PREDICTION-BASED INVESTMENT STRATEGIES IN EUROPEAN BOND MARKETS.

Daily returns of the European corporate bond market are predicted using a penalized Lasso regression or Random forests. Predictions are utilized in investment strategies that allocate resources into risky position: interest-rate hedged corporate bonds or unhedged corporate bonds and risk-free position proxied by the Euro-Bobl futures that track the German government bonds. The strategies are more profitable with a lower risk than their passive alternatives, which only invest in corporate bonds (hedged or unhedged), but the daily rebalancing can be costly. Therefore, we examine the break-even transaction costs and suggest two approaches to lower the overall costs. Overall, even with costs, active strategies based on prediction can achieve higher returns with a lower risk. [ABSTRACT FROM AUTHOR]