Barbell allocation
Put most of what you hold in something very safe and a small part in high-variance bets, avoiding the moderately risky middle.
- Time cost
- 45 min
- Output
- An allocation with the safe end’s failure conditions stated.
- Steps
- 5
Use when
- You cannot estimate the risk of medium-risk options with any confidence.
- The upside is unbounded and the downside must be capped.
- A small number of large wins would matter more than a steady average.
Do not use when
- The medium-risk option is genuinely well-understood — then it may simply be better.
- You need steady income rather than optionality, which the barbell does not produce.
Inputs required
- A definition of safe that survives the bad case
- Bets whose downside is capped at what you commit
Procedure
- 01
Size the safe end first
How much must survive any outcome? That share goes into something genuinely low-variance. Typically 80–90%.
- 02
Check the safe end is safe
Test it against the scenario that would hurt the risky end. If both fail together, it is not a barbell — it is a concentrated position in disguise.
- 03
Cap each risky position
Each bet at the risky end must be able to go to zero without touching the safe end. That cap is the whole structure.
- 04
Spread the risky end
Several uncorrelated small bets rather than one. The point is exposure to a large win, and you do not know which one it will be.
- 05
Avoid the middle deliberately
Medium-risk positions carry real downside without real optionality, and their risk is the hardest to estimate. That combination is what the structure exists to exclude.
Characteristic failure mode
Worked example
Someone allocating their working time across a year.
- 0185% to reliable work that covers costs.
- 0215% across three speculative projects, each able to fail entirely.
- 03Check: would the reliable work also disappear in the scenario that kills all three? No — different sector.
Result
Costs are covered in every scenario, and there is exposure to three uncapped outcomes. No position in the middle, where a project would be large enough to hurt and too small to matter.
Where to go next