The Ghost in the Machine
Why Australia's SMA Revolution Is Running on Empty Data
The Australian SMA was sold as a revolution. Direct ownership. Radical transparency. Tax smarts. Bespoke portfolios at scale.
Instead, what emerged looks suspiciously like yesterday's multi-manager architecture in a shinier wrapper: funds wrapped in models, models wrapped in platforms, platforms wrapped in advice groups. And underneath the marketing, the data that should prove the advantages is either missing, mangled, or incapable of answering the questions that matter.
This is an autopsy of the Australian SMA. Not because the structure is dead. Because the story is.
The Clean List vs the Dirty Reality
The modern SMA is supposed to be a precision instrument.
To make it measurable, the industry has converged on a deceptively simple spec sheet of seven pieces of "starter data" that every SMA should carry like a passport:
On paper, that list looks like the DNA of a clean, modern market. You can compare managers, interrogate performance, and see what sits in your portfolio. It's tidy. It's comforting. It fits neatly in a Professional Planner PDF.
In practice, this list behaves more like marketing theatre than operational truth.
Because when you peel back the slick dashboards and the clever UX, you rarely see a live, joined-up data spine feeding those seven fields. You see legacy registry systems duct-taped to new UI. Static PDFs being scraped to simulate "live" holdings. Manual overrides to force things to reconcile just before month-end.
We look like we're building a Tesla. Behind the scenes, it's a late-model Holden with a tablet glued to the dash.
The list is clean. The reality is anything but.
"Transparency" That Can't Answer Simple Questions
Holdings disclosure is the emotional centre of the SMA myth.
In a pooled fund, you own units in a black box. In an SMA, the pitch goes, you're supposed to see the gears turning. You should be able to look at your account and say: "I own 140 BHP, 60 CSL, 220 CBA. I know where my sector risk is. I can see my international exposure, not just a fund name."
But in the Australian SMA universe, that promise often dissolves under scrutiny.
The timing problem. The "Top 10" exposure report your adviser slides across the table might be based on positions as at 30 June, released 4–8 weeks later, filtered through a platform that only refreshes fund holdings monthly or quarterly. The client thinks they're seeing today. They're seeing a historical reenactment.
The identity problem. In a world of consultant-built models, dealer-group white labels, underlying mandate managers, and platform implementation teams, who is the actual manager of this SMA?
The "Product Identifier" field often behaves like an ontological riddle: the brand on the brochure is not the entity running the day-to-day book; the entity picking the stocks is not the one setting the asset allocation; the entity operating the platform is not the one bearing the headline reputational risk.
When an SMA underperforms or behaves strangely, you want to ask: Was it the asset allocation? Was it security selection? Was it sloppy execution, cash drag, or poor trading? Too often, the data cannot cleanly allocate the blame.
The industry is drowning in data, but starving for information. We have databases of trades and reconciliations. We can't cleanly answer the question: "Did this client get what the model promised?"
The Australian Anomaly: The Fund-of-Funds Matryoshka
To understand why Australian SMAs feel so off, you have to look at the basic building blocks.
In the US, much of the traditional SMA boom was built on direct equities: huge, liquid markets; near-zero brokerage; technology that could juggle hundreds of tickers per account; and tax-lot engines that systematically harvested gains and losses.
The Australian market took that language—"SMA," "direct ownership," "tailored"—and then quietly swapped the guts.
Instead of 30–50 direct stocks per sleeve per client, we built models populated with managed funds and ETFs, then wrapped them inside SMA infrastructure, then wrapped that inside platforms.
Open a typical Australian "SMA" and you find a global equity fund here, a bond fund there, a few ETFs for spice, sometimes another "multi-asset" fund sitting inside the mix. This is the Fund-of-Funds Trap.
On paper, it's "an SMA of carefully selected managers." In reality: you see that you hold the Magellan fund, not Magellan's underlying stocks. You see a bond fund code, not the duration or credit risk actually driving your returns. You see a global ETF, not the actual geographic or factor tilts inside it.
Holdings disclosure collapses to "wrappers I could have bought directly," while the platform quietly adds SMA administration fees, trading and implementation leakage, and additional complexity around tax parcels and instructions.
We are paying fees to wrap fees to wrap fees. And because the data is rarely presented as a true "all-in" total cost of ownership, this structure can pass itself off as innovation.
A Russian doll is charming when you're five. It's less charming when every layer has a basis point charge.
The Fractional Failure: When the Plumbing Calls the Shots
The second uniquely Australian twist is brutally simple: the market plumbing wasn't built to love fractions.
In the US, fractional share trading in listed securities is now mundane. If you have $500, you can buy $500 worth of a stock. The ledger tracks 0.0034 units. The system doesn't flinch.
Australia's settlement and custody world has, historically, taken a different view. CHESS and platform custody models want whole numbers. Many systems and controls assume whole parcels. Fractional ownership introduces headaches in record-keeping, corporate actions, and disclosure.
That sounds like a back-office detail. It isn't. It decides who gets to play.
Take a diversified Australian equity SMA with 25–30 holdings: some stocks at $5–$10 per share, some at $80–$100, a handful north of $200. Now give that model to a client with $20,000 and tell the algorithm: "Implement this exactly, please."
You immediately collide with reality. The model says "3.4% in CSL." 3.4% of $20,000 is $680. One CSL share costs, say, $300-plus. You can't buy 2.26 shares.
So the system rounds, compromises, or just leaves cash sitting on the side. Repeat this across multiple high-priced stocks and small balances and you get persistent under-exposure to the expensive names, higher-than-intended cash holdings, and a quiet, relentless performance leak called cash drag.
The report might proudly show "target cash: 2%." The lived reality for small accounts is often 10–15% cash, simply because the maths of whole shares doesn't bend.
And yet, very few SMA reports scream in red letters: "Warning: this account cannot physically hold the model given its balance."
Instead, the structural flaw is dressed up as "flexibility": "The SMA maintains a prudent allocation to cash." "We dynamically manage liquidity."
Recommended by LinkedIn
No. You are stuck in a whole-share world and pretending it's a feature.
Could custodians run internal ledgers that allocate fractional economic interests behind the scenes? Yes. Some already do, in limited ways and niches. But that quietly changes the nature of beneficial ownership and operational risk in ways that most retail clients do not understand, and that few disclosure documents explain in plain language.
So the industry sits in an uncomfortable middle: direct equity branding, whole-share plumbing, small accounts that mathematically cannot be what the brochure says they are.
The "Why" Crisis: Who Actually Wins?
SMAs are justified with three main claims: better tax (you control your own cost base and CGT); better management (professional oversight, institutional process); and better ownership (you "own the shares").
On a whiteboard, these benefits are real. In practice, the question is not "is it possible?" but "is it delivered, for this client, at this balance, on this platform, in this structure?"
The industry loves Time-Weighted Returns because it strips out cash flows, gives you a tidy number for "model skill," and looks beautiful in a factsheet. Clients do not live inside TWR. They live inside Money-Weighted Returns—what their dollars actually earned over their journey—after platform fees, SMA fees, underlying costs, and transaction costs, with cash drag, rounding, and messy real-world trading layered on top.
A model can honestly say: "We did 9% per annum over 5 years versus an 8% benchmark." And yet the median small-balance client in that model can honestly have experienced 4–5% per annum after every cost, timing effect, and implementation friction.
The gap between "model return" and "lived return" is where narratives go to die.
Now add tax. Yes, not inheriting embedded CGT can be valuable, especially for large, long-term, taxable accounts. Yes, you "can" manage tax lots cleverly in a direct-equity SMA. But if the client has $40,000 in the SMA, another $260,000 in super and other vehicles, a short holding period, and an SMA-of-funds structure on a platform, the marginal tax benefit may be swamped by higher fee layers, transaction costs, and the structural implementation issues described above.
Yet SMAs are increasingly sold down into the $30,000–$100,000 client segment as if the whiteboard benefits translate linearly.
We are democratising complexity faster than we democratise advantage.
SMA 2.0: What a Grown-Up Version Would Look Like
If all of that is "SMA 1.0," what does a credible "SMA 2.0" look like in Australia?
It's not just "more data" or "better dashboards." It is a set of design principles that put implementation truth and client outcomes back at the centre.
Kill the static model. Build constraint engines.
The era of "house model, pushed into 500 accounts with a button" needs to end. Not because scale is bad, but because pretending that one static list of stocks or funds is "tailored" is insulting.
SMA 2.0 needs parametric portfolios—engines, not lists. Inputs that include existing holdings, employer stock, tax profile, exclusions, risk budget, and account size. Explicit tracking-error ranges against a benchmark, with rules for how to trade off tracking error versus tax and trading costs.
If a client already holds $500,000 of BHP as employee stock, the engine should automatically exclude or severely scale back BHP in the Australian equity sleeve. The system should prove this with data, not rely on a manual notation that "tailored elements apply."
The "Tailored Elements Flag" stops being a binary field and becomes a log of actual client-specific constraints, with evidence of how they altered holdings, risk, and return.
Treat fractionalisation as infrastructure, not a marketing add-on.
SMA 2.0 has to choose: either build serious fractional capability inside the custodian/platform stack (with proper disclosure about the nature of the client's interest), or be brutally honest about minimum viable account sizes for direct-equity SMAs.
You cannot have it both ways. You cannot claim "one model for all" while the maths ensures small accounts are structurally under-invested and over-cash. You cannot celebrate direct ownership while quietly slipping into quasi-pooled behaviour behind the scenes.
If you can't implement a 30-stock model faithfully below, say, $80,000 or $100,000 without massive compromises, then don't offer it. Offer an ETF or pooled fund solution instead. Honesty beats theatre.
Cash drift should not hide in the background. It should be measured per account, reported against model target, and explained clearly to both adviser and client.
Hard-wire look-through as a standard, not a "nice to have."
If your SMA holds only direct stocks and bonds, holdings disclosure is straightforward. As soon as you introduce managed funds and ETFs, the old "Top 10 securities" list becomes borderline meaningless for risk analysis.
SMA 2.0 demands that any pooled vehicle inside an SMA must expose enough data to show sector, region, major positions (where appropriate), duration/credit for fixed income, and style/factor tilts. The platform aggregates those exposures across all layers so the client can see their true portfolio, not just the wrappers.
If a manager refuses to provide that look-through, you face a simple choice: don't put them in an SMA that trades on a transparency promise, or if you must, disclose loudly: "This holding cannot currently be fully looked through – here's what that means."
Partial transparency is not transparency. It's decor.
Net-of-everything reporting, at client level, as default.
Finally, reporting needs to grow up. Instead of model TWR in isolation, headline fee in isolation, and pretty charts that ignore slippage and cash drift, SMA 2.0 should elevate client-level Money-Weighted Return after all fees and realistic implementation costs.
It should include a clear "implementation leakage" metric: how far this client's experience diverged from the model due to cash, rounding, timing, and so on. It should include a total cost of ownership number that encompasses platform fee, SMA overlay fee, underlying MERs, and estimated transaction costs.
That number will be uncomfortable. Good. If an SMA cannot survive the X-ray of all-in cost versus benefit, it should not need a marketing budget. It should need a redesign.
The Final Question: Whose Problem Are We Actually Solving?
Strip away the brochures and the buzzwords and one truth remains: SMAs have been extraordinarily effective at solving one problem, the operational complexity of running centralised portfolios across hundreds or thousands of advised clients.
From the adviser's chair, a pooled fund switch is an RoA, signatures, settlement risk, and admin friction. An SMA switch is: update the model, push the button, let the system do the work. That is real value. It lowers the cost-to-serve, makes compliance tidier, and lets smaller practices behave like institutional CIOs.
But it is not the same as delivering truly personalised, constraint-aware portfolios, or proving superior client-level net outcomes against simpler, cheaper alternatives, or using data to shine a light on every layer of cost, leakage, and drift.
Right now, the SMA story is out of sync with the SMA reality. The seven "starter data" points are not wrong. They are simply insufficient if treated as a box-ticking exercise rather than the surface of a deeper accountability stack.
The real fork in the road is philosophical: Do we want SMAs to be an industrial control system for advice firms, a legitimate, efficient way to standardise investment implementation at scale? Or do we want SMAs to be a personalisation engine—a way to express client-specific constraints, tax profiles, and risk budgets in code and holdings?
If it's the former, say so. Call them what they are: centralised portfolio programs. If it's the latter, then the industry has work to do.
Work on market structure (fractionalisation and minimum viable balances). Work on data standards (look-through, identifiers, total cost). Work on reporting (MWR, implementation leakage, cash drift). Work on design (parametric portfolios, constraint engines, real tailoring).
The comforting line is: "The data exists. We just need the courage to read it." The uncomfortable extension is: once we read it, we may not like what it says about who currently wins from the SMA revolution.
And that is the moment where the ghost in the machine either becomes an actual brain, or finally gets exorcised.
Great piece here Ben. Reads like a Briefcase strategy deck I wrote in 2023, but far more polished :)
Thanks Ben Walsh - you've rased some important issues. Hopefully you receive some responsive comments from industry leaders within the SMA space.
The point about client outcomes being secondary in generation one feels accurate. If private assets are going mainstream, product design and disclosure have to mature well beyond marketing narratives.
Ben Walsh Some really good points here — particularly the often-overlooked question of accountability. Specifically: who is responsible (and ultimately liable) if investor portfolios are not actually aligned with the model portfolio an investor is subscribed to? That naturally leads to a few follow-on questions: • Who is independently checking that investor portfolios match the model? • What level of assurance can Superannuation Trustees rely on • Who is measuring and monitoring the level of deviation over time? • Who is analysing and publishing those results? • What does good look like (and what is not good) • And what disclosures are necessary so investors and regulators can genuinely assess outcomes? “SMA 2.0” feels like the right direction — and one we’ve been actively working toward for some time — helping managed account providers move from point-in-time construction to continually assured, client-centric portfolios.
Agreed. Achieving scale and genuine personalisation simultaneously is an impossible task (with the exception of every client being the same). Unless an SMA is constructed specifically for a client, promoting personalisation is the equivalent of selling fast food as fine dining.