GreenRock

Inspiration

We were interested in BlackRock's challenge statement which wanted to increase financial literacy. Since most of us having experiences in stocks or trading and downfalls, we thought it would be interested to extract the data from the Aladdin's API to see what kind of models or data they were using, and how they are able to accurately predict future portfolios. We knew that we wanted to make a web application that is helpful to the community educating them where they could run through a simulation to judge their financial judgements and how their portfolio would change.

What it does

Our project, GreenRock, extracts data from Aladdin's API and processes the financial data to help consumers make better financial decisions. We also incorporated an interactive simulation using machine learning models and past data extracted from the Performance Analysis API to predict future portfolio changes. Using the data, we also made an informational page, demonstrating the growth of your portfolio if you would have invested 10,000 dollars five years ago. This helps increase financial literacy because it shows users the impact of proper financial planning and its benefits. We also used neural networks to enable our program to determine whether stocks were underpriced of overpriced using a variety of factors.

How we built it

We used Java RESTful API to Build servlets that extract the data from the Aladdin's Performance Analysis API and we built a Java servlet. We used HTML, CSS, and Javascript to build an ornate and very user-friendly UI, as delivery is a key part of the impact of an application.

Challenges we ran into

One challenge we faced was integrating our front-end user input with our back-end to parse through the JSON dump. Since we used Java to parse through the API and retrieve the data, we had to get our user input value from our HTML fields from Javascript to Java, and project back to the UI. However, after several different attempts to resolve the problem included Ajax resolutions, IO attempts, we finally managed to successfully integrate our user input with our java servlet to successfully search the JSON dump and retrieve the corresponding level attribute.

Accomplishments that we're proud of

We are proud that we were able to successfully extract and understand the data from Aladdin Performance which the employees said was tricky itself, as the documentation is very vague, and this was not their actual complete API. Using this data, we were also able to successfully model future predictions by placing the user in the past to show the financial growth or decline of his/her investment. We then compared this prediction with the data we extracted to see that we were fairly accurate. We are also very proud of being able to determine if stocks are overpriced or underpriced using neural networks.

What we learned

We gained a greater knowledge of how BlackRock's API works. We all gained a better understanding of parsing through JSON dumps attempting to retrieve the "level" attribute which is what essentially predicts if your portfolio investment will increase or decrease. We learned about various events that can account to stock fluctuation. We also gained a lot of insight on how neural networks functioned and it was a entire different learning experience in a different aspect for every team member. Overall we gained a lot of knowledge regarding the Aladdin API which we can use in the future when making trading decisions, as well as sharpened our software skills.

What's next for GreenRock

Although the API that BlackRock gave us did was not their complete API with their future prediction data and models, using the previous data we extracted from the database we were able to test our prediction models by testing with the user being in the past. We could potentially use this simulation to predict future portfolio changes especially if integrated with advanced Machine Learning models which can track changes in news, merges between companies, or similar situations that have occurred in the past to predict how the stock will be affected. We could also expand upon the existing rushed neural network to further train it to reduce the standard of error in its predictions if stock is overpriced or underpriced.