本研究旨在探討總體經濟與市場情緒狀態能否提升加密貨幣橫斷面七日持有期超額報酬之樣本外可預測性，並區分統計可預測性、橫斷面排序能力與投資組合經濟價值三個層次。本文以 CoinMarketCap 日頻資料建立研究樣本，以 32 個個幣特徵（即各幣種自身的市場特性指標）為基準，依序納入總體經濟、市場情緒等狀態變數與特徵—狀態交互項，形成七組預測變數集，並比較 OLS、PLS、LASSO、Elastic Net、Random Forest 與 XGBoost 六種模型。實證結果顯示，加密貨幣橫斷面報酬具有樣本外可預測性；加入狀態變數後，Clark-West 檢定確認其具有統計上顯著的額外預測力，XGBoost 之樣本外 R² 由 1.14% 提升至 5.26%。然而，此一改善未同步反映於橫斷面排序能力，顯示狀態變數主要改善整體報酬水準預測，而非幣種間相對排序。SHAP 分析進一步顯示，總體經濟與市場情緒皆與模型預測報酬呈現具經濟直覺的非線性關聯型態。在投資組合層次，等權的五分位排序下的多空投資組合具有顯著風險調整後超額報酬，但市值加權後的結果明顯弱化，顯示預測收益可能集中於中小型幣種。整體而言，個幣特徵主要提供幣種間相對排序的依據，總體與情緒狀態則主要協助判斷整體市場報酬環境，兩者的貢獻性質不同。就實務意涵而言，狀態變數對市場擇時與風險控管具有參考價值，但在考量交易成本、市場容量與放空限制後，不應被直接解讀為可無摩擦執行的交易超額報酬。

This study examines whether macroeconomic and market sentiment states improve the out-of-sample cross-sectional predictability of seven-day cryptocurrency excess returns, distinguishing among three layers: statistical predictability, cross-sectional ranking ability, and portfolio economic value. Using daily CoinMarketCap data, we construct a research sample based on 32 coin-level characteristics — individual market indicators for each cryptocurrency — and sequentially incorporate macroeconomic and sentiment state variables along with feature-state interactions to form seven predictor sets, comparing six models: OLS, PLS, LASSO, Elastic Net, Random Forest, and XGBoost. The empirical results show that cryptocurrency cross-sectional returns are predictable out of sample. After incorporating state variables, Clark-West tests confirm statistically significant incremental predictive power, with XGBoost's out-of-sample R² increasing from 1.14% to 5.26%. However, this improvement does not carry over to cross-sectional ranking ability, suggesting that state variables primarily improve the prediction of overall return levels rather than the relative ranking across coins. SHAP analysis further reveals that macroeconomic and sentiment variables exhibit economically intuitive nonlinear association patterns with model-predicted returns. At the portfolio level, equal-weighted long-short portfolios generate significant risk-adjusted excess returns, whereas value-weighted results weaken substantially, indicating that predictive gains may be concentrated among small- and mid-cap coins. Overall, coin-level characteristics primarily provide information for ranking coins relative to one another, while macroeconomic and sentiment states mainly inform judgments about the overall market return environment — two distinct contributions that should not be conflated. In terms of practical implications, state variables offer reference value for market timing and risk management, but should not be interpreted as frictionless tradable excess returns once trading costs, market capacity, and short-selling constraints are taken into account.

Abdi, F., & Ranaldo, A. (2017). A simple estimation of bid-ask spreads from daily close, high, and low prices. Review of Financial Studies, 30(12), 4437-4480.
Amihud, Y. (2002). Illiquidity and stock returns: Cross-section and time-series effects. Journal of Financial Markets, 5(1), 31-56.
Bakshi, G., Gao, X., & Zhang, Z. (2024). Using machines to advance better models of the crypto return cross-section. Available at SSRN 4986862.
Bianchi, D., & Babiak, M. (2022). A risk-based explanation of cryptocurrency returns. Available at SSRN 3935934.
Bianchi, D., Babiak, M., & Dickerson, A. (2022). Trading volume and liquidity provision in cryptocurrency markets. Journal of Banking & Finance, 106547.
Cakici, N., Shahzad, S. J. H., Będowska-Sojka, B., & Zaremba, A. (2024). Machine learning and the cross-section of cryptocurrency returns. International Review of Financial Analysis, 94, 103244.
Clark, T. E., & West, K. D. (2007). Approximately normal tests for equal predictive accuracy in nested models. Journal of Econometrics, 138(1), 291–311.
Cong, L. W., Karolyi, G. A., Tang, K., & Zhao, W. (2022). Value premium, network adoption, and factor pricing of crypto assets. Available SSRN 3985631.
Corwin, S. A., & Schultz, P. (2012). A simple way to estimate bid-ask spreads from daily high and low prices. Journal of Finance, 67(2), 719–760.
Cosemans, M., & Frehen, R. (2021). Salience theory and stock prices: Empirical evidence. Journal of Financial Economics, 140(2), 460–483.
Datar, V. T., Naik, N. Y., & Radcliffe, R. (1998). Liquidity and stock returns: An alternative test. Journal of Financial Markets, 1(2), 203–219.
Filippou, I., Rapach, D. E., & Thimsen, C. (2024). Cryptocurrency return predictability: A machine-learning analysis. Available at SSRN 3914414.
Garfinkel, J. A., Hsiao, L., & Hu, D. (2023). Disagreement and the cross section of cryptocurrency returns. Available at SSRN 4345640.
George, T. J., & Hwang, C. Y. (2004). The 52-week high and momentum investing. The Journal of Finance, 59(5), 2145–2176.
Goyal, A., & Welch, I. (2008). A comprehensive look at the empirical performance of equity premium prediction. Review of Financial Studies, 21(4), 1455-1508.
Gu, S., Kelly, B., & Xiu, D. (2020). Empirical asset pricing via machine learning. Review of Financial Studies, 33(5), 2223-2273.
Jiang, X., Rodríguez, I. M., Jr., & Zhang, Q. (2022). Macroeconomic fundamentals and cryptocurrency prices: A common trend approach. Available at SSRN 4275078.
Lewellen, J., and Nagel, S. (2006). The conditional CAPM does not explain asset-pricing anomalies. Journal of Financial Economics, 82(2), 289-314.
Lin, M., Liu, Y., & Sheng, V. N. K. (2025). Analysis of the impact of macroeconomic factors on cryptocurrency returns: Based on quantile regression study. International Review of Economics & Finance, 97, 103757.
Liu, Y., Tsyvinski, A., & Wu, X. (2022). Common risk factors in cryptocurrency. Journal of Finance, 77(2), 1133–1177.
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.
Llorente, G., Michaely, R., Saar, G., & Wang, J. (2002). Dynamic volume-return relation of individual stocks. Review of Financial Studies, 15(4), 1005–1047.
Mohrschladt, H. (2021). The ordering of historical returns and the cross-section of subsequent returns. Journal of Banking & Finance, 125, Article 106064.
Wang, J.-N., Liu, H.-C., & Hsu, Y.-T. (2024). A U-shaped relationship between the crypto fear-greed index and the price synchronicity of cryptocurrencies. Finance Research Letters, 59, 104763.
Yang, D., & Zhang, Q. (2000). Drift-independent volatility estimation based on high, low, open, and close prices. Journal of Business, 73(3), 477–492.