Risk Tolerance
Risk tolerance is the amount of decline a portfolio can fall by without its owner deciding to sell. It is a separate question from risk capacity, which is how much loss the financial plan itself can absorb, and most questionnaires measure a stated attitude rather than any actual behaviour.
Everything about a portfolio’s design depends on one input that nobody can measure reliably, including about themselves. That is worth acknowledging before building anything on top of it.
How it works
Risk tolerance is the size of decline you will hold through without selling. Not the size you say you would hold through, and not the size you believe you would — the size you actually will.
It matters because selling at the bottom converts a temporary decline into a permanent loss. A portfolio held through a 40% fall recovers; the same portfolio sold at the bottom does not, and recovery from there requires a 66.67% gain.
So a nominally worse allocation held through a fall beats a better one abandoned in it, which is the specific reason this is not a soft subject.
Tolerance against capacity
Capacity is arithmetic: how much loss can the plan absorb. Someone needing the money in two years has low capacity regardless of temperament. Someone with a secure income and a thirty-year horizon has high capacity regardless of how they feel.
Tolerance is behavioural: how much loss will you sit through. The two are independent, and the lower of them governs.
High tolerance with low capacity is the dangerous combination, because nothing stops you until the money is needed and is not there.
A worked example
Take 100,000 in a market that falls 40%.
Fully in equities it falls to 60,000 — a loss of 40,000, needing 66.67% to recover.
At 60% equities it falls to 76,000 — a loss of 24,000, needing 31.58%.
At 40% equities it falls to 84,000 — a loss of 16,000, needing 19.05%.
State it as money, not as a percentage. “A 40% fall” is abstract; “watching 40,000 disappear and doing nothing about it for two years” is the actual question, and people answer the two differently. The recovery arithmetic is in the drawdown recovery calculator.
Why questionnaires miss
A risk questionnaire is completed in a quiet moment with no money moving. It measures a stated attitude, and the thing it is trying to predict is a decision made under stress with a falling balance on the screen.
Those two correlate weakly. Most people discover their real tolerance the first time it is tested, which is an expensive way to find out.
On this site’s shared series, 95% of bars sit below a prior peak and the longest stretch under
water ran 73 bars, while the series finished up 3.61%. The figures are in
research/series-measurements.json. Being below a high is the ordinary state, so the question is
not whether it will be tested but when.
Building for uncertainty about yourself
If you have not been through a severe fall, assume less tolerance than you would like to have. The cost of being too conservative is a slightly lower return; the cost of being too aggressive is selling at the bottom, and those are not symmetrical.
A cash buffer raises effective tolerance directly, because a fall does not force a sale when spending is covered from elsewhere.
Costs
Revising an allocation during a fall costs the spread and, in a taxable account, realises the loss. On this site’s shared series a round trip measures about 2% of the median bar range of 0.493.
Deciding beforehand is free and deciding during is expensive, which is the whole argument for writing the allocation down while nothing is happening.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 4 have a title about risk
tolerance, at a median of 4,282 views across 4 channels — and 50% use beginner-shaped language.
Asset allocation appears in 4 videos at 5,289 and diversification in 3 at 487. The counts come from
site/rank_investing.py, which deduplicates by video id.
Four videos at a 4,282 median. The input that every allocation decision depends on has essentially no coverage, while the products that sit inside the allocation have hundreds of videos each.
The answer to the question on that chart is that reducing now locks in the allocation you would have wanted before the fall, at the price of the fall. If the allocation was wrong, it was wrong at the top and the correction should have happened there. Changing it at the bottom is paying full price for the lesson — and the usual result is being under-allocated for the recovery as well.
When it fails
The failure is discovering your real tolerance during the event it was supposed to be measured for. Someone completes a questionnaire, is placed in an aggressive allocation, holds comfortably through several ordinary years, and then sells everything in the third week of a severe decline. The questionnaire was not wrong about their attitude; it simply had no way to test their behaviour, and the only test that exists costs whatever the decline had reached by the time they took it.
The second failure is confusing tolerance with capacity. The lower of the two governs.
A third is setting it once and never revisiting. Capacity changes with age, income and obligations.
A fourth is expressing it as a percentage. Money is what people actually respond to.
A fifth is assuming past calm implies tolerance. An untested allocation is an untested assumption.
And a sixth is adjusting during the fall. That is the one moment the decision is most expensive.
Related
Asset allocation is where the answer becomes weights. Diversification reduces how severe the test is likely to be. And time horizon is the capacity side of the same question.
The version of the question worth asking is not how you feel about a 30% fall. It is what you did the last time one happened. If the answer is that you have never experienced one, then the honest position is that you do not yet know, and the allocation should reflect that rather than an assumption about how brave you will turn out to be.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.