MacroQuant 2026: Forecasting & Trading Gold and Silver

Background

Gold and silver are globally traded precious metals that serve as inflation hedges, safe-haven assets during market stress, and industrial commodities linked to economic growth. Unlike isolated asset forecasting problems, precious metals are deeply connected to global macroeconomic forces such as interest rates, inflation expectations, currency movements, equity market performance, volatility regimes, and liquidity conditions.

Core Objective

Participants must build an end-to-end quantitative system that forecasts future returns for Gold and Silver Futures across three horizons and converts forecasts into trading decisions.

• 1-day ahead return
• 30-day ahead return
• 6-month ahead return

Dataset Description

The dataset integrates global precious metals futures, currency markets, equity indices, volatility indices, commodity markets, U.S. macroeconomic drivers, and Indian macroeconomic indicators.

Data begins from January 1, 2005 and extends to the most recent available date.

Challenge Components

1. Multi-Horizon Forecasting

Build time-series models to predict Gold & Silver returns across different horizons.

• 1-Day Return = log(Pₜ₊₁ / Pₜ)
• 30-Day Return = cumulative log return over next 30 trading days
• 6-Month Return = cumulative log return over next 126 trading days

2. Systematic Trading Strategy

Convert predictions into daily trading signals and portfolio allocation decisions.

• Daily position signals (+1 Long, 0 Hold, -1 Short)
• Risk-managed trading strategies
• Transaction cost: 0.05% per trade

Evaluation Metrics

Forecasting Evaluation (45%)

• Primary: RMSE (Root Mean Squared Error)
• Secondary: MAE, R², Directional Accuracy
• Multi-horizon performance averaged

Trading Evaluation (55%)

• Primary: Sharpe Ratio (risk-adjusted return)
• Secondary: CAGR, Maximum Drawdown, Sortino Ratio, Hit Ratio

Scoring Breakdown

Forecast RMSE (Test)

30%

Sharpe Ratio (Test)

30%

Multi-Horizon Robustness

15%

Rules & Constraints

Data Split

Period	Usage
2005–2018	Training
2019–2021	Validation
2022–2025	Test

Key Rules

• No future data usage
• No external data unless explicitly allowed
• Code must be reproducible
• All preprocessing steps must be documented
• Trading strategy must incorporate transaction costs