Scaling Time-Series Forecasting Models to Cope with the Multi-Verse with Ray

Explore a 25-minute conference talk on scaling time-series forecasting models using Ray to handle the multi-verse approach in financial markets. Learn how JPMorgan Chase leverages probabilistic forecasting at scale to improve time-series models for various finance industry use cases. Discover the concept of back-testing to develop probabilistic models based on multiple slices of past data, addressing non-stationarity and outlier influence. Understand the need for large-scale compute and distributed ML model development in probabilistic forecasting and non-stationary environments. Gain insights into potential applications for forecasting stock, commodity, energy prices, interest rates, and exchange rates across different time horizons. Examine JPMorgan Chase's platform for large-scale ML-based time-series forecasting, built on Ray's capabilities for efficient cloud-based distributed computations using Kubernetes. Delve into the platform's features, including probabilistic regression, feature engineering, selection, hyper-parameter optimization, and analysis metrics. Recognize the potential of distributed time-series forecasting powered by Ray to drive improvements in efficiency, profitability, and sustainability in the financial sector.

Scaling time-series forecasting models to cope with the multi-verse with Ray