Transaction vs holdings-based performance calculations
It’s important for your firm to know the pros and cons of each method.
Whilst both methods provide inputs to analysis providing a breakdown of the returns, enabling the effectiveness of various investment strategies to be evaluated and explained, it is important to understand that the latter is limited in accuracy and granularity – although is less demanding in terms of data requirements.
Both have their strengths and weaknesses; holdings-based calculations, for example, cannot be reconciled to performance returns. A transaction-based methodology can be used as an operational tool, unlike a holdings-based one, and can call out all sources of excess return, but is more difficult to implement and requires accurate and complete data, according to the CFA Institute.
But by working with software that can easily implement transaction-based calculations and offers robust data analytics tools, you can glean the insight from this more accurate calculation method. Simply put, transaction-based returns are generally known to be more accurate, in that it is possible to perfectly reconcile aggregated security-level transaction-based returns to total portfolio returns. By contrast, holdings-based methods are not able to do this, and research is beginning to come out that supports this. The Journal of Performance Measurement Spring 2018 edition featured a paper detailing the differences between the two types of attribution and concluded that large residuals may be apparent in holdings-based attribution.
With all of this said, there are certain features and details you should be looking for in portfolio analysis software. So, what are some of the questions you should be asking vendors about potential solutions when it comes to performance calculations?
What performance return calculations do you offer?
While we believe in, and much research shows, the superiority of transaction-based performance
calculations, it is important that you have choice when it comes to your attribution solution.
That’s why our Revolution product offers full transaction-based returns calculations as well as two holdings-based methods – one based on quantity and another on market value. Of the two, we consider the market value model superior, because it reconciles to the total assets of the portfolio. We aim to put control and choice in your hands.
Again, it should be reiterated that we do recommend implementing the full transaction-based methodology – including detailed cash and equivalent data – as the most accurate performance calculation.
How does your solution calculate transaction-based returns?
Not all calculation methods are created equal.
At the total portfolio or share class level, performance should be calculated using total asset return, based on market values and flows. This is in addition to allowing non transaction-based returns calculated from share prices and corporate actions.
At the security and strategy level, returns should be calculated using a true time-weighted return (for either a daily or elementary period) as well as using a modified Dietz return. Returns may be determined based on start, middle or end, or in fact, any fractional time of day transaction timings which can be controlled at multiple levels in the data and analysis structure. This includes transaction type along with individual transactions.
At both the security and strategy level, as well as the total portfolio and share-class level, Revolution calculates a number of different types of returns, at both the local and base level.
How do you calculate different return types?
Local and base returns are both calculated from their respective market and transaction values provided.
All of the currency, base and local returns take into account the actual exchange rates at the times of the valuations and thus are not approximations of end-of-day rates. This means that the calculations made in downstream attribution processing are in fact true currency rates and reflect the possible differences in return between portfolio and benchmark calculations.
We also offer trading and price returns, which are important not just for fixed income attribution processing, but for any asset type. Capital and income returns are also available; these can be quite useful when it comes to fixed income analysis, and also – like mentioned above – can be calculated for all asset types, taking into account changes in the accruals included in the valuations.
Finally, gross and net returns are included and are valuable for inclusion in various reporting, such as GIPS composite reporting.
How accurate are your calculations?
Something that many vendors who provide analysis software struggle with is being able to account for and recognize specific data conditions that could lead to no return being calculated or the calculation of an “abnormal return,” i.e. a return much different compared to the expected rate of return.
Such returns can indeed be mathematically correct but become onerous and time-consuming for an analyst to investigate or explain. Unfortunately, a number of vendor products struggle with this, only offering partial solutions that still ultimately require manual work on the part of the performance analyst.
That’s why it’s important to work with a system that has a comprehensive, transparent, accurate and – most importantly – automated solution that can easily reconcile these concerns when they arise, enabling you and your team to spend time on more value-add activities.
How do you handle data management
Measuring performance is ultimately a data management concern, rather than a calculation challenge.
A good vendor solution should be designed with simplicity, scalability and customization in mind to automate as much as possible and streamline workflow. The performance analyst should be able to use the system to easily investigate and resolve any concerns with the data.
With Revolution, market and index data are fully integrated, allowing you to focus on the analysis rather than data management tasks. Further, we aim to provide you with multiple options when it comes to reporting and extracting data from the platform. Revolution also supports multiple data models; this gives your firm the flexibility to choose which data model suits each portfolio and investment process.
While there are many factors and nuances to keep in mind when it comes to measuring performance attribution, the above considerations are vital for your firm to consider when choosing a vendor solution. Ultimately, the best kind of technology is that which helps you achieve strategic business goals in a seamless manner.
To further help you consider the evaluative points laid out above on a vendor-by-vendor basis, the next page is a guiding assessment checklist. If you’re interested in learning more about StatPro, click here to request your personalized Revolution demonstration and one of our product specialists will be in touch.
What attributes should an optimal transaction-based performance analysis system have?
- Allows for both base and local currency valuations and transactions to be imported and used in the calculation of returns.
- Has the flexibility to apply different transaction timing treatments or apply any form of ‘limit case’ adjustments to prevent generating abnormal returns under certain conditions.
- Provides flexibility to set or import transaction timing at the system, portfolio, transaction type and transaction level.
- Security transactions and their associated cash transactions are always treated with equivalent transaction timing.
- Provides transparency on any automatic adjustment that is provided.
- Features options to allow detailed cash to be loaded, reconciled and measured, in addition to the standard generation of cash lines.