Working in quantitative research over the Summer, things that seemed relatively simple such as finding stock data required a lot of syntax and code that was not intuitive to get results. We hoped to fix this.

What it does

It is a website which takes in voice commands to return in-depth stock data or results from back tested trading strategies.

How we built it

We built a website that uses javascript to record audio from the user and save it. This audio is processed by the Google cloud for speech to text transcription. A python program anlyzes the meaning of the text and looks through ticker data which is passed into plotly to create a .html graph.

Challenges we ran into

We ran into problems incorporating the python back end with our website as it was built with PHP. We got around the problem by having the website automatically download a file rather than pass it to a python program.

Accomplishments that we're proud of

We are proud thagt we were actually able to incorporate trading algorithms not just metrics plotting. We are also proud that we have a UI and not just a terminal. The goal of our project was ease of use for trading "beginners" and that helps substantially.

What we learned

If we plan on incorporating a python back end, we need to be more mindful of website structure and plan it out ahead of time. We also learned about API authentification and natural language frameworks.

What's next for Talk To Trade

We need more algorithms and more stocks to analyze. For a production level tool we need to add in a lot of aliases for the various stocks. We wouild also like to make it useful for people who are actively trading by adding in the capability to purchase and sell stock from the website.

Share this project: