominant) of the time, even after controlling for players’ ability, gender and risk attitude. Also, we find that gender and risk preferences play a role in leaders’ behavior when theory predicts indifference. Finally, the experimental distribution of final performance difference is also close to what theory predicts.