I Spent Over 3,000 Hours Developing an AI-Powered Finance Assistant. The Good, The Bad, The Ugly, and The Now

The Ugly

I would consider NexusTrade my life’s work. It started off as a passion project primarily designed for myself. I called it “The Money Printer” because the idea was that by using an automated platform, I would “print money”. I remember creating a thread on Reddit’s algotrading subreddit. And the people LOVED it.

The platform was clunky, ugly, and rigid. It wasn’t extensible, and it was literally impossible to express the vast majority of ideas. But the people of Reddit loved it, and convinced me it was the coolest thing since sliced bread.

The love and enthusiasm I got from those comments ignited a fire in me. From that point on, I was obsessed. I decided to pivot and create a platform — not just for myself, but for others to use as well.

The Bad

I spent the next year or so making significant changes to the UI. I did this by myself, and I am no designer. I wouldn’t even consider myself a frontend engineer. But I was driven, and passionate about the problem I was solving. And I did the best I could to make a professional-looking application.

The UI was a lot better than the first version of the platform. And it was more configurable. It even had some revolutionary features, like the genetic optimization engine, that optimized trading strategies using genetic algorithms.

But it had its flaws. Backtests were slow, and often froze my computer if the duration was longer than 2 years. Subsequently, this caused most genetic optimizations to be impossible. You could only run a very small population size, and even then, it would take days. And while it was more configurable than “The Money Printer”, it still couldn’t express highly complicated ideas.

Want deeper insights on the flaws on NextTrade? Check out this article I wrote about the transformation.

With this realization, I decided to open-source my work and all but give-up. I wrote this article on Medium, and that was it.

I was pulled back in because of you guys.

The Good

The reception I received on that article reignited the spark I had from when I started the project. I decided to rebuild the entire system from scratch.

Rather than diving straight into the code, I spent a great deal of time architecting a better solution. I redesigned what “an algorithmic trading strategy” meant, and made it possible to express virtually any idea within a no-code user interface. At the same time, I also made the system faster by re-implementing the core trading logic in Rust, a language that’s as fast as C++, but safer for production applications.

I then started small quality of life improvements, like a notification system and a tracing system for strategy execution transparency. I thought of every little detail needed for a powerful investing platform.

This included figuring out how orders should work, especially in cases where a user submits an order at the same time as their automated strategy. I thought in-depth about genetic optimization, and how to control the process to optimize a parameter between certain bounds. I invested a significant amount of time simplifying the platform, creating traditional SaaS options (like a Forgot Password button), integrating an email system, and a whole lot more.

But the most important thing I did was pivot the platform yet again. I created Aurora, the AI-Powered Finance Assistant. Initially, she was only able to generate trading strategies from plain English, so users didn’t have to learn how to navigate the UI to do so. But seeing how valuable she was for the platform, I began to quickly add more features to her.

And with this came another seismic shift. NexusTrade was no longer an “algorithmic trading platform”. No, it became an “AI-Powered Investing Research Platform”.

The Now

Don’t get me wrong. The platform is more than capable of algorithmic trading. It offers the most feature-rich set of options for traders looking to automate their strategies in the market, and that feature-set will only continue to expand.

But from using the platform, I realized I didn’t use the automated investing features all that often. I mean, after creating and deploying a strategy, all you can really do is wait to see how it performs.

What I spent most of my time doing was financial research. And Aurora made this easy. I was able to compare stocks to one another quickly, by just asking her to perform an analysis.

This analysis taught me how to evaluate companies based on objective fundamental metrics, and how to formulate trading ideas based on company valuations. It’s one of the reasons why I decided to purchase Nvidia call options, even though most people would say it was overvalued, helping me gain over $15,000 within the past year.

And with this, I knew that my chat was capable of doing a lot more. Last week, I finished a design for an AI-Powered stock screener, and then quickly implemented an MVP shortly after that.

This stock screener will prove to be the most powerful and useful feature to date. Having the ability to sort through millions of records instantly is unbelievable, especially because its accessible to the average person by using natural language.

Concluding Thoughts

The only thing I’ve invested more time in than this project is my education. It’s my passion project; something that I look at in awe because of the quality of my work and the sophistication of the system. As a software engineer, I’ve built many systems in my life, but nothing that makes me more proud than NexusTrade.

I strongly believe that I am creating something truly innovative, and something that has the potential to revolutionize how people approach trading and investing. Yesterday, I wrote an article for my blog, Aurora’s Insights, that basically says this: in one year from now, utilizing AI with your investing is not going to be a “want”. It’s going to be a “need”. I’ve shown with my platform that AI vastly expedites the process of doing financial research. I meticulously designed the entire system to be fast, configurable, and have a suite of valuable features that any real investor would die to use.

I understand that these tools are novel, and because of that, people will want to wait for more social proof to make the decision to try it out. I think that time waiting is a waste. Being first at a new technology like this is often how you become the best, and learning how to use the technology now (before everyone else does) will put you at an advantage compared to your competition. NexusTrade is easy-to-use, insanely powerful, and free to get started. So what are you waiting for exactly?