Forecast Intraday FX Volatility for a London Liquidity Desk
Overview
What this challenge is about.
You receive 18 months of tick-level mid-quote data for six FX pairs plus a calendar of scheduled macro events. Resample to 1-minute bars, engineer realized-volatility features, and train at least two models (e.g., a gradient-boosted tree and a small temporal Convolutional Neural Network or CNN) to predict the next 30-minute realized volatility. Back-test out-of-sample on the most recent 3 months with a strict walk-forward split. Report root mean squared error (RMSE) and QLIKE loss against the EWMA baseline, conditional on macro-event windows. Wrap conclusions in a model risk memo that names the deployment risks (data drift, look-ahead bias, regime change) in plain language.
The Brief
What you'll do, and what you'll demonstrate.
Build and validate an intraday FX volatility forecaster that beats an EWMA baseline, especially around scheduled macro releases.
Earning criteria — what you'll demonstrate
- Apply machine learning to a realistic quant-finance forecasting problem
- Design walk-forward back-tests that avoid look-ahead bias
- Evaluate volatility models with both statistical and economic metrics
- Communicate model risk to a non-ML governance audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Data Scientist
End-to-end feature engineering on tick data with honest evaluation builds the core data-scientist muscle of moving from raw data to a defensible model.
This challenge sharpens
- feature-engineering
- time-series-forecasting
- backtesting
Machine Learning Engineer
Productionizing this notebook would be the natural next step; the reproducible pipeline and risk checklist are the artifacts MLEs hand to model risk teams.
This challenge sharpens
- python
- backtesting
- model-validation