ETF Model Portfolios for Advisors: How Risk-Managed Models Work

An ETF model portfolio is a ready-made allocation built entirely from ETFs, which an advisor applies across client accounts. A risk-managed ETF model portfolio adds a rule set that changes its own exposure when market conditions change.

You evaluate one by asking three things. What triggers the change. How fast it can fire. And what it costs you when the trigger is wrong.

We measured that third one. Take a plain 200-day rule on a 60/40 model. Over 19 years it cut the worst drawdown from 35.8% to 15.2%. It gave up 0.7 points of annual return to do it. Dates and method below.

Most advisor due diligence stops at the holdings list. That's the wrong page. The holdings tell you where the model is today. The rule set tells you where it goes next.

What is an ETF model portfolio?

It's a maintained allocation, usually five to fifteen ETFs, with target weights for each one. The advisor subscribes through a custodian, a TAMP, or a rebalancing platform. The manager maintains the sleeves and publishes the trades. The platform applies them to every linked account.

The advisor still owns the client relationship and the suitability call. The model handles security selection and rebalancing. That split is the whole appeal.

Who builds model ETF portfolios for financial advisors?

Three kinds of builders, and the difference matters.

• ETF sponsors. Large asset managers publish models built mostly from their own funds.
• Third-party strategists. Independent managers build models from any sponsor's ETFs, sometimes their own.
• Home offices. Broker-dealers and large RIAs run internal models for their own advisors.

Ask one question of every builder. What share of the model sits in the sponsor's own products? A high share isn't wrong. It just means the fund lineup chose the holdings, not a neutral screen.

Then ask where the model actually runs. The delivery platform decides how fast a published trade reaches a client account. Our guide to rules-based risk management platforms covers that layer.

What does an ETF model portfolio cost?

Two layers. The first is the expense ratio of each ETF inside the model, which the client always pays. The second is a model or strategist fee, which some builders charge and others waive.

Sponsor models often carry no separate model fee. The sponsor earns through the underlying funds instead. Third-party strategists usually charge a fee on top, quoted in basis points. Add both layers before you compare models. Then add your platform fee.

Conservative, moderate, or aggressive: which ETF portfolio model fits?

Most builders publish three to five risk tiers. The tiers differ by one thing, the equity share.

• Conservative. Roughly 20% to 40% equity. Built for income and short horizons.
• Moderate. Roughly 50% to 70% equity. The 60/40 family lives here.
• Aggressive. Roughly 80% to 100% equity. Built for long horizons and nothing else.

One trap. An aggressive ETF portfolio model with a risk rule is not the same thing as a moderate strategic one. The first carries full equity risk until the rule fires. The second carries less risk every single day.

Pick the tier on the client's horizon. Pick strategic or risk-managed on the client's behavior. Two separate decisions, and advisors blur them constantly. For the allocation layer underneath both, see our guide to portfolio asset allocation models.

Strategic vs risk-managed ETF model portfolio: what changes?

A strategic model holds fixed weights. A 60/40 model stays near 60/40 and rebalances on the calendar. Its risk rises and falls with the market.

A risk-managed model lets the weights move. Something other than the calendar decides when. That second part is where the diligence work lives. The industry files this family under tactical asset allocation, and our guide to that term runs the full comparison.

A conservative strategic model isn't the same thing either. It lowers risk permanently by holding less equity all the time. A risk-managed model tries to carry risk when risk is paying, and step back when it isn't. It seeks to reduce drawdowns without a permanent haircut to exposure.

Both are legitimate. A client who can't tolerate volatility at all wants the first one. A client who needs growth but sells at the bottom wants the second.

How does a rule set actually decide to move?

Four steps. Simpler than most managers make them sound.

1. Measure. The model reads a defined input on a set cadence. Daily closing price is the most common one.
2. Compare. That reading gets tested against a threshold. A moving average, a volatility band, a price reversal level, a credit spread.
3. Classify. The result puts each holding into a state. Risk-on or risk-off, usually, sometimes with a middle setting.
4. Trade. Holdings that changed state move into the defensive sleeve, or back out of it.

Then ask where the lag sits. There's always a gap between measuring and trading. A model that measures on the close and trades the next open carries one extra session of risk. That gap is real, and it never shows up in the brochure.

Price-trend rules are the most common kind in model portfolios. Our piece on trend following covers how they behave alone. Volatility targeting covers the other main family.

Can an ETF model portfolio tilt toward factors?

Yes, and plenty do. A factor tilted ETF model portfolio reworks the core equity sleeve. It buys value, momentum, quality, size, or low-volatility funds instead.

A tilt and a risk rule solve different problems. A tilt changes what you own. A rule changes how much you own. You can run both.

Ask what the tilt is funded from. A momentum sleeve carved out of the equity allocation raises tracking error against a cap-weighted benchmark. One carved out of bonds raises outright risk. Our guide to factor investing covers which factors have held up and over what windows. For the weighting question sitting underneath it, see equal weight against cap weight.

How do I evaluate a risk-managed ETF model portfolio?

Five questions, in this order.

1. What is the signal? Ask what data the rule set reads. Price, volatility, breadth, credit spreads, the yield curve. "Proprietary" is not an answer.
2. How often can it move? A model that rebalances quarterly can't respond to a three-week crash. Match the response speed to the risk you're hiring it to handle.
3. What does it move into? Cash, T-bills, long duration, low-volatility equity, gold, or a buffer ETF sleeve. Every defensive sleeve carries its own failure mode. Long bonds failed in 2022.
4. How long has the underlying strategy run live? Ask about the strategy, not the wrapper. A young ETF can hold a strategy with years of live history in separate accounts.
5. What is the whipsaw record? Every rules-based model gets faked out. Ask how often it happens, and what an average false signal costs. We counted it for a basic rule further down this page.

If a manager can't answer the fifth question, nobody has measured it. That tells you something on its own.

What does a risk-managed ETF model portfolio actually return?

We built the simplest honest version and measured it. Two models, same 60/40 target, same bond sleeve, rebalanced on the last session of each month.

The strategic model holds 60% in a broad S&P 500 index fund. The other 40% sits in a core US bond fund. The risk-managed model is identical, with one rule added. When the equity fund closes below its 200-day moving average, that sleeve moves to Treasury bills. It moves back when the fund closes above the line. Signal on the close, trade the next session.

Measured on adjusted closing prices from May 31, 2007 to October 22, 2026. The start date is set by the Treasury bill fund's own trading history, not by us.

• Full period, 19.4 years. Strategic returned 7.8% a year, worst drawdown 35.8%. Risk-managed returned 7.1% a year, worst drawdown 15.2%.
• 2008. Strategic lost 21.1%. Risk-managed gained 1.7%, and its deepest drawdown that year was 6.6%.
• 2010. Strategic gained 12.1%. Risk-managed gained 2.6%, and its drawdown was deeper, 11.1% against 8.4%.
• 2020. Strategic gained 14.7%. Risk-managed gained 9.4%. Drawdowns were 21.6% and 13.1%.
• 2022. Strategic lost 15.8%. Risk-managed lost 13.3%. Drawdowns were 20.5% and 13.8%.

Read 2010 twice. The rule gave up 9.5 points of return and still delivered the worse drawdown. A choppy up year can beat a signal on both counts at once.

That's the whole trade in one line. Under a point of annual return bought a worst drawdown less than half as deep. Whether a client should pay that is a conversation, not a calculation.

This is our own measurement of a generic published rule on public index fund prices. It is not a THOR strategy, and no manager would run exactly this.

What does a model ETF portfolio do in a real drawdown?

Test any model you're evaluating against two dates.

The first is 2022. The S&P 500 peaked at 4,796.56 on January 3, 2022. It closed at its low of 3,577.03 on October 12. That's roughly a 25% decline over nine months. Slow, grinding, with several sharp rallies inside it.

Bonds gave no cover that year. The Bloomberg US Aggregate lost about 13% in 2022, its worst calendar year on record. A strategic 60/40 model fell into the mid-teens, because both sleeves fell together.

The second is 2020. The S&P 500 peaked at 3,386.15 on February 19, 2020 and bottomed at 2,237.40 on March 23. That's about 34% in 23 trading sessions. The index recovered the whole loss by late August.

Ask the manager what the model did across both. Not the return. The positioning. When did it cut exposure, and when did it come back?

Those two drawdowns punish opposite behavior. 2022 rewarded getting defensive and staying there. 2020 punished exactly that. A model that handles one well often handles the other badly. You want to know which one you're buying. We ran the two side by side in risk-on and risk-off models compared.

How often does a risk rule get faked out?

We counted it on the same model and the same 19.4 years. The rule went defensive 48 separate times. It sat in Treasury bills for 23.8% of all sessions.

Thirty-eight of those 48 defensive spells lasted under 30 sessions. The median spell ran five sessions. So most exits were short, and most short exits were noise.

Ten ran 30 sessions or longer. The longest covered May 22, 2008 to May 29, 2009. That single spell is why the 2008 line above reads the way it does.

Ask every manager for this count. If nobody can produce it, nobody has measured it.

Is an ETF model portfolio the same as a separately managed account?

No. Both leave the client holding positions directly, which is where the confusion starts. A model holds ETFs, so the client owns fund shares. A separately managed account can hold individual securities. That opens up tax-loss harvesting position by position.

Models win on simplicity and minimum account size. Accounts win on tax control. The taxable balance usually decides it.

When does a risk-managed model portfolio fail?

This is the part worth reading twice.

• V-shaped reversals. March 2020 is the case study. Our measured rule went defensive on March 6 and did not come back until May 26. It missed the first two months of the recovery.
• Choppy, directionless markets. Signals fire, reverse, and fire again. Each round trip costs a little. 2010 is the measured version above, where the rule cost 9.5 points and still took the deeper loss.
• Correlated defensives. If the defensive sleeve is long bonds, 2022 shows what happens when the hedge falls alongside the asset.
• Taxable accounts. Moving exposure realizes gains. A model that looks clean in an IRA can look expensive in a brokerage account. ETFs are tax-efficient inside the wrapper, not when a rule sells them.
• Client patience. In a long uptrend, a risk-managed model lags a plain index fund. That gap is what the insurance costs. Clients forget that by year three.

The last one is the one that actually loses accounts. Set the expectation in the first meeting, in writing. It lands harder with a retiree. Sequence-of-returns risk makes the order of losses matter more than the average.

How does THOR build its models?

THOR runs systematic ETF models, including THOR Low Volatility, THOR Index Rotation, and THOR AdaptiveRisk Dynamic. The rule sets read price behavior and shift exposure between risk-on positions and defensive ones.

The underlying strategies have run live in separate accounts since 2020. The ETF wrappers came later. So when you evaluate THOR, ask about the separate account history. That's where the live record sits.

On construction mechanics, see how advisors build risk-managed portfolios. On the narrower goal of holding value through a decline, see capital preservation strategy.

Definitions

• ETF model portfolio. A maintained allocation built from ETFs that an advisor applies across client accounts.
• Strategic model. A model that holds fixed target weights and rebalances on a schedule.
• Risk-managed model. A model whose exposure changes on a defined rule set rather than a fixed schedule.
• Strategist fee. A charge for the model itself, separate from the expense ratios of the ETFs inside it.
• Signal. The measurable input that triggers a change in exposure.
• Defensive sleeve. Where the model puts money when it steps out of risk assets.
• Whipsaw. A signal that fires and reverses quickly, costing the model on both legs.
• Risk tier. One version of a model, set apart from its siblings by its equity share.
• Factor tilt. An equity sleeve weighted toward a trait such as value or momentum instead of market size.
• Live track record. Results from real money, as opposed to a backtest.

One last check before you fund anything. Ask for the model's live start date, and the date of its most recent rule change. If the rules changed after the track record started, the record before that change describes a different model.

Past performance is not indicative of future results. Index figures and measured rule results cited above are derived from public market data for illustration only. They do not represent the performance of any THOR fund or strategy, and this material is not investment advice.

Talk it through

THOR builds adaptive index strategies for advisers and allocators. Fifteen minutes with our founder and CIO gets you the mechanism, the index history, and an honest read on where it fits. Book a time.