Inspiration
Let me paint you a picture. Every single day, billions of dollars change hands on the stock market. And you know who's on the right side of those trades? Not you. Not us. The algorithms. The quant desks. The hedge funds with servers physically colocated next to the exchange just to shave off microseconds.
We got sick of it.
Why should the little guy always be eating the hedge fund's leftovers? The data is out there. The signals are screaming. Nobody's listening because there's too much noise. So we said fine. We'll build the machine that cuts through it.
That's SIGNAL100. And we're just getting started.
What it does
We scan the top 500 US stocks every single day. Every one of them. We pull in historical market data, quarterly filings, technical indicators, and we feed it all into a machine learning model that does one thing. It tells you where the price is going tomorrow.
Not a guess. Not a vibe. A prediction. Built on real data, trained on real markets, tested against real results.
That's SIGNAL100. Five hundred stocks. One signal. Tomorrow's edge, today.
How we built it
Black Coffee. That's how we built it.
We ingested real market data, prices, volume, momentum, every technical indicator you can think of. We fed it into a machine learning model and told it one thing. Find us the trades worth taking.
The model learned. It got smarter. It started spitting out signals, clean, ranked, confidence-scored, right into a dashboard that any trader can read in under three seconds flat.
No noise. No fluff. Just the signal.
The dashboard is sharp, dark, fast. It looks like something out of a trading floor on the 47th floor of a Manhattan skyscraper. Because that's exactly the energy we were going for.
Challenges we ran into
Here's the truth nobody tells you at hackathons. The market is the hardest opponent you will ever face.
We built a model that looked brilliant on paper. Backtested beautifully. Then we threw real data at it and it humbled us fast. The market has seen every trick. It punishes overconfidence harder than anything.
We had to tear it down and rebuild. Twice. We had to stop trusting what looked right and start trusting what tested right. That's the discipline. That's the difference between a team that blows up in six months and a team that's still standing in year five.
We also had to build a dashboard that didn't just work. It had to feel like money. Because if you're going to give someone signals that move their portfolio, the product has to inspire confidence the second they look at it.
Accomplishments that we're proud of
Honestly, just getting all 500 stocks into this thing is something we're proud of. That's not a small lift.
That's 500 tickers, hundreds of data points each, historical prices, quarterly filings, technical indicators, all flowing into one model, cleanly, reliably, every single day.
Most people scope that down. We didn't. We built it to handle all of it because anything less felt like leaving money on the table.
That's the accomplishment. We didn't cut corners. We built the full thing.
What we learned
Building SIGNAL100 full stack taught us more than any course ever did.
On the backend we learned how to build a data pipeline that actually holds up under real load. Ingesting 500 stocks worth of historical prices, filings, and indicators without it falling apart took real engineering. We learned the hard way that clean data is not a given. You have to fight for it.
On the ML side we learned that a model that performs in training means nothing until it performs in production. We learned about overfitting, about walk-forward validation, about the difference between a model that backtests well and one that actually generalizes.
On the frontend we learned how to take complex financial data and make it readable in under three seconds. That is harder than it sounds. Data visualization at this scale requires real thought about layout, performance, and user experience.
And across the whole stack we learned how to make every layer talk to each other cleanly. Database to model to API to UI. Getting that pipeline end to end, fast and reliable, was the real win. We came in knowing bits and pieces. We left knowing how to build a full product from the ground up.
What's next for Signal100
We are not done. Not even close.
The first thing on the roadmap is a better model. What we built works but we know it can be smarter. We are looking at deeper architectures, more features, better training data, and a feedback loop that learns from every prediction it makes. The model gets better every day we run it. We want to accelerate that.
The second thing is live tracking. Right now we predict tomorrow's price. Next we want to track it in real time. Live price feeds, intraday signals, alerts the second something worth acting on hits the screen. No delays. No lag. Just the signal the moment it matters.
The goal is simple. A platform that not only tells you where a stock is going tomorrow but watches it get there with you in real time.
We built the foundation. Now we build the future.
Log in or sign up for Devpost to join the conversation.