Broome Street Labs builds forecasts, live models, and data infrastructure that turn partial, messy, fast-moving data into numbers your team can act on.
Everything from the statistics to the data pipelines to production, so your team can make decisions with confidence.
Models that update as new data arrives, with uncertainty that is calibrated against how often past forecasts were actually wrong.
Polling averages, weighting, and subgroup estimates that account for house effects, sparse samples, and data that doesn't agree with itself.
Pipelines that turn messy public and proprietary data into clean, monitored feeds and tools your team uses every day.
Recommendation, tagging, and bandit systems that keep learning in production after launch.