How to Compare DFM Performance: A Framework for UK Advisers

Put two discretionary factsheets side by side. Both say Balanced. One reports 24 months, gross of the annual management charge, from a model portfolio. The other reports five discrete twelve-month periods, net of all charges, from a composite of live client accounts. Both numbers are honest. Neither can be compared with the other, and an investment committee that puts them in the same column has produced a document that looks like due diligence and functions as decoration.

DFM performance comparison is the part of manager selection that advisers most often get wrong, precisely because it looks like the easy part. The numbers arrive pre-formatted. The work is establishing whether they describe the same thing.

The headline number is a marketing artefact

Strategy names in UK discretionary management are unregulated. Balanced at one firm can carry a materially different equity weighting to Balanced at another, and neither is misrepresenting anything. Before a single return is compared, the risk mandate has to be matched on substance: strategic equity range, permitted use of alternatives, currency hedging policy and the volatility band the manager actually targets.

The same applies to the vehicle. A model portfolio return is a theoretical construct showing what the strategy would have delivered to an account invested at inception, rebalanced on schedule, with no cash drag and no client-specific constraint. Real accounts have all three. The gap between the model and the client experience is not a rounding error, and it widens in bespoke services where individual constraints are the point.

Four variables to normalise before you compare anything

Variable	The trap	What to ask for
Time period	Since-inception or cherry-picked windows that start after a drawdown	Discrete twelve-month periods, at least five, same start and end dates across providers
Fee basis	Gross of AMC, or net of AMC but gross of underlying OCFs, platform and custody	Net of all charges the client bears, with the deducted components itemised
Risk mandate	Identical labels, different equity ranges	Strategic asset allocation ranges and realised volatility, not the label
Population	A model, or a hand-picked representative account	A composite covering every portfolio managed to that mandate

Normalise those four and the comparison starts to mean something. Skip any one of them and the exercise is presentational.

Composites are the question that separates the serious

The single most useful question in a DFM performance comparison is this: does this number come from a composite, and what was excluded from it?

A composite groups every discretionary portfolio a firm runs to a given strategy and reports them as one asset-weighted return. It exists to stop a manager choosing which client to show you. The Global Investment Performance Standards set out how composites should be constructed, requiring firms that claim compliance to include every discretionary fee-paying portfolio in at least one composite and to present a defined minimum history rather than a period of their choosing.

Compliance is voluntary and a great many capable UK discretionary managers do not claim it. The absence of a GIPS claim is not a red flag. What is a red flag is a firm that cannot explain, in writing, how its published figures were built. Three follow-up questions do most of the work.

• Which portfolios were excluded from this composite, and on what stated policy?
• Are terminated accounts included for the periods they were managed, or dropped on exit?
• Has any third party verified the calculation, and can we see that verification?

A firm that answers all three in a week is telling you about its operational infrastructure. So is one that sends a marketing deck instead.

What each layer strips away

Structural diagram, not scaled data. The width of each band is illustrative only. The point is that most published DFM comparisons stop at the first or second row while the client lives on the last one, and for portfolios held largely outside wrappers the final step can be the largest single deduction. The wrapper-level detail sits in our guide to the tax implications of DFM portfolios for UK HNW clients, and the cost layers in DFM charges explained.

Return without risk context is half a number

Two managers delivering the same five-year return, one with a maximum drawdown twice the other’s, have not performed equally. For most private clients the second one has performed better, because the client who lived through the deeper drawdown was more likely to leave at the bottom.

The measures worth requesting are unglamorous and consistently informative: realised volatility over the same discrete periods, maximum drawdown and the time taken to recover from it, and capture ratios in rising and falling markets. Downside capture in particular tells you whether a defensive mandate was defensive when it counted, which is a question a five-year annualised figure cannot answer.

Peer context helps here, and UK advisers commonly use the Asset Risk Consultants private client indices as a reference for discretionary outcomes across risk bands. Treat any peer universe as directional. Constituent firms self-select, methodologies differ, and a quartile ranking is a conversation starter rather than a conclusion.

Dispersion is the number almost nobody publishes

Ask a bespoke provider for the range of client outcomes within a single strategy over the last three years, from best to worst, and watch what happens.

A firm with tight dispersion is running a genuinely centralised process with disciplined implementation. Wide dispersion means the result depended on which manager was assigned to the client and when they invested, which makes the composite average a poor guide to what your particular client should expect. Neither answer is automatically disqualifying. A provider that has never calculated it, and cannot produce it, has told you something about its investment governance that no factsheet will.

This is also the question that distinguishes a bespoke service from a model in expensive packaging. Near-zero dispersion in a service charging bespoke fees is worth investigating, a point covered in model portfolio services vs bespoke portfolios.

Hold providers to the standard the rules already set

The FCA’s rules on past performance in COBS 4.6 exist for retail communications, but they make a serviceable minimum standard for anything a provider hands an adviser. They require past performance information to cover complete twelve-month periods, to identify the reference period and the source of the data, to avoid presenting past performance as the most prominent feature of the communication, and to carry a prominent warning that the figures refer to the past.

Use it as a filter. A performance pack that presents an eye-catching cumulative number without discrete periods, without a stated source, and without a defined universe is not the beginning of a comparison. It is a reason to go back and ask for the pack again.

A comparison your file can survive on

Selection decisions get reviewed. The reviewer, whether that is your compliance function, an acquirer’s diligence team or a supervisor, will not be assessing whether you picked the manager with the highest return. They will be assessing whether the comparison behind the choice was reasoned and reconstructable.

That means the file records the basis, not just the outcome:

1. The discrete periods used, identical across every provider assessed.
2. The fee basis, stated explicitly, with the deducted components listed.
3. The risk mandate evidence, showing that the strategies compared were genuinely comparable.
4. The population behind each number, composite or otherwise, with exclusions noted.
5. The risk measures considered alongside return.
6. The reasoning, including why a provider with weaker headline numbers was or was not preferred.

The last point is the one most often missing and the one that carries the most weight. A committee that selected the second-best performer for articulated reasons has evidenced a process. A committee that selected the best performer with no recorded reasoning has evidenced a spreadsheet.

The mechanics of running that process, from charter to minutes, are set out in how to run an investment committee in a UK advice firm. The broader provider assessment sits in due diligence when choosing a discretionary fund manager, and the periodic review obligation once a manager is appointed is covered in ongoing DFM oversight.

The bottom line

Most DFM performance comparisons fail before the analysis starts, because the numbers being compared were never describing the same thing. Fix the inputs first: same periods, same fee basis, same risk mandate, same portfolio population. Then ask the two questions providers rarely volunteer, about composite construction and client dispersion. What survives that process is a short list built on evidence, and a file that explains itself two years later when nobody remembers the meeting.

If you are reassessing your discretionary panel this year, the practical starting point is to request identical discrete-period, net-of-all-cost data from every incumbent and challenger at the same time. The variation in what comes back, and how quickly, is itself a finding.