On the predictability of fund manager performance
- 1 day ago
- 2 min read
If you are a fund analyst or selector, do you value predictability in a fund manager?
By “predictable”, I mean having a reasonable sense that, given the current market, whether a manager is likely to generate positive or negative returns, or outperform versus their benchmark and peers.
Predictable performance also means it is reasonable to rely on the historical fund performance in selecting funds. If we understand the drivers of a fund’s returns, we can form a more informed view of likely outcomes.
Fund ratings only work if the fund performance is somehow predictable. If performance were entirely unpredictable, ratings would have little practical value. Institutional investors like pension funds and insurers also demand a degree of consistency and predictability in investment outcomes when they pick managers.
Predictability in performance can come from how fund managers structure their investment processes. Investment firms impose disciplined, repeatable processes. A value manager, for example, is expected to apply that style and skills consistently. By design, such a fund will exhibit exposure to the value factor, which in turn makes its performance partly predictable.
Admittedly, this is a simplification. Many managers blend styles and adapt over time. But even then, firms require clear processes and defined objectives. Without some level of predictability, winning mandates from institutional investors or getting picked by wealth managers becomes difficult.
However, predictability has a downside: it invites replicability.
The rise of smart beta illustrated this clearly. Returns from many active strategies, once thought to be skill-driven, were shown to be replicable through systematic factor exposures, often at lower cost. This has intensified fee pressure, particularly as regulators push for clearer demonstrations of “value for money”.
A new challenge is now emerging: AI and large language models. If a manager’s decisions and outcomes can be modelled and predicted, what prevents them from being replicated?
A recent NBER working paper[1] , “Mimicking Finance” by Cohen, Lu and Nguyen [2], explores this directly. The authors find that:
71% of mutual fund trade directions can be predicted
Predictability is higher for managers with longer track records and in less competitive categories
Less predictable managers significantly outperform, while highly predictable ones underperform
Importantly, “unpredictability” here does not refer to erratic behaviour. It refers to the portion of performance that cannot be explained by common factors: beta exposures, macro conditions, or the business cycle. In other words, it captures genuinely idiosyncratic skill.
This raises a useful framing: the more a manager’s returns can be explained, the easier they are to replicate. The harder they are to explain, the more likely they are to represent true alpha.
I will return to this theme in the next Tricio Monthly Insights.
James Chu, CFA
Head of Investment Solutions
[1] See recent blogs from Joachim Klement: https://klementoninvesting.substack.com/p/unpredictability-creates-alpha-stock, and https://klementoninvesting.substack.com/p/unpredictability-creates-alpha-fund
[2] L. Cohen, Y. Lu, Q. H. Nguyen, Mimicking Finance, NBER Working Paper 34849, February 2026 (http://www.nber.org/papers/w34849)
Comments