Bitcoin Price Prediction: Peer-Reviewed Evidence and Social Media Discourse

Battle Count: 77

Highly relevant to quantitative trading. The paper directly addresses the central question of whether any model can beat the naive baseline for Bitcoin at horizons relevant to trading (1 day to 6 months). Key findings with direct trading implications: (1) no model reliably beats naive at 1-6 month horizons across regimes, (2) daily predictability exists but may not survive transaction costs, (3) statistical significance does not imply economic value, (4) stock-to-flow and Metcalfe's Law models have failed formal testing, (5) cross-crypto spillovers offer limited exploitable edges. The methodological recommendations (walk-forward, multi-regime holdout, naive baseline, DM test) are directly applicable to any quantitative trading strategy evaluation.

This is a survey/methodology paper, not an implementation paper. It does not propose or implement a new model. The complexity lies in understanding the evaluation methodology taxonomy and applying the proposed standards to one's own research. The recommended evaluation protocols (walk-forward, multi-regime holdout, naive baseline comparison, Diebold-Mariano testing) are standard in financial econometrics and relatively straightforward to implement, though the multi-regime holdout design requires careful period selection.

As a survey paper, reproducibility depends on the underlying studies cited. The paper provides detailed tier classifications and study characteristics (Appendix Table 3) enabling readers to verify claims. However, the social media component relies on Grok searches with acknowledged opaque retrieval coverage and limited search reproducibility. No code or data repository is provided for the survey itself. The paper recommends code and data release for future work but does not provide its own.