避險基金為金融市場中的一種投資商品，其特質是可透過基金經理人融合財務工
程與金融方面的專業知識操作，然而其具有不透明性且操作不易，對於一般投資者來
說具有相對高的投資風險，故避險基金在市場上持續交易的時間值得探討。而傳統上
可使用Cox 比例風險模型來分析在市場上的存活時間，但其有一限制為存活時間與設
限時間須獨立，但在金融市場中會影響避險基金存活時間的未必只有破產事件，經理
人操作手法或是環境因素都會使基金撤銷市場，故存活時間與設限時間可能是相依的，
若使用傳統的存活分析方法可能會使得預測出現誤差，使用關聯結構模型(Copulas)
可以進一步改進此問題。
在本研究中，會使用傳統的Cox 比例風險模型與關聯結構改進後的模型進行分析
並比較其預測結果。所使用的實際資料為Lipper/TASS 數據庫在西元1990 年到2018
年之間的避險基金資料，在實例分析前，會先進行模擬實驗，假設存活時間與設限時
間是相依的情況下生成資料並套用Cox 比例風險模型與關聯結構改進的模型，觀察何
種資料型態下關聯結構模型表現較好。

Hedge fund is an investment commodity in the financial market. However, due to
information asymmetry, it is difficult to operate and is associated with relatively high
investment risks for ordinary investors. Therefore, it is worthy to examine how the
hedge fund operated in the market. In the financial market, the survival time of hedge
fund may not only be affected by the bankruptcy event. The operation techniques of
managers or environmental factors might also cause the fund to withdraw from the
market. Therefore, the survival time and the censor time may be dependent. As a result,
the Cox proportional hazards model that assumes independent relationship between
survival time and the censor time may cause estimation error. In contrast, copulas model
may provide more accurate estimation as the copulas model is not constraint by the
independence constraint.
In the study, Cox proportional hazards model and the copulas model will be used to
analyze and compare results. The simulation experiment will be carried out prior to case
analysis, and simulation data will be generated under the assumption that the survival
time and the censor time are dependent, and the Cox proportional hazards model and
copulas model will be applied to observe which data type makes copulas model
performs better.

[1]Cox, D. R. (1972). Regression models and life‐tables. Journal of the Royal Statistical Society: Series B (Methodological), 34(2), 187-202.
[2]Emura, T., & Chen, Y. H. (2018). Analysis of survival data with dependent censoring: Copula-Based Approaches. Singapore: Springer.
[3]Fortin, I., & Kuzmics, C. (2002). Tail‐dependence in stock‐return pairs. Intelligent Systems in Accounting, Finance & Management, 11(2), 89-107.
[4]Fredheim, M. (2008). Copula methods in finance. VDM Verlag Dr. Müller.
[5]Harrell, F. E., Califf, R. M., Pryor, D. B., Lee, K. L., & Rosati, R. A. (1982). Evaluating the yield of medical tests. Jama, 247(18), 2543-2546.
[6]Hartmann, P., Straetmans, S., & Vries, C. D. (2004). Asset market linkages in crisis periods. Review of Economics and Statistics, 86(1), 313-326.
[7]Ishwaran, H., & Kogalur, U. B. (2007). Random survival forests for R. R news, 7(2), 25-31.
[8]Ishwaran, H., Kogalur, U. B., Blackstone, E. H., & Lauer, M. S. (2008). Random survival forests. The annals of applied statistics, 2(3), 841-860.
[9]Jaworski, P., Durante, F., Hardle, W. K., & Rychlik, T. (2010). Copula theory and its applications (Vol. 198). Berlin: Springer.
[10]Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations. Journal of the American statistical association, 53(282), 457-481.
[11]Kleinbaum, D. G., & Klein, M. (2005). Survival analysis: a self-learning text (Vol. 3). New York: Springer.
[12]Kalbfleisch, J. D., & Prentice, R. L. (2011). The statistical analysis of failure time data. John Wiley & Sons.
[13]Nelsen, R. B. (2007). An introduction to copulas. Springer Science & Business Media.