This article demonstrates how to use Amazon EMR and Apache Iceberg to backtest index rebalancing arbitrage strategies while avoiding look-ahead bias in financial research.

• Index rebalancing arbitrage exploits price discrepancies from ETF index changes by longing added stocks and shorting removed stocks
• Look-ahead bias occurs when future data inadvertently influences historical backtesting, leading to overly optimistic results
• Apache Iceberg tagging creates named snapshots of market data at specific points in time for accurate historical analysis
• Iceberg time travel enables querying data as it existed at tagged snapshots, preventing future information leakage
• Experiment tested three trade entry points: announcement day, effective date, and ETF holdings registration date
• Results showed effective date provided best returns across most holding periods tested
• EMR on EKS provides scalable infrastructure for managing entire investment research lifecycle
• Iceberg tagging also supports GDPR compliance and maintains data lineage through branches

The approach successfully addresses backtesting challenges in quantitative finance by combining EMR's processing power with Iceberg's data versioning capabilities to ensure research accuracy.