Case Study
One-Click Extraction: Precision and Efficiency in Every Trade Request
Learn How ING Automates Sales Trader Workflow With Sense Street
Accuracy
Average Latency
Challenge
Fixed Income sales face a very high volume of daily RFQs, especially in volatile markets. ING aims to enable its team to focus on high-value tasks, like engaging with clients or winning trades, while minimizing errors and reducing time spent on booking tickets.
Time-Consuming Process
Error-Prone Ticket Booking
High RFQ Volume
Sense Street’s STW solution helps broker-dealers automate the recording of both executed and missed RFQs, ensuring compliance with MiFID II requirements.
It also captures valuable data points that offer insights into client behavior and trading activity.
Finding a Better Solution for Your Manual Workflows
Sense Street integrates API calls directly into ING’s real-time Sales-Trader Workflow (STW), making Fixed Income sales teams more efficient.
This integration automates the booking of voice Requests for Quotes (RFQs) received from buy-side clients, eliminating the need for manual data entry and reducing redundant transcription of chat information. As a result, Sense Street enables sales teams to allocate their time and resources to more strategic tasks, ultimately enhancing overall productivity.
Deployment & Results
Sense Street interactively ingests data from ING and provides a structured output in return, which is then integrated into the STW platform for immediate use.
This results in low latency, high precision and a low-error process.
The partnership between Sense Street and ING has led to efficiency improvements:
The STW solution consistently delivers accuracy* above 97%.
The accuracy of the model has improved significantly, from 85% during the training phase to 97% after a year in production. This is achieved by combining a human-in-the-loop approach that continuously aligns with ING's usage.
Filling STW is now just one click, streamlining workflow and allowing sales teams to focus on client engagement and trade execution.
*Average F-score across all extracted fields.