Which team won the Netflix Prize in 2009?

BellKor's Pragmatic Chaos won the Netflix Prize in 2009.

The Netflix Prize was an open competition launched in 2006 to improve Cinematch, Netflix’s movie-rating prediction system. Contestants received anonymized customer-rating data and competed to reduce the error in predicting how users would rate films. The grand prize was $1 million for a 10% improvement over Netflix’s benchmark.

BellKor's Pragmatic Chaos was formed by merging leading groups including BellKor in BigChaos and Pragmatic Theory. Its members included researchers associated with AT&T Labs, Yahoo, Commendo Research & Consulting and Pragmatic Theory. The team used an ensemble of techniques rather than relying on one simple recommendation formula.

Netflix announced the winner on September 18, 2009, and awarded the prize at a ceremony three days later. The team achieved a test RMSE of 0.8567. The Ensemble had matched its result, but BellKor's Pragmatic Chaos submitted its qualifying solution about 20 minutes earlier, and the contest rules used submission time as the tie-breaker.

The winning algorithm was not ultimately deployed because integrating it into Netflix’s production system was judged too costly relative to its expected benefit.