Naive Forecasts Reign Supreme in Bitcoin Price Prediction

The world of Bitcoin price forecasting has become a colorful collection of methods, ranging from basic scarcity models to complex machine-learning systems.

Each approach enters the price-prediction contest against a simple yet effective opponent: naive forecasts that use only current market information.

A preprint by Carlos Baquero of the University of Porto found that no model has demonstrated durable superiority over the appropriate naive benchmark at horizons of one to six months across several market regimes.

The literature contains hundreds of papers, but none have been able to beat the naive forecast in a way that holds up outside of the period in which they were designed.