The forecasting literature has presented overwhelming evidence that the aggregation of heterogeneous expectations leads to improvements in forecast accuracy; however, outperforming a simple equal weight- ing scheme has proved challenging. This paper proposes an aggregation scheme of subjective bond return expectations based on the historical accuracy of professional interest rate forecasters. Our aggregate belief proxy outperforms equal weight and median weight combinations and is comparable to statistical projections even if its dynamics are quite different. With this measure at hand, we study the relationship between quantities of risk and compensation for risk and demonstrate a strong link to subjective expectations even if this is difficult to detect using realised returns.