Since about the 90’s, there has been a popular strategy among investment professionals to allocate 60% of their portfolio to equities and 40% to bonds. The theory is that when stocks fall, bonds rise, so investors can increase returns and decrease risk by diversifying across both asset classes. The strategy has fared well during the market crises of 2000 and 2008. However, during the 2020 market crash, many funds deploying this type of risk parity strategy suffered because equities and bonds crashed simultaneously.

The fluctuating prices of stocks and bonds causes the portfolio weight of each asset class to drift over time. To remedy this, the portfolio is periodically rebalanced. Although when rebalancing the portfolio, the weights don’t need to be fully corrected. In fact, the rebalancing mechanism can significantly affect a strategy’s performance. The tranche method is to only partially correct the weights when rebalancing. 

AQR and Bloomberg studied the differences between rebalancing techniques and found that during times of high volatility, fully rebalancing a portfolio led to over-corrections and higher turnover than the tranche method. Their study showed that rebalancing only 25% of the way to the target allocations made the portfolio take longer to return to the target allocations, but the process led to smaller trades.

We sought to reproduce these results and test the hypothesis that tranche rebalancing is more effective than fully rebalancing. So during Idea Streams episode #1, both rebalancing techniques were tested using the 60/40 risk parity strategy.

Our Process

For the equities component of the 60/40 strategy, we chose the SPY ETF. For the bonds component, we searched for government bond ETFs with the largest volume. Of the bond ETFs with the largest volume, SHY was chosen since it had the longest history to test. We wanted to backtest the tranche rebalancing strategy over the longest period possible to determine its performance during different market regimes and crises. With the inception date of SHY being July 22, 2002, we set the start of the backtest to January 1, 2003.

Setting the portfolio weights of each security to 60/40 is done by emitting 2 insights at the very beginning of the backtest, each with a 20-year duration. This sets the target weight for each security to be 60% and 40%, respectively. The portfolio construction model used in this algorithm was the InsightWeightedPortfolioConstructionModel, so the PortfolioTargets sent to the execution model always instruct the execution model to expose 60% of the portfolio to SPY and 40% to SHY.
To ensure the execution model performs tranche rebalancing instead of fully rebalancing the portfolio to the target weights, a custom execution model was created. The custom execution model places orders on a monthly basis. However, it only corrects the portfolio weights 1/12th of the way to the target weights.


To compare the rebalancing methods, we first created a benchmark backtest where the target weights of the portfolio are fully corrected at the start of each month. During the rebalances of March and April 2009, we noticed large portfolio churn. For instance, the image below shows the portfolio buying 487 shares of SPY in March and selling 304 shares of SPY in April.

In contrast, the image below shows the tranche rebalancing method resulted in smaller trades for SPY during the same time period. This rebalancing technique led to the algorithm purchasing 197 shares of SPY in March and buying 146 more shares in April. The trades were smaller and the portfolio avoided over-correcting during the two months.

There were several other metrics we noticed between the two rebalancing techniques:

Full Rebalancing Tranche Rebalancing
Fees $555.61 $486.60
Drawdown 35.1% 32.3%
Sharpe ratio 0.595 0.615

In conclusion, the idea that the tranche rebalancing is more effective than fully rebalancing the portfolio held up in our experiment. We saw a reduction in fees, drawdown, portfolio churn, and an increase in the Sharpe ratio. For a full walkthrough of how we built the Tranche Rebalancing Risk Parity strategy, watch the Idea Streams episode below.