Benchmark Timing Constraints in Active Portfolio Optimization
Summary
The document compares active portfolio optimization framed around residual return and residual risk with formulations that maximize active return relative to tracking error. It focuses on whether a manager should constrain portfolio beta to one to separate stock selection from benchmark timing. The answer distinguishes active return, defined relative to the benchmark, from residual return, defined after scaling benchmark return by portfolio beta. These coincide when beta is one.
The response characterizes the approaches as different objectives: the cited Grinold and Kahn framework emphasizes stock selection and discourages relying on benchmark timing, while the other cited texts optimize active return without an explicit beta constraint. It suggests applying the beta constraint when using those formulations to control benchmark timing. This is an opinion-based explanation rather than a worked optimization derivation; it does not specify the full constraints, risk model, or how to estimate beta in practice.
Key ideas
- Active return subtracts benchmark return, while residual return subtracts beta-scaled benchmark return.
- The two return measures coincide when portfolio beta equals one.
- A beta-one constraint can separate stock selection from benchmark timing in an optimization.
- The answer describes the cited frameworks differently but gives no full derivation or implementation details.
Tags
Full text
# How to deal with benchmark timing in quantitative portfolio management?
# How to deal with benchmark timing in quantitative portfolio management?
In Grinold & Kahn (2000), the authors emphasized the separation of stock selection and benchmark timing in active portfolio management. So if we avoid benchmark timing, the optimal portfolio's beta should be 1 and we need to add this constraint to our portfolio optimization problem:
where h_P is the vector of portfolio weights. The objective is the portfolio's residual return adjusted by residual risk [see p.139-p.142 in Grinold & Kahn (2000)].
But I can't find this constraint in other books on quantitative portfolio management such as Chincarini & Kim (2007) and Qian, Hua & Sorensen (2007). In these two books, the portfolio optimization problem is constructed as the maximization of portfolio's active return adjusted by tracking error without any constraint on portfolio's beta [see p.281 in Chincarini & Kim (2007) and p.35 in Qian, Hua & Sorensen (2007)]:
where f is forecasts of excess return, f_{PA} is portfolio's active return, h_{PA} is active weights, i.e. h_{PA} = h_P - h_B, and h_B is benchmark weights. (Solve this problem we can get h_{PA}, then h_P = h_{PA} + h_B)
Questions: What's the relationship between these two kinds of portfolio optimization problems? Why doesn't the "maximization of active return adjusted by tracking error" consider benchmark timing and constraint on portfolio's beta? How should we deal with benchmark timing in the framework of Chincarini & Kim (2007) and Qian, Hua & Sorensen (2007)?
Reference
Grinold & Kahn, 2000, Active Portfolio Management: A Quantitative Approach for Producing Superior Returns and Controlling Risk
Qian, Hua & Sorensen, 2007, Quantitative Equity Portfolio Management: Modern Techniques and Applications
Chincarini & Kim 2007, Quantitative Equity Portfolio Management: An Active Approach to Portfolio Construction and Management
## Answer by KaiSqDist (score 1)
https://quant.stackexchange.com/a/82180
Currently reading up on APM by Grinold & Kahn (2000) as well. Let me try to give my opinion. The active $r_a$ and residual $r_\varepsilon$ returns are defined differently:
$$r_a = r_p - r_b$$
$$r_\varepsilon = r_p - \beta_p r_b$$
where the two are the same if and only if the PM avoids benchmark timing $\beta_p = 1$ entirely.
What's the relationship between these two kinds of portfolio optimization problems?
I think they are different. The first (Grinold & Kahn, 2000) focuses on maximizing the active return without benchmark timing because the value-added (as the text suggests) -
"A more subtle reason, suggested in Chap. 6, "The Fundamental Law of Active Management," is that there is less chance of deriving substantial value added through benchmark timing." (Page 110/621)
Therefore, the optimization based on the authors focuses more on superior stock selection rather than benchmark timing (which is what the other two authors consider).
Why doesn't the "maximization of active return adjusted by tracking error" consider benchmark timing and constraint on portfolio's beta?
As shown above, the active return by construction does not have any constraint on $\beta_p$. My guess is that those authors (in the latter two books) do not decompose active returns into benchmark timing and stock selection.
How should we deal with benchmark timing in the framework of Chincarini & Kim (2007) and Qian, Hua & Sorensen (2007)?
Follow Grinold & Kahn (2000)'s technique.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.