Based on Kalman filter，this article use state space model to estimate time varying optimal hedge ratio of china's copper futures market and compare hedge performance of it with that of CC-GARCH model, VECM model，VAR model and OLS model．Hedging effectiveness is measured using the percentage of variance reduction and the percentage of sharp ratio reduction．We find that in terms of the two different measurement of hedging effectiveness，state space model based on Kalman filter perform significantly better than other models．The conclusion is robust to hedge periods．The results of the comparison of dynamic CC-GARCH model with static models depend on the duration of the hedge．VECM model perform worst and the hedging performance of VAR model does not significantly surpass that of simple OLS model．The risk of econometric models includes model·misspecification risk and estimation risk．Although the model-misspecification risk of advanced econometric model may be smaller than simple models，its estimation risk is greater and the total risk is uncertain．We find the solution of Kalman filter agrees with Bayesian rules，as suggestes Kalman filter approach outperform other models in dealing with estimation risk．