Building a portfolio framework that survives its own assumptions
Most portfolios are not undone by bad forecasts. They are undone by the absence of rules that still function once the forecast proves wrong.
Key takeaways
- Write the constraints before the allocation; the allocation is the output, not the starting point.
- Every assumption should carry an observable that would falsify it.
- A fixed review cadence converts panic decisions into scheduled ones.
A portfolio is an argument about the future written in positions. The problem is that arguments about the future are wrong at a predictable rate, and the cost of being wrong is rarely distributed evenly. A framework earns its keep not when the thesis holds, but in the interval between the thesis breaking and the investor admitting it.
That interval is where most permanent damage occurs. It is not caused by the market moving; it is caused by decisions taken without prior agreement on what would constitute a change of mind. When the rules are written after the drawdown begins, they are written by the drawdown.
Constraints first, allocation second
The conventional sequence — form a view, express it in weights, then bolt on risk controls — inverts the correct order. Constraints are the more durable object. A view has a half-life measured in quarters; a constraint reflects something structural about the investor: liquidity needs, obligations, career risk, tax position, temperament under loss.
Written first, constraints reduce the allocation problem to something tractable. Written last, they become negotiable, and anything negotiable is renegotiated at exactly the wrong moment.
- Liquidity floor: the capital that must remain untouched by market outcomes, expressed in months of committed spending rather than a percentage.
- Concentration ceiling: the largest single exposure permitted, sized so that a total loss on it remains an inconvenience rather than an event.
- Drawdown tolerance: the peak-to-trough decline the investor can carry without altering behaviour — measured honestly, not aspirationally.
- Complexity budget: the number of positions and instruments that can genuinely be monitored, not merely owned.
Every assumption needs a falsifier
An assumption without an observable that would contradict it is not an assumption; it is a preference. Useful frameworks pair each thesis with the specific evidence that would retire it — a change in policy direction, a break in a spread relationship, a deterioration in the cash generation of an underlying business.
The purpose of writing down what would make you wrong is to make being wrong cheap.
This discipline also protects against a subtler failure: thesis drift. A position bought for one reason and held for another is an unrecorded new decision, taken without scrutiny. Falsifiers make drift visible.
Cadence beats reaction
Continuous monitoring and continuous decision-making are different activities and should be separated. Information can arrive at any time; portfolio changes should arrive on a schedule, with clearly defined exceptions for pre-agreed triggers.
A fixed cadence — monthly monitoring, quarterly rebalancing, annual framework review — does two things. It ensures neglected exposures are examined even when nothing dramatic has happened, and it converts the urge to act during volatility into a queued item rather than an executed trade.
Stress the framework, not just the portfolio
Scenario work usually asks what happens to positions. The more revealing question is what happens to the process: which rule breaks first, which constraint becomes uncomfortable, which decision would be deferred. If a scenario produces a portfolio outcome the investor can absorb but a process outcome they cannot, the framework is the exposure.
The test of a framework is not whether it performed well in a favourable regime. It is whether it remained usable in an unfavourable one — legible enough to follow under stress, and specific enough that following it did not require optimism.