Bloomberg: Pension不一定是按日历rebalance,也有可能根据return objective rebalance,涨的过多就卖
• One way to test the hypothesis of whether pensions rebalance quarterly is to determine whether signals derived from quarter-to-date performance offer superior forecasting ability to those derived from monthly performance for the final month of each quarter. To take a look at this issue, I employed a similar simple back-testing protocol to that used in my original study from 2017. Specifically, I looked at the quarter- and month-to-date relative performance of the SPX and Bloomberg Barclays U.S. Aggregate Bond Index with one week to go before the end of the quarter and calculated the P/L from going short one unit of the outperformer and long one unit of the underperformer.
• Now, in most cases the signals from the two methodologies were identical, as indeed they are this month as well. Yet insofar as any discrepancies between the two show up, they suggest that the monthly signal is the more accurate. I tested across three different time frames, and generally speaking the monthly signal generates somewhat superior returns, a slightly higher hit ratio, and consistently stronger correlations with price action in the last week of the quarter.
• Another way to look at things is through the prism of extremes. Since 1990, the 10 largest quarters of equity outperformance over the course of the three months have seen equities continue to outperform during the final week to the tune of 15 bps. Looking at the 10 largest periods of equity outperformance for the final month of the quarter, equities have underperformed over the final week by 93 bps.
• Now there is one caveat to all of this that suggests that pensions and endowments don’t totally ignore the calendar. Since 1990, the annualized return of trading signals from months that are not a quarter-end (0.25%) are paltry compared to monthly signals from quarter-end months (5.8%.) So one way to look at the apparently superior forecasting of monthly signals above is that it is simply noise.
• Yet my pension friend is adamant that the sort of epic signals that rebalancing models suggest are at work simply don’t exist. Of course, that rather raises the question of how and why markets tend to mean-revert at month- and quarter-end if it isn’t allocators who are doing the trading. The next few days should provide an interesting test case of the theory, in any regard. By this time next week I guess we’ll know if Dr. Evil has gone to work for a pension fund in his retirement.
#stocktalk
• One way to test the hypothesis of whether pensions rebalance quarterly is to determine whether signals derived from quarter-to-date performance offer superior forecasting ability to those derived from monthly performance for the final month of each quarter. To take a look at this issue, I employed a similar simple back-testing protocol to that used in my original study from 2017. Specifically, I looked at the quarter- and month-to-date relative performance of the SPX and Bloomberg Barclays U.S. Aggregate Bond Index with one week to go before the end of the quarter and calculated the P/L from going short one unit of the outperformer and long one unit of the underperformer.
• Now, in most cases the signals from the two methodologies were identical, as indeed they are this month as well. Yet insofar as any discrepancies between the two show up, they suggest that the monthly signal is the more accurate. I tested across three different time frames, and generally speaking the monthly signal generates somewhat superior returns, a slightly higher hit ratio, and consistently stronger correlations with price action in the last week of the quarter.
• Another way to look at things is through the prism of extremes. Since 1990, the 10 largest quarters of equity outperformance over the course of the three months have seen equities continue to outperform during the final week to the tune of 15 bps. Looking at the 10 largest periods of equity outperformance for the final month of the quarter, equities have underperformed over the final week by 93 bps.
• Now there is one caveat to all of this that suggests that pensions and endowments don’t totally ignore the calendar. Since 1990, the annualized return of trading signals from months that are not a quarter-end (0.25%) are paltry compared to monthly signals from quarter-end months (5.8%.) So one way to look at the apparently superior forecasting of monthly signals above is that it is simply noise.
• Yet my pension friend is adamant that the sort of epic signals that rebalancing models suggest are at work simply don’t exist. Of course, that rather raises the question of how and why markets tend to mean-revert at month- and quarter-end if it isn’t allocators who are doing the trading. The next few days should provide an interesting test case of the theory, in any regard. By this time next week I guess we’ll know if Dr. Evil has gone to work for a pension fund in his retirement.
#stocktalk