Case Study: TradeSmith improves financial time series forecasting with Nixtla TimeGPT

TradeSmith - Customer Case Study

TradeSmith, a financial technology company serving over 60,000 active investors, faced challenges in scaling its investment forecasting to tens of thousands of daily financial time series. Their legacy tree-based models struggled with accuracy, speed, and the cumbersome integration of numerous exogenous variables. To overcome this, they turned to vendor Nixtla and its TimeGPT service.

Nixtla implemented its TimeGPT model into TradeSmith's production pipeline, which delivered over 100,000 forecasts monthly. The solution outperformed the legacy xgboost models in both accuracy and inference speed, while also streamlining the feature engineering process. This allowed TradeSmith to reliably generate over 22,000 daily forecasts, providing their investors with superior, actionable insights.

TradeSmith

Michael Carr

Chief, Quantitative Research