Quandela/Challenge_Swaptions

Challenge Swaptions

Introduction

Financial markets are among the most demanding real-world environments for machine learning: data is noisy, incomplete, non-stationary, and often scarce exactly where predictions matter most. Interest rate derivatives are a prime example of this difficulty.

We introduce a challenge proposed during the Mil’HaQ Fest, the first AI and Quantum hackathon by Mila and Quandela. The problem is co-designed by Mila and the AMF (Autorite des marches financiers du Quebec) and deployed across multiple international hackathons. The challenge focuses on pricing swaptions using quantum machine learning, with an emphasis on time-series prediction and missing data imputation. The dataset presented here is purely artificial and was generated for this challenge.

More information on this challenge can be found on Quandela Training Center. And no prior background in quantitative finance is assumed, we will guide you all the way ;) .

Load From Hugging Face

Install the datasets library if needed:

pip install datasets

Load the level 1 (future prediction) data from the Hub:

from datasets import load_dataset

ds_level1 = load_dataset(
    "Quandela/Challenge_Swaptions",
    data_files="level-1_Future_prediction/train.csv",
    split="train",
)
print(ds_level1)

Load the level 2 (missing data prediction) data from the Hub:

from datasets import load_dataset

ds_level2 = load_dataset(
    "Quandela/Challenge_Swaptions",
    data_files="level-2_Missing_data_prediction/train_level2.csv",
    split="train",
)
print(ds_level2)