Information Ratios, Manager Skill, and Predictive Limits

Summary

The document distinguishes a manager’s historical information ratio from a prediction that the manager will continue to outperform. It asks whether managers with high past information ratios tend to retain above-median ratios, but the response offers a caution rather than evidence of persistence: the information ratio itself should not be treated as predictive. A historical performance statistic describes results over the chosen period; it does not by itself establish durable skill or forecast future results.

The response also emphasizes that an information ratio is only as meaningful as its benchmark. It is more straightforward to assess a fund with a strict mandate and a closely aligned benchmark. For a flexible multi-strategy fund, finding a benchmark that captures the manager’s opportunities and exposures can be difficult, making the ratio hard to interpret. The document gives no empirical autocorrelation analysis, sample, or test for manager persistence. Its main lesson is therefore about the limits of inference from the metric and the importance of benchmark fit, rather than a quantitative conclusion about how stable information ratios are.

Key ideas

• A manager’s trailing information ratio is descriptive and does not alone predict future performance.
• Benchmark choice strongly affects how useful an information ratio is.
• Strictly mandated funds are easier to compare with a closely aligned benchmark.
• Flexible multi-strategy funds are difficult to assess because their exposures may not fit a single benchmark.
• The response provides no empirical test of information-ratio autocorrelation or manager persistence.

Tags

# Do managers information ratios exhibit autocorrelation? Ie. are they stable over time?


# Do managers information ratios exhibit autocorrelation? Ie. are they stable over time?












I'm reading through Active Portfolio Management, and I can't get my head around Information Ratio's real world applicability.

In table 5.6 it lists some empirical infomation ratios:

However, there is no discussion of whether these are stable over time. Is there any indication that a 90th percentile manager will have an above median IR in the near future?

It is all good to look back and say "Alice performed remarkably well in the trailing 5 years", it's a very different statement to say "Alice is shown to be more skilled by recent performance". The first statement is descriptive, the second is predictive.

Any thoughts on how best to interpret this?

## Answer by amdopt (score 1)

https://quant.stackexchange.com/a/70958

Nothing about IR is predictive (like most metrics used to analyze managers). The usefulness of IR is highly dependent on picking a benchmark to which the manager is highly correlated. Analyzing active mutual funds with strict mandates is easy, whereas a multi-strat hedge fund that can do whatever it wants is nearly impossible.

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.