Ensemble Risk-Adjusted Ranking
Make ideas compete on expected return and risk instead of allocating by quota
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 99%
The model replaces a star investor or fixed sector sleeves with an ensemble of specialist research clusters. Each cluster develops company estimates, but every idea is exposed to scrutiny from other fundamental and quantitative colleagues. Candidates then enter one common ranking based on expected return over a defined horizon adjusted for estimated risk. The ranking, rather than a top-down regional or sector quota, becomes the portfolio roadmap: higher-ranked stocks receive larger weights, while lower-ranked stocks receive less capital or none. As a holding's price rises toward its target, its prospective return falls and it moves down the list. The team then trims or sells it and recycles the proceeds into higher-ranked opportunities, creating an explicit and repeatable buy, sizing, and sell discipline.
Origin
Causeway designed this ensemble structure after Ketterer and Harry Hartford concluded that star-manager systems create succession risk and embed one person's behavioral biases in client portfolios.
Core principles
- 01Dispersed responsibility reduces dependence on a star investor.
- 02Professional criticism helps expose individual biases.
- 03Stocks should compete across sectors rather than receive fixed sleeves.
- 04Position size should rise with risk-adjusted attractiveness.
- 05Falling expected return should trigger capital recycling.
How to run it
- 1
Distribute research ownership
Assign industries to specialist clusters with accountable leaders. Let each cluster develop deep knowledge and produce comparable investment cases.
Pro tip Choose specialists whose analytical styles complement one another.
Watch out Specialization should not become ownership of a protected portfolio sleeve.
- 2
Challenge every thesis
Expose each candidate to critique from other fundamental analysts and quantitative colleagues. Test assumptions, evidence, valuation, and risk before ranking it.
Pro tip Invite junior colleagues to challenge senior analysts so overlooked evidence can surface.
Watch out A nominal review process fails when hierarchy suppresses dissent.
- 3
Estimate return and risk
Set a common investment horizon and calculate the expected annualized return to a defensible target. Assess company-specific and systematic risks using consistent standards.
Pro tip Use scenario ranges where risks cannot be represented by one number.
Watch out Inconsistent targets make cross-sector rankings misleading.
- 4
Create one competitive ranking
Combine all sector candidates into a shared list ordered by risk-adjusted expected return. Avoid guaranteed positions for any cluster, sector, or region.
Pro tip Review borderline rankings collectively when small input changes alter the order.
Watch out Do not confuse ranking precision with certainty about future returns.
- 5
Translate rank into weight
Give larger portfolio weights to higher-ranked securities and smaller weights to lower-ranked ones, subject to portfolio risk controls. Let aggregate sector and regional allocations emerge from the selections.
Pro tip Apply liquidity and concentration limits after ranking rather than replacing the ranking.
Watch out Blindly following rank can create unintended correlated exposure.
- 6
Recycle capital dynamically
Update prices, targets, and risks continuously. When appreciation reduces a holding's expected return and rank, trim or sell it and move capital to stronger opportunities.
Pro tip Predetermine review triggers to reduce attachment to successful holdings.
Watch out Do not retain a company merely because the original thesis worked.
In the wild
Causeway compares Volkswagen's expected return and estimated risk with Barclays and every other researched company. If Volkswagen offers the stronger risk-adjusted return, it ranks higher and receives a larger weight. As its share price converges toward the two-year target, its remaining return declines, moving it down the ranking.
→ Portfolio weights change through a consistent cross-sector process rather than a manager's regional or sector forecast.
A consumer-staples cluster may have fewer holdings when staple valuations imply weak prospective returns, while a financials cluster contributes more highly ranked ideas. If staple prices later fall enough to improve expected returns, those companies can rise in the same shared ranking.
→ Sector exposure responds to security-level opportunity without imposing fixed sleeves.
Common mistakes
Guaranteeing equal sector sleeves
Fixed allocations protect weak ideas from competition and can force capital into unattractive sectors. Let comparable security-level rankings drive selection.
Allowing one final wizard
A star allocator reintroduces the concentration of judgment the ensemble was designed to remove. Use transparent shared rules and collective scrutiny.
Failing to lower ranks after gains
Price appreciation reduces prospective return even when the company remains excellent. Update the ranking and recycle capital when better opportunities emerge.
Is it for you?
Best for
It is best for collaborative investment organizations with multiple specialist research teams and a common valuation discipline.
Not ideal for
It is not ideal for teams that cannot compare expected returns and risks consistently across different industries.
From the transcript
“Whatever the biases are, you don't want to have any one person's biases or one or two people embedded in your client portfolios all the…”
“we work in a ranking system the highest risk adjusted return stocks down to the lowest”
“we now know that we should have larger weights in the stocks that are higher ranking and less weights in those that are lower ranking.”
From the episode
Sarah Ketterer Discusses Investment Management
Sarah Ketterer