Problem statement - Currently Investors face challenges in effectively managing their portfolios, particularly in bucketizing stocks for short, medium, and long-term investments, and maintaining optimal balance through timely portfolio rebalancing.
Collecting data from diverse sources like Stock exchange platforms, Textual data from news websites and user reactions from social media platforms like X and cleanse, processing them to build a model that can identify the top performing stocks based on actual trading numbers and user sentiment is currently a big technical challenge in Portfolio Management Industry.

We will demonstrate how easy it is to build an AI Data solution with our service that can identify the top performing stocks and solve these business/technical problems using Microsoft Fabric, Azure Open AI and LangChain.

The idea is to create a Portfolio Management Service in Microsoft Fabric that can be leveraged by financial advisors to use AI for Asset Management without having the deep knowledge of AI/ML.

What the solution offers -

1) Data -
a) Fundamental data of the stocks of various exchanges by time and volume for last two decades.

2) Data Cleaning and Engineering -
a) Capability to clean and normalize data for various recommendation models.
b) Adding more mathematical features like integral derivatives over the fundamental features.

3) Range of trained ML and AI Models -
a) They can train/reuse a combination of models like decision tree to get the top n stocks and their weightage to invest in for a period of time.
b) The software can use the capabilities of ChatGPT to eliminate few of the stocks from the selected stocks using sentiment analysis.
c) The CEO/LT of the company can get real time crisp positive, negative and neutral news of their company as well as competitors. This uses LangChain along with Azure OpenAI to perform real time sentiment analysis from above captured data.

4) Strong Back testing Framework -
a) The product will offer a strong unbiased back testing framework to verify that the model that the user came up with is working as per expected.

5) Create Automated PPT of Recommended Stocks
a) Create the ppt explaining why the given n stocks were selected and their financial analysis.

6) Tracking the portfolio -
a) The tracking mechanism will actively do news and sentiment analysis and recommend re-balancing of the portfolio. The tracking mechanism can be customized as per user needs.

We are proposing this service as a key component of the industry solution feature of Microsoft Fabric. Each of the aforementioned functionalities will be represented by its own dedicated tile in the Fabric project. This design empowers Asset Managers to seamlessly load all relevant data into the project, clean and normalize it, add custom mathematical fields, run pre-trained ML/AI models, conduct sentiment analysis on text data, and finalize the stock bucket. Dedicated tiles will display the criteria for stock selection, based on the previous steps, and provide alerts when portfolio rebalancing is needed. This comprehensive solution offers a robust, out-of-the-box toolkit, positioning itself as a one-stop solution for all asset management needs.