Mizuho and IBM build market forecasting tool that mimics brain’s learning schemes


Mizuho Financial Group has teamed up with IBM to develop an AI-based financial market forecasting tool that mimics the human brain’s learning scheme.

The tool uses artificial neural network technology, called dynamic Boltzmann machine (DyBM), that has been developed by IBM Research’s Tokyo team. The technology utilises time-series prediction logics to extract similar trend dates and create future market trend predictions.

It’s now being put to work in the financial giant’s global markets company, in the asset liability management and treasury portfolio operation.

Mizuho tapped outputs of multiple forecast models using market data for the last two decades to identify past dates which have high degree of similarity with the current market. Then the firm mapped the subsequent price trends of each similar date to develop the forecasting tool, which predicts future price trends and volatility.

Mizuho has embraced AI, working with Nomura Research Institute recently on two separate projects, using the technology to help monitor and analyse communications between sales reps and investors as well as to eliminate lengthy and complex derivative contract procedures.



Source link