Quality Control: How to check for errors in your investment performance

May 5, 2021
15 min
Quality Control: How to check for errors in your investment performance

Recent investment performance calculation mistakes at Pennsylvania Public School Employees’ Retirement System (“PSERS”) have highlighted the importance of quality control reviews and raises questions about where risk exists, how these risks can be mitigated, and what role independent verifications should play in the quality control process.

What happened at PSERS?1

An error in the return calculation for Pennsylvania’s $64 billion state public school employee retirement plan has had serious implications for its beneficiaries and those involved in the calculation mistake.

In 2010, the plan, which was already underfunded, entered into a risk-sharing agreement where employees hired after 2011 would pay more into the plan if the return (average time-weighted return) over a specific time period fell below the actuarial value of asset (AVA) return of 6.36%.

In December 2020, the board announced that the plan had achieved a return of 6.38%, a mere 2 basis points above the minimum threshold. But in March the board changed its tune, announcing that the calculation was incorrect and the 100,000 or so employees hired since 2011 (and their employers) should have actually paid more into the plan.

What’s worse is PSERS also announced that the FBI is investigating the organization, although details of the probe have not yet been released.

According to PSERS, a consultant, that had calculated the return, came forward and admitted to the calculation error. But the board also said that it is looking into potential cover up by its staff. From what we know, at least 3 independent consultants were involved in providing data used for the calculations, calculating the returns, and verifying the returns. So, with all these experts involved, how could this happen and what can your firm do to avoid a similar situation?

Key issues to address in an investment performance quality control process

Firms should develop sound quality control processes to help identify errors before results are published. Often these processes either do not exist or are insufficient to identify issues. Following a robust quality control process that considers the key risks involved and then finds ways to mitigate these risks greatly increases the accuracy of presented investment performance.

Although we do not yet know the cause of the errors found in the PSERS case, we can highlight a few primary reasons errors occur in investment performance reporting. Primarily, errors found in published performance results are caused by:

  • Key Issue # 1 – Issues in the underlying data (e.g., incorrect or missing prices, unreconciled data, missing transactions, misclassified expenses, or failing to accrue fixed income)
  • Key Issue #2 – Mistakes in calculations (e.g., manual calculations that fail to match the intended methodology)
  • Key Issue #3 – Errors in reporting (e.g., publishing numbers that do not match the calculated results)

A robust quality control process should specifically address all three of these areas.

Considerations when designing a robust quality control process

Key Issue #1 – Issues in the underlying data

As they say, garbage in, garbage out. It is important to ask and address questions confirming the validity of data before it is used to calculate performance. Specifically, consider how the data used in the calculations is gathered, prepared, and reconciled before completing the calculations. Is there any formal signoff from the operations team confirming that the data is ready for use? Has a review of the data been conducted by an operations manager prior to this confirmation being made?

While deadlines to get performance published can be tight, taking the time to ensure that the underlying data is final and ready to use before performance is calculated can prevent headaches later on.

The following is a list of issues to look for when testing data validity:

  • Outlier performance – Portfolios performing differently than their peers may indicate a data issue or that the portfolio is mislabeled (i.e., tagged to a different strategy than it is invested in).
  • Differences between ending and beginning market values – Generally, we expect a portfolio’s market value at the end of one month and the beginning of the next month to be equal (unless using a system where external cashflows are recorded between months and differences like this are expected). Flagging differences can help identify data issues.
  • Offsetting decrease/increase in market value – Market values that suddenly increase or decrease and then return to the original value may have an incorrect price or transaction that should be researched.
  • Gaps in performance – A portfolio whose performance suddenly stops and then restarts may have missing data.
  • 0% returns – The portfolio may have liquidated and may no longer be under the firm’s discretionary management.
  • Very low market values – The portfolio may have closed and is only holding a small residual balance, which should be excluded from the firm’s discretionary management.
  • Net-of-fee returns higher than gross-of-fee returns – Seeing net returns that are higher than gross returns could indicate a data issue unless there are fee reversals you are aware of (e.g., performance fee accruals where previously accrued fees are adjusted back down).
  • Gross-of-fee returns and net of-fee returns are equal – If gross-of-fee and net-of-fee returns are always equal for a fee-paying portfolio, it is likely that the management fees are paid from an outside source (paid by check or out of a different portfolio). The returns labelled as net-of-fee in a case like this should be treated as gross-of-fee returns.

Key Issue #2 – Mistakes in calculations

Mistakes happen, but there are ways to reduce their frequency and impact. First, you’ll want to consider how manual your performance calculations are as well as the experience of the person completing the calculations.

Let’s face it, Excel is probably the most widely used tool in performance measurement, especially for smaller firms. While many firms likely find Excel to be a user-friendly tool for calculating performance statistics, it has its limitations. Studies have shown that up to 90% of spreadsheets contain errors and spreadsheets with lots of formulas are even more likely to contain mistakes. Whether it’s not properly dragging down a formula or referencing the wrong cell, fundamentally, the biggest problem is that users do not check their work or have carefully outlined procedures for confirming accuracy.

Although this may seem obvious, having a second set of eyes on a spreadsheet can save you from the embarrassing headache of having to explain errors in performance calculations. It is even better if this review is a multi-layered process. Having someone review details as well as someone to do a high-level “gut-check” to make sure the calculations and results make sense can reduce this risk. Depending on the size of your firm, this may be easier to accomplish with a third-party consultant, where you serve as a final layer of review.

Having this final “gut-check” can help prevent avoidable errors prior to publication. We find that this final “gut-check” is best performed by someone who knows the strategy intimately rather than a performance or compliance analyst, as these individuals may be too focused on the calculation details to take a step back and consider whether the returns make sense for the strategy and are in line with expectations.

If you use software to calculate performance, you can significantly reduce the risk of manual error, but due diligence should still be performed from time to time to manually prove out the accuracy of the calculations completed in the program. This does not need to be done every time but should be conducted when introducing a new software system and any time changes are made to the program.

Key Issue #3 – Errors in reporting

It may seem silly, but many performance reporting errors come from transposing strategy and benchmark returns in presentations or placing the return of one strategy in the factsheet of another. Therefore, it is important to consider how the final performance figures make it from the system or spreadsheet into the performance presentations. Are they typed? Copy and pasted? Or are the performance reports generated directly out of a system? It’s not enough to complete the calculations correctly, the final reports must also be accurate, so adding a step to review this is crucial.

A similar review process to the one described above can really make a difference, but ultimately, understanding the vulnerabilities of your performance reporting will help you design quality control procedures that address any exposure.

Calculations completed by external performance consultants

Whether performance is calculated internally or by a third-party performance consultant, the same key issues should be considered when designing the quality control process. Due diligence should be done on the performance consulting firm to evaluate the level of experience the firm has with calculating investment performance and what kind of quality control process they follow prior to providing results to your firm. This information will help you determine what reliance you can place on their procedures and what your firm should still check internally.

For example, outsourcing performance calculations to an individual or single-person firm likely necessitates a more in-depth review since this individual would not have the ability to have a second set of eyes on the results prior to providing them to your firm. However, even larger performance consulting firms with robust quality control processes may not have intimate knowledge of your strategies, meaning that, at a minimum, a final “gut-check” should be done by your firm prior to publication.

Reliance on independent performance verification firms to find errors

Many firms that hire performance verification firms rely on their verifier to be their quality control check; however, this may not be a good practice for a variety of reasons. If this is a common practice at your firm, you may want to check the scope of your engagement before relying too heavily on your verifier to find errors.

Verification is common for firms that claim compliance with the Global Investment Performance Standards (GIPS®). But even firms that claim compliance with the GIPS standards and receive a firm-wide verification are required to disclose that, “…Verification does not provide assurance on the accuracy of any specific performance report.”

This is because verifiers are primarily focused on the existence and implementation of policies and procedures. While their review may help identify errors that exist in the sample selected for testing, it specifically does not certify the accuracy of presented results. While the verification process is valuable and often does turn up errors that need to be corrected, regardless of the scope of your engagement, a robust internal quality control process is likely still warranted.

Firms that are not GIPS compliant may engage verification firms for various types of attestation or review engagements like strategy exams or other non-GIPS performance reviews. In these situations, the scope of the engagement may be customized to meet the needs (and budget) of the firm seeking verification. A clear understanding of exactly what is in-scope and specifically what the verifier is opining on when issuing their report is key.

Situations where the engagement entails a detailed attestation tracing input data back to independent sources, confirming that calculations are carried out consistently, and verifying that published results match the calculations, allow for heavy reliance on the verifier as part of your quality control process.

Alternatively, when the scope merely consists of a high-level review confirming the appropriateness of the calculation methodology, a much more robust internal quality control process should be applied.

Knowing the scope of the engagement your firm has established with the verification firm is an important element in determining how much reliance can put on their review and findings, which can then be incorporated into the design of your own internal quality control procedures.

Key take-aways

Mistakes happen in investment performance reporting, but a robust quality control process can greatly mitigate this risk. Understanding the risks that exist, designing processes to test these risk areas, and understanding the role and engagement scope of all consultants involved are essential items in designing a quality control procedure that work for your firm – and hopefully one that will help you avoid situations like what happened with PSERS.

If you are not sure where to begin, we have tools and services available to help. Longs Peak uses proprietary software to calculate and analyze performance. Our software helps flag possible data issues and outlier performers and also produces performance reports directly from our performance system.

In addition, our performance consultants are available to work with your team to help identify potential vulnerabilities in your performance reporting process and can help you develop better quality control procedures, where needed.

Questions?

If you would like to learn more about our quality control process or any of the services we offer (like data and outlier testing) to help improve the accuracy and reliability of investment performance, contact us or email Sean Gilligan directly at sean@longspeakadvisory.com.

1 For more information on PSERS, please see this article from the Philadelphia Inquirer.

Recommended Post

View All Articles
Key Takeaways from the 30th Annual GIPS® Standards Conference in Philadelphia
Key takeaways from the 30th Annual GIPS® Standards Conference, including AI’s role in investment performance and verification, SEC exam priorities, and proposed trade error guidance.
November 18, 2026
15 min

CFA Institute hosted the 30th Annual Global Investment Performance Standards (GIPS®) Conference on November 10 and 11, 2026 in Philadelphia, Pennsylvania. Thirty years in, the conference still does what it does best, bringing the performance community together to work through the issues that land on our desks. It was good to reconnect with familiar faces and to see many first timers in the room.

‍

The theme was "Shaping What is Next in Investment Performance," and one topic ran through nearly every session, even those not specifically dedicated to it: artificial intelligence. It came up in verification, the SEC discussion, the GIPS Standards update, and across the questions attendees submitted all week. While AI seemed to dominate many of the discussions, other themes included SEC examination priorities, proposed guidance on trade errors, performance-based fee calculations, the retailization of private markets, and the asset owner view of GIPS compliance. Here are the takeaways worth passing along.

‍

AI Took Center Stage

The audience polls reflected both enthusiasm for AI and caution about its use. Seventy-five percent of attendees expect AI and automation to have the greatest impact on verification over the next five years. At the same time, 84% are uncomfortable fully delegating final approval of client communications to AI. Roughly 80% also believe the GIPS standards need little adjustment in response to AI, with only minor tweaks necessary.

‍

CFA Institute's framing seemed reassuring for firms already experimenting. AI is treated as one more tool in the performance process, much like an accounting or performance system. Using it does not, by itself, jeopardize a firm's GIPS compliance. What matters is how firms oversee its use. CFA Institute is researching model bias and ethics to decide whether changes to the standards are needed.

‍

The practical sessions showed how much is already possible, from building efficient frontier analysis and tracking error decomposition to drafting attribution commentary with an AI agent in a fraction of the usual time. The consistent message from presenters was that the human must be in the loop. One speaker summed it up well: delegate tasks, but do not delegate responsibility. Some errors are harder to spot than an obvious hallucination, like when the numbers are right, but the interpretation is wrong. The expectation is a clear chain of evidence behind any result, with a qualified person reviewing it before it goes out.

‍

That logic carried into record keeping. When AI contributes to a calculation, a disclosure, or a GIPS Report, firms should retain enough information to support and recreate the result, as verifiers or regulators may ask for it. In practice that means documenting how AI is used in your policies and procedures, keeping version control over prompts, and retaining the underlying data. The accountability does not transfer to the software.

‍

CFA Institute is gathering industry input through its AI and performance survey. Responses will help inform any future updates to the GIPS standards. If you work in performance, we encourage you to participate.

‍

The GIPS Standards Update: What Is New and What Is Coming

As of October31, 2026, 1,647 organizations across 51 markets claim compliance with the GIPS standards, including all of the top 25 global firms and 85 of the top 100. Adoption remains strongest in the US, UK, Canada, Switzerland, and South Africa, with Saudi Arabia now in the top ten.

‍

The headline project is a proposed Guidance Statement on Trade Errors, out for comment through December 15, 2026. Trade errors happen at every firm, and the proposal addresses how to handle them in portfolios and composites alongside a firm's error correction policy. The key points in the proposed new guidance are that a trade error policy would be optional, much like a significant cash flow policy, while error correction stays mandatory for compliant firms. The guidance would add a narrow new reason a firm may remove a portfolio from a composite. Presenters were clear that this is not a way to drop a poor performer by calling it an error, and that firms would need consistent materiality thresholds. If you have opinions you would like to share about this guidance before it is finalized, please submit your comment letters to gipsstandards@cfainstitute.org by 15 December.

‍

CFA Institute also released a Guide for Return Attribution Reporting and the third edition of its paper Reconciling the GIPS Standards and the SEC Marketing Rule. Further out, a GIPS 2030 consultation paper is taking shape, with net-of-fee presentation, model portfolios, after-tax reporting, and AI all on the list. One idea under consideration would let firms present the net return relevant to a prospect, not only the most conservative net return, which would help align the GIPS standards with the SEC Marketing Rule.

‍

SEC Marketing Rule and the Current State of Exams

Mark Dowdell of the SEC Division of Examinations gave another candid session. The 2027 exam program is squarely retail focused, with attention on fiduciary duties, conflicts of interest, liquidity, complex products, and private credit valuation, along with cybersecurity and AI.

‍

On the Marketing Rule, recent staff guidance has eased two pain points. Firms now have a safe harbor to show investment characteristics and extracted performance on a gross basis, as long as the related composite or total-portfolio gross and net returns appear with equal prominence over the same period and the gross figures are clearly labeled. Firms also have more flexibility to present net performance using actual fees even when those are lower than the anticipated fee, with appropriate disclosure. Also, it was clarified that returns that comply with the SEC Marketing Rule, but do not meet the requirements of the GIPS standards, may be presented as supplemental information for a GIPS compliant firm.

‍

Examination findings continue to highlight familiar weaknesses. Firms are still cited for policies that were never updated for the Marketing Rule and for annual reviews that lack real testing. Hypothetical and predecessor performance remain trouble spots, and examiners continue to check incentive fee calculations against the governing documents. On AI, the message was straightforward: if you use it to help review or sign off on performance, make sure to have policies around it, include human oversight, and create strong supporting documentation, because the staff will ask exactly how the AI tools are used. The SEC also cited firms that exaggerate their use of AI.

‍

Performance-Based Fees: Small Details, Significant Implications

A session on performance-based fees drove home a single point: small differences in contractual language can materially change what a general partner earns. Compounding frequency, catch-up provisions, high-water marks, and the choice between European and American style waterfalls can each swing the result. Whether carried interest is structured as a fee or an allocation of profit can also change its tax implications.

‍

The session emphasized the need for shared understanding of the calculation. Lawyers draft the terms, administrators calculate the fees, and auditors verify the result. The intent, the language, and the calculation all have to agree. Many attendees asked whether any system can handle bespoke fee calculations well or if the work still ends up in Excel. The vast majority still seem to be using Excel for these calculations although there is some shift to software solutions. A pending SEC proposal could extend the use of incentive fees to a broader set of public funds, potentially bringing these challenges to more of the industry.

‍

Private Markets and Retailization

As private markets expand into wealth and retirement channels, the recurring message was that access is only part of the equation. As one panelist put it, private markets are a decision inside plan design, not a decision instead of it.

‍

The harder issue is valuation. In semi-liquid and evergreen vehicles, the net asset value is not just a reporting mark, it is the price at which investors subscribe and redeem. This makes valuation accuracy more important than ever. Appraisal-based values can smooth volatility and delay the recognition of losses, and every private market metric ultimately rests on an estimate of what is still held. The guidance for practitioners was to clearly separate realized cash returns from the manager’s unrealized estimates, disclose smoothing and benchmark limits, and tailor the disclosure to the investor. It was also noted that even when AI supports a valuation model, the valuer remains accountable for the result.

‍

Asset Owner Perspectives

A panel of GIPS compliant asset owners, including the Public Investment Fund of Saudi Arabia, MassPRIM, and the Pennsylvania Treasury, shared why they pursued compliance when no client required it. In each case the decision came from the top and was about institutional excellence and the confidence that independently verified numbers give boards, oversight bodies, and the external managers they hire. One panelist described compliance by osmosis: once the organization commits, the whole team pays closer attention to the inputs that drive performance, and issues get caught before they grow.

‍

A poll showed that most asset owner RFPs now ask whether managers are GIPS compliant, and a growing number ask across all asset classes. This pulls compliance into less transparent parts of the market such as private assets. The panelists were candid about shared challenges, including lagged private market valuations, but were encouraging about the path forward for more asset owners to claim compliance. They noted that compliance is achievable even with a small team when the right custodian, verifier, and consultant are in place.

‍

This is one of the threads we watched most closely. Asset owner adoption raises the bar for everyone, because an asset owner that has done the work expects the same discipline, ethics, and full disclosure from the managers it hires. Helping asset owners and the managers they evaluate meet that standard is a core part of our work, and we expect this community to keep driving demand for GIPS compliance across asset classes.

‍

Final Thoughts

Even with a large focus on artificial intelligence, the message from Philadelphia was that the fundamentals have not changed. AI can improve the speed and quality of performance work, but trustworthy reporting still depends on transparency, consistency, documentation, and human accountability. The firms that come out ahead will treat AI as leverage for their expertise rather than a replacement for it, and keep their policies, records, and review processes a step ahead of the tools.

‍

At Longs Peak, we help firms and asset owners put these principles into practice. We help build practical, well-controlled performance and GIPS compliant processes, from composite construction, performance reporting, and Marketing Rule presentations to preparation for verification. Increasingly, it also includes helping clients document and oversee AI use within those processes. If you would like to talk through how any of these developments affect your organization, please contact us.

‍

GIPS® is a registered trademark owned by CFA Institute. CFA Institute does not endorse or promote this organization, nor does it warrant the accuracy or quality of the content contained herein.

If AI does much of the performance work that taught us,what will teach the next generation of performance experts?

‍

I recently sat in on a webinar with Hope Morrissey of BigPi on building AI tools inside a firm. It was practical and timely: at Longs Peak, we are building out our own workbench, and much of the conversation mapped onto choices we are making right now.

‍

But the question I asked at the end wasn't about tooling. It was about people.

‍

Most of us learned this profession by doing grunt work. We linked monthly returns by hand in Excel. We built and rebuilt composites line by line. We chased a 3 bp difference for an afternoon and found a fee posted to the wrong month. We made mistakes, and someone more experienced caught them and explained why.

‍

That work is going away, and maybe it should. AI and automation increasingly handle reconciliation, first-pass review and draft disclosures faster than any analyst. What's left, and what matters most, is judgment and expertise.

‍

Which raises the uncomfortable part: if the work that built today's experts disappears, how do we build tomorrow's?

‍

The grunt work was never really about the math

Looking back, the spreadsheets weren't the lesson. They were the delivery mechanism for three things that nobody taught us directly.

‍

  • A feel for what normal looks like. After linking monthly returns for a few months or running VLOOKUPs to map the right data to the right composite, you get to know what a normal month looks like and, perhaps more importantly, when one looks off.
  • A map of where things break. Every error made by hand taught us where errors live: the cash flow cutoff, the fee accrual, the stale price, or the account that should have left the composite two months ago.
  • Ownership of the number. When you built it yourself, you could defend it. You knew every assumption inside it.

That is the raw material of judgment. AI doesn't remove the need for it. It removes the accidental way we used to acquire it. So, the answer isn't to preserve busywork for its own sake. It's to identify what the grunt work was teaching and teach it on purpose.

‍

Judgment is the new edge

Shortly after the webinar, I came across a LinkedIn post built on comments NVIDIA CEO Jensen Huang made on the podcast A Bit Personal. Huang pushed back on the usual definition of smart: "somebody who's intelligent, solves [technical] problems." His view: "I find that's a commodity." The people he values are technically astute but can also "infer the unspoken" and "see around corners."

‍

That lands hard in performance measurement. For two decades our credibility rested on getting the math right: time-weighted returns, composite construction, dispersion, three-year standard deviation. Systems already do most of that. AI is now taking the rest.

‍

What remains is the judgment call, the “unknowables.” Is this account still discretionary? Does this cash flow policy make sense for how the strategy is actually marketed? Is this benchmark still the right comparison after the strategy drifted (or the benchmark got too concentrated)?

‍

Those questions don't have a formula and often require interpretation rather than automation. They have experience behind them. And that brings us back to the training problem.

‍

Rebuilding the apprenticeship on purpose

Historically, nobody had to think very hard about how someone learned the job. You hired smart people, gave them the work, and the work did the teaching. Judgment showed up eventually, usually starting somewhere around the end of the first year.

‍

That approach falls apart once the work itself is automated. If we want people who can make good calls, we have to create the experiences that used to happen by accident. I don’t think we have this fully figured out, but these are the five things I keep coming back to.

‍

1. Do it by hand once, then let the machine do it

‍

I still think every new analyst should calculate both time-weighted and money-weighted returns by hand, link a quarter of monthly returns, and put together a small composite from scratch, at least once. Nobody is going to do it that way in production. The point is that it's very hard to supervise a process you've never done yourself. If you don't know where a cashflow belongs in the formula, you won't notice when something is in the wrong place.

‍

Once they've done it, have them run the same data through your automation tools and compare. Sometimes the results match. If they don't, working out why (a different day-weighting convention, a fee treated differently, a cash flow on the wrong date) teaches more about how th calculation really works than any training deck I've seen.

‍

2. Make the reviewer the new entry-level job

‍

For most performance teams, the junior job was to produce the number and someone more senior checked it. Now the machine increasingly produces the number, so the junior job becomes checking it. That's a harder place to start, because we're asking people to review work they may not have spent years doing themselves.

‍

One way to close that gap is to build exercises out of past results. Take AI-generated output and plant real errors (built from anonymized or synthetic examples) seen over the years: a mispriced security, a trade booked a day late, a fee posted to the wrong month, an account that should have left the composite and didn't. Ask newer team members to find them before a senior reviewer does. They may not catch everything at first, and that's fine. Talking through what they missed is where the pattern recognition starts to form, much the way it formed when those mistakes were made by doing the work by hand.

‍

3. Write down the instinct before you investigate

‍

Experienced performance professionals can usually tell something is off before they can explain why. Dispersion that jumps dramatically in a calm quarter. An account that is consistently a composite outlier. A composite return that beats every account in it, by a lot. That instinct is valuable, but it isn't always right. Sometimes it's just bias, and the number is actually fine.

‍

A simple habit helps. When something looks wrong, have the analyst write down what they think is going on before they start digging. Once they've found the answer, have them go back and note whether their first read held up. After a few months they should have a real record of where their instincts are reliable and where they aren't. When those notes are tracked and talked through as part of training, senior team members can see how someone reasons, not just whether they got the right answer.

‍

4. Teach people to hear what isn't in the file

‍

Much of our value sits in the gap between the data and what's happening at the firm. An account quietly drops out of a composite and nobody mentions why. Marketing starts promoting a new strategy before anyone has asked whether it needs its own composite. A portfolio manager mentions "a bit of a style tilt" in passing on a call. None of that is in the file, and AI can only work from the context we provide. If no one recognize what matters and brings it to the table, the tool will never know it was there.

‍

That kind of listening is not new, but it is hard to teach in a classroom. The best way I know is to bring junior team members into the rooms where it happens, early, even if they don't say a word. On an in-house performance team, that might be the monthly attribution review with the portfolio managers, the product committee when a new strategy is on the table, or the kickoff call with your verifier. On our side as consultants, it's client calls and verification kickoffs. In-house teams actually have an advantage here because they sit down the hall from PMs, operations and compliance, and an informal conversation is often where changes show up. Encourage juniors to participate in those conversations instead of working only from the data feed.

‍

Either way, debrief afterward. Ask what they heard that wasn't in the data, and what they would follow up on. The first few answers maybe thin, so share what you heard. Over time they'll pick up on things they used to overlook. That's probably one of the clearest signs that someone is developing real judgment.

‍

5. Turn your own history into the curriculum

‍

Our best training material is our own history of judgment calls. Build a library of real cases (anonymized of course): the facts, what was decided, why, and what you'd do differently now. Include the ones you got wrong.

‍

Run "what could go wrong" exercises together to pass judgment along. Before a verification, exam, or new factsheet goes out, get the team together and ask: "It's six months from now and this went badly. What happened?" Have everyone write their answers down before anyone shares. Newer staff hear what experienced people worry about and why: an undocumented policy change, hypothetical performance shown without the required policies, a net-of-fee calculation that doesn't tie to the fee schedule. That's knowledge that usually takes years to pick up by accident.

‍

What this asks of leaders

This isn't only a training problem. It's a leadership choice about where the time AI saves actually goes.

‍

  • Reinvest the hours. If automation saves a senior reviewer ten hours a month, some of those hours belong to teaching. Otherwise, efficiency quietly eats the pipeline of future experts.
  • Hire for judgment. Give candidates a messy composite file and ask what worries them, rather than screening on credentials alone.
  • Build tools that explain, not just answer. As we build our workbench at Longs Peak, one question I keep asking is whether our tools show its reasoning. A tool that explains why it flagged an account teaches. A tool that just flags it doesn't.
  • Measure what matters. Performance teams measure investment results for a living, yet rarely measure their own judgment. Whether you sit inside an investment firm or advise one, track the issues your team caught before the verifier, auditor, compliance or a regulator did, and who caught them. Over time, that tells you more about how your people are developing than how quickly the reports go out.

‍

The experts of the future won't learn the way we did

That's not a loss if we're deliberate about it. The firms that get this right will use AI to move people faster from producing numbers to questioning them, and they'll treat that shift as a training design problem, not a happy accident.

‍

I don't think anyone has fully solved this yet. We’re trying to be intentional about our future pipeline. I'd like to hear how other performance teams are approaching it. What are you doing to build judgment in people who may never link a return by hand?

‍

And as always, reach out if you want help rethinking how your team builds judgment in performance measurement at hello@longspeakadvisory.com.

‍

‍

Sources

  • Jensen Huang quotes as reported by Upworthy (Feb 3, 2026), from his appearance on A Bit Personal with Jodi Shelton.

‍

Every so often we hear it: "Oh, you're one of the performance software tools."

It's an understandable assumption. We work in the same world as the software providers: composites, Global Investment Performance Standards (GIPS®), performance data, and the systems that produce all of it. From the outside, it can look like we're all selling the same thing.

‍

We aren't. And if you're deciding how to support your performance and GIPS compliance work, the difference is worth understanding.

‍

The short version: Longs Peak is a team of people who do the work with you, or for you. We're not a platform you log into. Behind the scenes, our tools help our team deliver faster, more consistent service. But no number leaves our hands without a real performance expert standing behind it.

‍

Software Gives You a Capability. Someone Still Has to Run It.

‍

A software provider hands you a tool. A good one can be powerful, and those systems offer a ton of capabilities. But once the tool is installed, the work to run it lands on your desk.

‍

Someone on your team still has to load the data, set up and maintain the composites, review the results, chase down the account that looks off, and know the GIPS standards well enough to make the judgment calls. Software can calculate a return. It can't decide whether a new account belongs in a composite, explain why a number moved, or figure out what to do when a custodian changes its data feed.

‍

What We Actually Do

‍

At Longs Peak, we get our hands dirty, so you don't have to. In practice, that means:

‍

  • We manage your composites. Membership decisions, inclusion and exclusion, periodic updates, and the returns that come out the other end. We work alongside     members of your team to get those questions answered and make the updates for you. The work gets done by a person who does it every day.
  • We act as your GIPS expert. When a new strategy launches, a policy needs revisiting, or a gray area comes up, you have someone who knows the standards and has seen the situation before.
  • We dig into your data. We look for outliers, accounts that are breaking composite rules, and other anomalies, before they end up in your reported performance and get caught in verification. Not only do we help identify potential issues, but we work with your team to help create the paper trail to provide an explanation if questioned by a verifier.
  • We help with system issues. When your system isn't producing what you need, we get inside it and help work out why.
  • We design the output. Once the data work is done, we turn it into beautifully designed, client-ready factsheets and reporting. And we understand this information from a performance, compliance, and design perspective.
  • We help you communicate what matters most. For your prospects or a board of directors: which statistics support your story, and which just add noise? Those are communication choices, and we help you make them.

None of this is advice delivered from a distance. We maywork with you as a remote team, but we're not remotely hands-off. We work insideyour data, your systems, and your deadlines, the same way a colleague down thehall would. And when something needs fixing, we roll up our sleeves and fix it,so it’s off your plate and your to-do list.

Ways to Work With Us

‍

Every firm is set up differently, so we don't force a single model.

‍

  • Fully outsourced. We run composite management on your behalf, start to finish. For firms without a dedicated performance team, this is often the cleanest answer.
  • Part of your team. We work alongside your people, taking on the pieces that are heavy, specialized, or just hard to staff. To your colleagues, we're simply part of the performance team.
  • Your GIPS expert on call. Your team runs the day-to-day, and we're there for the questions, reviews, and judgment calls that need someone who has seen it all before.
  • Your marketing team. Your data is done and ready to be put to work. We turn it into polished factsheets and reporting for prospects, consultants, and boards, and help you decide which parts of your story deserve the spotlight.

Many clients start in one place and shift as their needs change. That flexibility allows us to grow alongside you and offer as much or little support as you need.

‍

Why This Matters

‍

Most firms don't struggle with GIPS compliance because they misunderstand a requirement. They struggle because the knowledge lives in one person's head, the team is stretched thin, or a system issue has been quietly sitting in the queue for months.

Software doesn't fix any of those problems. People do. Having an experienced team involved reduces key-person risk, frees your own staff for higher-value work, and means fewer surprises when verification or a regulatory review comes around.

‍

A Question Worth Asking

‍

If the person who runs your performance work were out for a month, who would pick it up? And if the honest answer is "we'd figure it out" or you simply don’t have an answer, it may be time to think about who else should be in the room.

‍

We're happy to talk through where you are today and what kind of support would actually help. No software demo required. Email us at hello@longspeakadvisory.com.

‍

GIPS® is a registered trademark owned by CFA Institute. CFA Institute does not endorse or promote this organization, nor does it warrant the accuracy or quality of the content contained herein.