Explainable Economic Nowcasting

University of Sydney student research project developing transparent econometric models to predict Australian economic indicators 3-6 months ahead of official statistics

Economics & Econometrics Student University of Sydney
3-6
Months Lead Time
3
Core Indices
100%
Explainable Models
24hr
Update Frequency

About the Project

Oukson is a student-led research initiative exploring how alternative data sources can improve economic forecasting accuracy through rigorous econometric methods.

Research Focus

Mixed-frequency econometrics

Combining high-frequency alternative data with low-frequency economic indicators using MIDAS and state-space models

Causal inference

Establishing robust causal relationships between alternative data sources and macroeconomic outcomes

Policy evaluation

Real-time assessment of policy impacts using transparent, auditable models

Student Researcher

Currently pursuing Economics & Econometrics at the University of Sydney, with a focus on applied macroeconometrics and policy analysis.

📚 Major: Economics & Econometrics
🎓 University: University of Sydney
📅 Expected Graduation: 2027
🔍 Research Focus: Nowcasting & Policy

Three Core Indices

Each index leverages unique alternative data sources to provide early signals for key Australian economic indicators.

🛢️ Fuel-CPI Index

Predicts inflation trends by analyzing real-time fuel price data from NSW FuelCheck, capturing price transmission effects 2-3 months ahead of CPI releases.

Data Source: NSW FuelCheck API
Method: MIDAS Regression
Lead Time: 2-3 months
Update: Daily 6:00 AM

⚡ Power-CPI Index

Monitors electricity cost impacts on overall inflation using AEMO real-time pricing data, providing early warning signals for energy-driven inflation pressures.

Data Source: AEMO Real-time
Method: State-Space
Lead Time: 1-2 months
Update: Hourly

🎓 Student-Flow Index

Tracks international student trends using visa data and Chinese travel platforms, providing high-frequency nowcasts for Australia's education export sector.

Data Source: Visa + OTA Data
Method: VAR + Sentiment
Lead Time: 3-6 months
Update: Weekly Tuesday

Our Methodology

We prioritize explainable, white-box econometric models over complex black-box algorithms to ensure transparency and policy relevance.

Why Explainable Models?

❌ Black-Box AI Problems:

  • • Unexplainable prediction logic
  • • Regulatory unfriendly
  • • Fails during extreme events
  • • No causal understanding

✅ Oukson Approach:

  • • Every prediction has a causal chain
  • • Confidence intervals & uncertainty quantification
  • • Reproducible research methods
  • • Policy-maker friendly interpretations

Technical Stack

🐍

Python Econometrics

statsmodels, pandas, numpy, matplotlib

📊

Core Models

MIDAS, VAR, Local Projections, State-Space

🔄

Data Pipeline

Real-time APIs, automated cleaning, version control

📈

Deployment

Streamlit dashboards, FastAPI, GitHub Pages

Project Timeline

From MVP development to academic validation and beyond

September 2025

MVP Development

Building three core indices with basic data collection and modeling

85% Complete

October 2025

Academic Validation

Present findings to 2 economics professors for feedback and validation

December 2025

First Client

Secure first paying customer and refine product based on feedback

2026-2027

Scale & Graduate

Expand client base while completing Economics & Econometrics degree

Get In Touch

Interested in collaborating, providing feedback, or learning more about our research?

Academic Inquiries

For academic collaboration, methodology discussions, or research partnerships.

research@oukuaai.com

Business Inquiries

For pilot programs, commercial partnerships, or data access requests.

business@oukuaai.com

University of Sydney Connection

This research is conducted as part of ongoing studies in Economics & Econometrics. Academic supervisors and peer reviewers are welcome to reach out for discussion.