lego-learner

Using machine learning and principles from finance to trade lego sets

The Data

This project uses a combination of data from two API's

• Current Price - Bricklink API as of 05/10/2023
• List prices and features - Brickset API as of 05/10/2023

Check it out on Kaggle

The Model

Used a random forest and neural network and contrasted resulsts across the two models

The Results

The models were able to predict list price and current market price extremely well from our selected features.

Our model was able to consistently beat the S&P 500 and an equal weighted portfolio across all lego sets produced in that year.

We analyzed feature importanace for determining list price.

We used the model to make a price forecast for the next five years.

Detailed results, plots, and analysis are all in poster.pdf.