Financial institutions are regulated to maintain specific liquidity ratios to meet their liquidity needs under systemic stress environment extending up to 30 calendar days. The minimum LCR requirement specified by Basel III varies from 60% in 2015 to 100% in 2019. This calls for a close watch on the liquidity positions of financial institution which is cumbersome to manage on a manual basis; and hence, a need for a machine learning / artificial intelligence based solution.
This machine learning module uses historical, high-quality liquid assets, inflow, outflow and net cash flow to forecast values for the current day and computes the Liquidity Coverage Ratio (LCR). The module starts with a dashboard of LCR chart and its high-level components such as HQLA, Inflow, Outflow and Off-Balance sheet and upon drilldown exhibits the last 30 days of data related to each. It uses advanced time series analysis technique known as seasonal auto regressive integrated moving average (in short SARIMA) which is reliable to provide short-term forecasts on high-frequency data. The technique considers all the underlying time series components such as trend, seasonal, cyclical and irregular while generating forecasts, making it more accurate where seasonal and cyclical effects are dominant. Furthermore, there is no theoretical upper bound on historical volume of data that can be fed into this algorithm and hence, it is a highly-scalable solution.
Streamlined LCR Reporting:
LCR Dashboard: Increased data accuracy by consolidating LCR data into a single, easy to understand dashboard to ensure the timely arrival of data and feed AI models.
HQLA Market Analysis: Advanced AI analysis of LCR data to generate reliable short-term forecasts using historic data series and machine learning techniques. The analysis is able to determine trend and estimates for the day, use the data as approximation for processing, conduct back testing of approximation based on actual data, and use historic data with behavioral patterns to further enhance approximations.
Intra-day LCR Reporting: Faster LCR reporting would lead to a reduction in FTE operations costs and help firms move toward Intra-day LCR reporting through enhanced processes with increased predictability and reliability.
Monitoring: Managing the LCR at an optimal level to mitigate under borrowing or over borrowing either internally or externally hence reducing losses and increased profits.
Forecasted LCR: LCR for settlement for T day can be managed with forecasted LCR.
Increased Visibility and Predictability: Machine learning from large data of 30 years along with market factors can provide ML opportunity for long-term visibility and predictability.
To learn more about our Artificial Intelligence solutions for Cognitive Machine Learning and the work we’re doing email us at email@example.com
How we are innovating
Comprehensive integrated valuation for illiquid OTC products and a predictable real-time OTC price for traders to use in their models or use for auto-quoting.Read More