Portfolio reporting: how to report return, risk, attribution and costs properly

In depth11 min read · 12 sources

Good portfolio reporting answers four questions: how much did the portfolio earn, was that good compared with a suitable benchmark, what risk was behind it, and what did it cost? Here you will see the building blocks of a report, work through the difference between time-weighted and money-weighted return with an example, and learn what the GIPS standards, MiFID II, the key information document and fund reports require, and which mistakes make reports look better than they are.

The building blocks of good reporting

A report should make a decision possible: carry on, reallocate, change the manager. To do that, it has to be complete, comparable and honest. The building blocks are similar in private portfolio statements, fund reports and institutional mandates; only the depth differs:

The individual metrics have their own articles: risk metrics, Sharpe ratio, tracking error, performance attribution and look-through for seeing through funds. This article is about how they combine into a report.

Time-weighted or money-weighted?

The time-weighted return (TWR) chains together the returns of the sub-periods between two cash flows. Deposits and withdrawals do not affect it. It measures what the investments achieved and is suitable for comparison with a benchmark and with other managers. The money-weighted return (MWR) is the internal rate of return of all cash flows: it measures what the investor actually earned on their money, including the timing of their cash flows. Modified Dietz is an approximation of the MWR that needs no iteration.

For composites, the GIPS standards require time-weighted returns, calculated at least monthly and with a valuation on every day with large cash flows. Money-weighted returns are only permitted if the manager controls the external cash flows and the strategy has, for example, a fixed term, fixed capital commitments or a significant share of illiquid investments, as is typical for private equity CFA Institute (2019).

Periods and annualisation

You convert multi-year returns to an annual figure using the geometric mean, not by dividing. For periods of less than a year, scaling up is misleading: a good half-year turns into an apparent annual return that nobody achieved. The GIPS standards therefore explicitly prohibit it CFA Institute (2019). See also annualising and CAGR.

A good report shows fixed periods (year to date, one, three, five and ten years, and since inception) as well as the individual calendar years. That way nobody can pick a favourable starting point, and a bad year does not disappear into the average. You convert volatility and tracking error from monthly to annual figures with √12; this assumes that the monthly returns are independent.

Benchmark: the right yardstick

Without a benchmark, a return says little: 6% in a year in which the stock market gained 20% is weak for an equity portfolio but respectable for a cautious mixed portfolio. According to Bailey (1992), a valid benchmark needs six properties:

• unambiguous: the names and weights of all securities are clearly defined,
• investable: the investor could also buy the benchmark passively,
• measurable: its return can be calculated regularly,
• appropriate: it fits the manager's style and risk,
• current: it reflects the manager's present investment views,
• specified in advance: it is fixed before the start of the evaluation period.

By these criteria, peer groups of other funds are of limited use: their composition is not known in advance and they are not investable. In a second paper, Bailey (1992) proposes tests of benchmark quality, including coverage of the portfolio's holdings, low turnover, a tracking error below that against the overall market, and similar style characteristics. For a mixed portfolio, the right benchmark is usually a blended index with fixed weights, reset regularly, with the same currency hedging as the portfolio.

Ex-post and ex-ante risk

Ex post means measured from actual returns (volatility, maximum drawdown, tracking error, Sharpe ratio). For composites with monthly returns, the GIPS standards require the annualised standard deviation over the last three years for both the composite and the benchmark CFA Institute (2019). Ex ante means estimated from a model, for example the value at risk for the next month, the expected tracking error from the covariance of the holdings, or a stress test. Ex-ante figures are only as good as their model; a report should state the method, the data period and the confidence level.

The two views belong together. If the realised tracking error is persistently far above the expected one, either the model underestimates the risks or the portfolio was managed differently from what was planned.

Attribution in the report

Attribution explains where the active return comes from. For it to be useful in a report, three things have to be right: the segments correspond to the decisions actually taken (asset class, region, security selection), the effects add up exactly to the active return across all periods, and the weights reflect the actual exposure, which for futures and overlays means more than just market values. The basics are in the article performance attribution, and the particular features of mixed portfolios in multi-asset attribution.

Costs: what has to be disclosed

Banks and asset managers have to disclose costs under MiFID II. Under Article 50 of Delegated Regulation (EU) 2017/565 of the European Commission (2017), all costs and associated charges of the service and of the financial instruments must be aggregated in the disclosure and expressed both as a cash amount and as a percentage; third-party payments are itemised separately. Ex post, the cost disclosure comes annually and personalised, based on the costs actually incurred, together with an illustration of how the costs cumulatively reduce the return.

Before you buy a fund or a structured product, you receive the key information document (KID, Basisinformationsblatt) under the PRIIPs Regulation (2014). It has fixed sections, including “What are the risks and what could I get in return?” with the summary risk indicator and performance scenarios, and “What are the costs?” with the total costs in euros and percent. UCITS funds were exempt until the end of 2022 and until then provided the key investor information document (KIID) instead.

Standards and obligations

• GIPS: the 2020 edition of the CFA Institute's Global Investment Performance Standards has been in effect since 1 January 2020 CFA Institute (2019). They are not required by law but are regarded as best practice CFA Institute (2019). At their core is the composite: all actual, fee-paying, discretionary mandates must be included in at least one composite, so that nobody shows only their best accounts. A presentation shows at least five years and builds up to ten years, with the benchmark, the dispersion of mandate returns and the three-year standard deviation for each year; returns are stated after transaction costs, and the presentation says whether they are calculated before or after management fees.
• MiFID II: ex-ante and ex-post cost disclosure with an illustration of the effect of costs, see above.
• PRIIPs: key information document with risk indicator, performance scenarios and costs, before purchase.
• Fund reports: under Germany's Capital Investment Code KAGB (2013), the management company (Kapitalverwaltungsgesellschaft) prepares an annual report for each investment fund (Sondervermögen) containing a statement of assets, purchases and sales, an income and expense statement, changes in the fund's net assets, a comparison of the last three financial years and, for retail funds, the total expense ratio. For UCITS, it is published in the Federal Gazette (Bundesanzeiger) no later than four months after the end of the financial year, and a retail fund's semi-annual report no later than two months after the reporting date.
• ESG: funds under Article 8 or 9 of the SFDR (2019) also report in their periodic reports on how they have met their sustainability characteristics or objectives; more in the article ESG ETFs.

This overview is general information on reporting standards, not legal advice. The versions of the regulations in force at the time are authoritative.

Who the report is for

A report for all three is usually good for none of them. A proven approach is a single page with the key figures up front (return, benchmark, risk, costs), followed by the detail for those who need it.

Common mistakes and window dressing

• Survivorship bias: funds or strategies that were abandoned are missing from the statistics. This raises the average return and creates apparent persistence, as Brown et al. (1992) show (see survivorship bias).
• Cherry-picking periods: if you start just after a crash, you look good. Fixed periods and calendar years prevent this.
• Wrong benchmark: a mixed fund against a money market rate, an equity fund with small caps against a large-cap index. That measures risk, not skill.
• Return without risk: 12% a year with leverage and a 40% drawdown is something different from 8% with a 15% drawdown.
• Money weighting for savings plans: showing only the time-weighted fund return conceals that your own result can differ because of the timing of the cash flows, usually downwards, as Dichev (2007) shows for the investors in entire stock markets.
• Gross instead of net, simulated instead of real: returns before costs and backtests must be labelled as such; a backtest is not a track record.

For advanced readers

• Modified Dietz versus true TWR: Modified Dietz assumes that the return accrues evenly over the month. With large cash flows in volatile markets, it deviates noticeably. That is why GIPS requires a valuation on the day of large cash flows.
• Arithmetic and geometric excess return: portfolio 8.05%, benchmark 6.80%: arithmetically 1.25 percentage points, geometrically 1.0805 / 1.0680 − 1 = 1.17%. The geometric form chains across periods without a residual; more on this in multi-asset attribution.
• Ex-ante versus ex-post tracking error: the expected tracking error applies to today's weights, the measured one to the drifted weights of the past. For that reason alone the two differ, even without any model error.
• Dispersion within the composite: how widely the returns of individual mandates are spread around the composite shows whether all clients were treated similarly. GIPS requires a measure of internal dispersion for this if more than five portfolios were included for the year.

What the research shows

Dichev (2007) compares the returns of stock markets with the money-weighted returns of their investors: because of the timing and size of inflows and outflows, actual investor returns are systematically below buy-and-hold returns in almost all major international markets. For reporting, this means that both returns belong in a private investor's report. Brown et al. (1992) show how leaving out abandoned funds distorts performance studies and simulates persistence that does not exist.

On benchmarks, Bailey (1992) formulated the properties still cited today and showed that peer groups of other managers do not meet them. Brinson et al. (1986) made clear that investment policy, i.e. the benchmark itself, explains most of the variation in returns; a report should therefore show the benchmark and the active deviation separately. A handbook covering all the metrics and methods of performance measurement is Bacon (2008).

How Portfolio Lab calculates it

The Backtest reports time-weighted returns: inflows and outflows from a savings plan or withdrawals change the portfolio value but not the return metrics; all costs are included. The app does not calculate a money-weighted return (internal rate of return). The metrics range from total return, CAGR and volatility through the Sharpe, Sortino and Calmar ratios, maximum drawdown, and VaR and CVaR 95% per period, to beta, alpha, tracking error, information ratio and upside and downside capture against a benchmark. The trailing returns show 3 months, year to date, 1 year, and 3, 5 and 10 years annualised.

Via Export, you can create a report as a PDF (A4) or as an HTML file with interactive charts, currently only in German. It contains the composition, performance, metrics, the largest drawdowns, annual returns (with inflation, if set), trailing returns, crisis periods and, if taxes are enabled, a tax overview; at the top are the assumptions such as period, currency, rebalancing, inflows and outflows, additional costs, benchmark, risk-free rate and, where applicable, trading costs and leverage. Fund costs (TER) are already reflected in the prices; ongoing additional costs, for example for discretionary portfolio management, are deducted on top by the backtest, while the benchmark runs without them. With trading costs, the app also shows the final value without these costs. The difference is the cost disclosure in euros. Attribution against a benchmark is on the Risk page. More in the backtest methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.