Abstract:
This research studies and develops trading strategies for fixed income securities based on bond pricing error and yield curve factor error, which is the deviation of the yield curve factors from their historical or expected values on a particular trading day. For both pricing-error based and factor-error based strategies, the trading signals are derived by two models which are the moving average and time-series models. This empirical study is conducted for the German and U.S. government bond markets. We found the mean-reversion effects on the pricing error of the bonds in both markets and found that the pricing-error based strategy with the moving average model outperforms the benchmarks in both markets. In addition, we found that the strategy based on the deviation of the level also outperforms the benchmarks in both markets, even though the mean-reversion effects are insignificant for the level of the yield curve. This suggests that a short-term average is a good mean forecast for the level of the yield curve. Furthermore, we found that a time-series model can improve the performance of a moving average model for the strategy based on the slope and curvature of the yield curve. Combining two single strategies together, we found some improvement in the standardized returns due to diversification in the German government bond market, while the benefit of diversification is not high enough to compensate the lower return in the U.S. government bond market.
