Ticketmaster / Live Nation · 2019 to 2020
Marketplace Pricing & Reinforcement Learning
Reinforcement-learning policies balancing real-time supply and demand across a national two-sided live-entertainment marketplace.
Role · Senior Product Data Scientist. Pricing product owner.
The problem
Live event inventory is perishable, supply is fixed the moment it goes on sale, and demand arrives in bursts driven by news and sentiment. Static pricing leaves money on one side of the market and empty seats on the other. The constraint that matters is not modeling difficulty; it is that promoters, venues, and artists all hold veto power over the price.
Approach
- 01
Owned the pricing product surface, not only the model: the levers promoters and venues actually touch and the guardrails they set before the policy is allowed to move anything.
- 02
Deployed reinforcement-learning policies that adapt to real-time demand signals within rights-holder constraints.
- 03
Instrumented outcomes so every pricing decision could be evaluated against a counterfactual instead of against last year's gross.
Outcome
Pricing products deployed across a marketplace operating at national scale.