ETF prices have a richer set of drivers than the equities they're built from — creation/redemption activity, basket arbitrage, dealer inventory, and the interaction with underlying constituents all shape intraday price formation. This paper looks at how modern AI techniques can be applied to short-horizon ETF price forecasting, framed by the constraints and microstructure realities I observed during my years working alongside ETF Market Making technology at Citadel Securities. It's written for an audience that already understands the mechanics of ETFs and wants to know where machine learning meaningfully changes the forecasting picture — and where it doesn't.
Read the paper: Predictive Power in ETF Markets: Leveraging Artificial Intelligence for ETF Price Forecasting.