stocker

If I were to ask you: whats the best smell in this world? The logical answer is- the smell of money! With this AI based project, you can accurately predict where the real profit is!

Comment

Inspiration

With the market being in such a volatile condition as it is right now, tools like this can really help you out in a big way. As Sun Tzu said: "In the midst of chaos, there lies opportunity". The current market is chaos, so why not make it an opportunity?

What it does

This simple yet accurate stock predictor simply takes the last 10-13 years worth of stock data of any given stock, and builds a model on it. Then, it uses the same data, which gets split into 90:10 ratio, to train and test the model, in that order. Then, it gives, what it thinks will be the future trend in the form of a plot, all within the same kivy window.

How we built it

This was built using the knowledge that I gained from my AI-ML class in college this semester. The front-end was built within the same python script using kivymd. The main focus in this project was accuracy, and that, with the model layers, in combination with the optimizer, and the data processing, yields a result that is both accurate, and reliable.

Challenges we ran into

The biggest challenge was efficiency. Since its money that's on the line while investing, the biggest challenge was to get it accurate enough to the point where you could make money without the risk of losing it all. That was achieved beautifully by the AI in this project

Accomplishments that we're proud of

After about a week of testing the project, I can proudly say that this is able to grasp the market trends and give out the predictions that turn out to be very close to the actual trends, at only 10 epochs, it gives staggering accuracy.

What we learned

I've learnt that AI is capable of more than people know of. The kind of learning that would take a normal person years, such as reading the market trends, can be done by an AI within seconds of training.

What's next for stocker

Stocker is already a powerful tool that can be used to predict the stock prices. What could be even better, is that it regularly checks the prices of a specific set of companies every 3-4 hours, and when it detects a certain threshold of profit that can be made on a particular stock, it sends you a text to invest in that particular stock, along with how much to invest, which will be an inversely dependent function of the loss function for that stock.

Built With

Share this project:

Updates