Investment Performance Outlier Testing

Sean P. Gilligan, CFA, CPA, CIPM

January 29, 2020

Back to topic
GIPS Compliance

For any firm that aggregates portfolios of the same strategy into a composite, or otherwise groups portfolios by mandate, how do you know that each portfolio truly follows that strategy? The answer is outlier testing.

Why Utilize Composites?

The GIPS standards require firms managing separate accounts to construct composites, which aggregate all discretionary portfolios of the same strategy. However, even for firms that are not GIPS compliant, the use of composites is considered best practice when reporting investment performance to prospective clients. Composites offer a more complete picture than presenting performance of a model or “representative portfolio” – which usually leave prospects wondering whether the information is truly representative or if the portfolio presented was “cherry picked.”  

When creating and maintaining composites, firms must ensure that portfolios are included in the correct composite for the right time period – the period for which you had full discretion to implement the composite strategy for that portfolio. This is achieved by following a clearly documented set of policies and procedures for composite inclusion and exclusion. However, what happens when changes are made to a portfolio and those changes are not communicated to the person maintaining the composite?

In an ideal world, information in your firm would flow perfectly so that the person maintaining your composites knows exactly what is happening with the firm’s clients. In reality, client requests commonly result in small or temporary changes to the portfolio (e.g., halt trading, raise cash) that are not formally documented in the client’s investment guidelines or investment policy statement.

Without formal documentation of these changes, information may not flow down to the manager of your composites. While these minor or temporary changes may not affect the client’s long-term objectives, they may cause the portfolio to deviate from the strategy, requiring (at least temporary) removal from its composite. When these restricted portfolios are left in the composite, they often become performance outliers and create “noise” in the composite results. This “noise” prevents the composite from providing a meaningful representation of the portfolio manager’s ability to implement the strategy. This will also interfere with your prospective clients’ ability to analyze and interpret your performance results.

Why test for performance outliers?

Testing for performance outliers prior to finalizing and publishing performance results can help your firm remove this “noise” and can prevent costly errors in performance presentations. Firms that lack adequate composite construction policies and controls to ensure the policies are consistently followed often end up with errors in their composite presentations. In fact, it is very likely that errors in your performance exist. It is rare for us at Longs Peak to conduct an outlier analysis where no issues are found. Outlier testing should be completed quarterly and at a minimum, before any related verification or performance examination.

Many firms, especially those that are GIPS compliant, rely on their verifier to catch errors in their composites. We do not recommend this and suggest firms perform testing internally (or with the help of a performance consultant like Longs Peak) because:

  1. Verifiers only test a sample and will likely not catch all of your issues.
  2. Verification may happen months after the performance has been published. When errors are found, it may require redistribution of presentations with disclosures regarding prior performance errors.
  3. When verifiers find errors, they generally increase their sample size as well as their assessment of engagement risk. These two things lead to more time spent on the verification and a potential increase in your verification fee.

Even if not GIPS compliant, when firms use composites, regulators may test to ensure the composites are a meaningful representation of the strategy. In addition to improving accuracy, testing for performance outliers can help your firm‘s composites meet the standards expected by regulators.

How can performance outliers be identified?

Testing for performance outliers involves reviewing the performance of portfolios within the same composite or strategy to test if they are performing similarly. This testing allows you to flag any portfolios that may be performing differently so you can evaluate if their inclusion in the composite is appropriate.

For example, if your firm has a Large Cap Growth composite, testing performance outliers would involve compiling the return data for all of your Large Cap Growth portfolios, identifying which portfolios performed materially different from their peers, researching why they performed differently, and then taking the appropriate action if an issue is discovered. This may sound like a daunting task, but it doesn’t have to be. Let us walk you through this in more detail.

Some firms simply look at the absolute difference between each portfolio’s monthly return and the monthly return of the composite. While this may be straight forward, relying only on the absolute difference to determine outliers does not take into consideration the size of the return and the normal distribution of portfolio returns in the composite. For example, if you set a threshold to look at all portfolios that deviate from the composite return by 50bps, the result for a composite with low dispersion and a total return of 2% would very be different than a composite with higher dispersion and a total return of 20%.

In the outlier analysis Longs Peak conducts for clients, we use standard deviation in conjunction with a comparison of the absolute differences to identify the outlier portfolios that require review. Utilizing standard deviation allows us to identify portfolios that are truly outside the normal distribution of returns for each period. For example, reviewing all portfolios that are more than 3 standard deviations from the composite mean will provide the portfolios outside the normal distribution of returns for that period, regardless of the size of the return or the level of dispersion in that composite.

What to consider when reviewing outlier performance

The severity of the outlier

The larger the outlier, the more likely it is that the portfolio has an issue that would require it to be removed from the composite. We typically start by looking at the most extreme outliers first. Generally, we look at portfolios with performance periods flagged with +/-3 standard deviations from the mean return for the period. By addressing these first (including removing them if it is determined they do not belong in the composite), we are able to re-run the outlier test to assess what outliers exist without these extreme cases disrupting the analysis.

Once these extreme outliers are addressed, we move on to review the portfolios that are +/-2 standard deviations and even +/-1.5 standard deviations, if needed. We keep reviewing accounts with returns closer and closer to the composite’s mean return until we are consistently confirming that the portfolios do in fact belong in the composite and errors are not being found.

Each firm will be different in how much they need to drill down to get to a point of comfort that no more errors exist. If your composite is managed strictly to a model, the outliers will be very clear and easy to identify. If each portfolio you manage is customized, more research is often needed to determine if the outlier performance is simply a result of the portfolio’s customization or if the portfolio was included in the wrong composite.

How often the portfolio is an outlier

Longs Peak’s performance outlier reports show a portfolio’s performance, the number of standard deviations it is from the mean each month, and the number of months the portfolio was an outlier throughout its history in that composite. Our reports also show whether there was a cash flow during that period or not. The following are examples of outlier frequencies we evaluate:

Infrequent: If you see that a portfolio is only an outlier for one month and that month had a large cash flow, then you will know that the portfolio is likely only an outlier for that period because of the cash flow and, often, no further research is required.

Frequent: If you can see that the portfolio is an outlier for most of the months under review, then you will know that there is likely an issue with this portfolio.

As of a specific date: If you can see that the portfolio was not an outlier historically, but became a frequent outlier from a certain month forward, this may indicate that a restriction was added or that the strategy changed as of that period. The portfolio may then need to be reclassified to the appropriate composite or flagged as non-discretionary.

The most common causes of outlier performance and how to address performance outliers

Common causes of outlier performance:

  • Data issues – When outliers are extreme, it is likely that there is an issue with the data. Examples include a pricing issue that caused a material jump in performance or a late dividend hitting a portfolio that is closing and had most of its assets already transferred out. These issues are often easily addressed, depending on the circumstance of each case.
  • Cash flows – If a portfolio is only an outlier for one month and during that month the portfolio experienced a large cash flow, this is likely the reason for the outlier performance. If the portfolio had high cash for a period of time around the cash flow and the market moved during that period, this portfolio likely would perform differently than its fully invested peers. Nothing needs to be done in this scenario since the outlier performance is explained and there is no indication that the portfolio is invested incorrectly or grouped with the wrong portfolios.
  • Legacy positions or other client restrictions – If your clients hold legacy positions that you are restricted from selling or have other similar restrictions, this will likely cause these portfolios to perform differently when compared to their unrestricted peers. Depending on your composite construction rules, unless immaterial, these portfolios likely need to be excluded from the composite. With these portfolios removed, other outliers may appear that were not as noticeable when the restricted portfolios were included. It is important to refer to your firm’s composite construction policies, which should outline clear parameters for when restricted portfolios should be included/excluded in composites.
  • Portfolio categorized incorrectly – A portfolio may appear as an outlier because it was placed in the wrong composite. This often happens if a portfolio’s composite changed and it was not removed from its prior composite. If this is the case, the portfolio must be removed (after the change) and added to the new composite based on the timing outlined in your firm’s composite construction policies.
  • Portfolio managed incorrectly – Performance outlier analysis may help identify a portfolio that is managed to the wrong strategy. For example, it is possible that the portfolio is grouped with the correct portfolios, but the wrong strategy was implemented in the portfolio. This is one of the most important errors that performance outlier testing can identify because it means that the client is actually not having their money managed to the strategy for which your firm was hired. In this case, the portfolio would need to be rebalanced to the correct strategy. Likely, a review of the history would need to be conducted as well to ensure the client was not disadvantaged by the error.
  • High dispersion between portfolio managers – Especially when more than one portfolio manager is implementing the same composite at your firm, material differences may exist in the way they each manage the strategy. Outlier performers may be due to differences in the portfolio managers’ discretionary management. If the composite is being sold as one cohesive product, it is important to identify where the portfolio managers deviate and determine if they can work more closely together to avoid high dispersion or if the strategy should actually be run as two different products.

When researching outlier performance, keep in mind that, on its own, a portfolio’s performance deviating from its peers is not a valid reason to remove the portfolio from its composite. You need to determine the root cause of the deviation and remove the portfolio from its composite only if the root cause was client-driven. If the deviation was caused by tactical, discretionary moves made by the portfolio manager, the portfolio must remain in the composite as its performance is still a representation of the portfolio manager’s implementation of the strategy.

Ready to implement performance outlier testing at your firm?

While it is best practice to create a flow of information that will allow portfolios to proactively be included/excluded in the correct composite at the appropriate time, testing for performance outliers acts as a back-up plan to catch anything that was missed.

If analyzing your composite data to identify performance outliers is not something you have the resources to do internally, Longs Peak is available to help. Longs Peak offers both consulting and reporting services that can assist your firm with outlier analysis. Conducting outlier analysis should be done at least quarterly to help ensure your firm is managing your portfolios consistently and are reporting strategy or composite performance that is meaningful and accurate. Please contact us to discuss how we can help implement this practice for your firm.


If you have questions about investment performance, composite construction, or the GIPS standards, we would be love to talk to you. Longs Peak’s professionals have extensive experience helping firms with all of their investment performance needs. Please feel free to email Sean Gilligan directly at