BlogCases

AI/ML Solution for Investment Risk Forecast and Assessment

Cases

Company's activities: A large financial company (+10,000 employees) with a turnover of $1,000,000,000 that provides consulting services in the field of long-term investment.

Business Task

To build graphs of popularity and users' attitudes (sentiment analysis) to companies and public figures based on public and private sources of information, which would increase the accuracy of forecasting and risk assessment by 75%.

Proposed solution

To build an AI/ML solution for:

  • automatic collection of extensive data (>100 TB) from many sources - public (media, Twitter) and private (email, Dropbox, Google Drive);
  • NLP analysis of all the data obtained to identify entities (person, company, country, etc.) and their attitude (from negative to positive) with reference to time;
  • provision of a convenient search tool through an extensive amount of data;
  • provision of a user-friendly interface for building graphs of the entity's popularity and users' attitude towards it.

1. Gather, analyze, store the information in real-time received from the following sources:

Public:

  • NYTimes
  • Sec.gov
  • Seekingalpha
  • Twitter

Private:

  • Email
  • Dropbox
  • Google Drive
  • Amazon S3
  • Google Bigtable
  • SQL-databases
  • HTTP Upload form

2. Perform an NLP analysis to identify entities and sentiment value to them with reference to time.

3. Record the information received in separate databases for quick search, analytics and building graphs with reports.

4. Provide a user tool for building graphs of the entity's popularity and users' attitude towards it.

AI/ML Solution for Investment Risk Forecast and Assessment

5. One of the key features of this solution is that the flow of these processes is flexible, and an analyst can build the desired flow on his/her own using the user interface (UI), even without any experience with programming languages.

AI/ML Solution for Investment Risk Forecast and Assessment

Value delivered by Dysnix

This tool helps not only to collect and structure the information from completely different sources but also to get a complete picture of the changes related to the company's or person's popularity and people's attitude towards them over a long period of time. Moreover, this information can be used for the prediction and assessment of the investment risk.

Industry

Finance Services

Asset Management

Location
New York, USA
Project duration
2019
Our team
5 (CTO, Back-end Dev, Front-end Dev, DevOps, Data Scientist)
Technologies used

ETL: Apahe NiFi

Python

spacy.io

Tensorflow

Keras

Scrappy

Kubernetes

Kafka

InfluxDB

Services provided

Google Cloud Platform

Google Cloud Storage