Horizon and Risk Tolerance
Two different constraints that are frequently treated as one. Horizon is a fact about when the money is needed; tolerance is a fact about what will actually be held through.
MadStockAlerts Research · Updated August 29, 2026
What to take away
- Horizon is objective: when the money is needed, and in what amounts.
- Capacity for risk is what the plan can survive; tolerance is what the person will survive.
- The binding constraint is usually the lower of the two.
- Stated tolerance is measured in calm periods and tested in bad ones.
- A plan abandoned at the bottom is worse than a more modest plan that was held.
MAD Academy Training Video · 0:45
Two Different Questions
Horizon is when you need the money and tolerance is what you can watch happen. The lower of the two governs.
This lesson is part of a Stock Alerts + Tools plan.
Three separate questions
| What it is | How it is established | |
|---|---|---|
| Horizon | When the money is needed | A fact about a plan, and mostly knowable |
| Capacity | How much variability the plan can absorb | Arithmetic: contributions, other assets, flexibility of the goal |
| Tolerance | How much variability the person will hold through | Behavioural, and only reliably revealed in a decline |
Someone can have a thirty-year horizon, ample capacity, and no tolerance for a thirty percent decline. In that case the tolerance binds, because the allocation that will be abandoned is not the one being held.
- HorizonWhen the money is needed. A fact, and mostly knowable
- CapacityWhat the plan can absorb. Arithmetic
- ToleranceWhat the person will hold through. Revealed only in a decline
- The binding constraintWhichever of the three is lowest
Why horizon matters
A longer horizon changes what a decline means. Money needed next year cannot recover from a decline; money not needed for twenty years has time for one, and the historical record shows that longer holding periods have narrowed the range of realised equity outcomes considerably.
This is frequently overstated into a claim that equities are safe over long horizons. The historical record for one market over one period is a limited sample, and long periods of poor real returns have occurred. The narrowing is real; a guarantee is not.
The more robust version is about the sequence rather than the average: money not needed soon does not have to be sold at a low, and the ability to avoid selling at a low is most of what a horizon buys.
Why stated tolerance is unreliable
Tolerance is usually established by a questionnaire completed in a calm period, asking how a hypothetical decline would feel. The answers systematically overstate what the person will do when the decline is real, their account balance is the one falling, and the news is explaining why it will continue.
- A hypothetical loss is imagined; a real one is experienced, and the two are not the same stimulus.
- Declines arrive with narratives that make continuation feel likely, which is absent from a questionnaire.
- Prior experience of a full cycle is the single best predictor, and many investors have not had one.
- Behaviour in the last decline is far better evidence than any answer about the next.
The practical version: the allocation that matters is the one that will still be held after a bad year, and the best available evidence about that is what was actually done during the last bad year.
Horizon is rarely a single date
Treating a portfolio as having one horizon is usually a simplification. Money is needed in instalments over decades, which means part of it has a short horizon and part a very long one.
That is the reasoning behind approaches that segment a portfolio by when the money is needed, holding near-term requirements in stable assets and long-term ones in growth assets. It is a way of making the horizon explicit for each portion rather than averaging it into one number.
This describes a structure rather than recommending one. Whether segmentation suits a particular situation depends on circumstances a professional would assess.
Estimating tolerance more honestly
Stated tolerance is unreliable, and a few substitutes are better because they rest on something other than a hypothetical.
- What was actually done in the last substantial decline, which is direct evidence rather than a prediction.
- The dollar amount rather than the percentage: a thirty percent decline reads differently expressed as a specific sum.
- Whether the plan continues to work if the portfolio is worth substantially less for several years.
- Whether contributions would continue during a decline, which is when they matter most.
The second item is the most effective single change. Percentages are abstract and amounts are not, and the same allocation frequently reads as acceptable in one form and unacceptable in the other.
Someone with no experience of a full cycle has no direct evidence available, which is worth acknowledging rather than substituting with a questionnaire. The conventional response is to start at a level that is comfortable and adjust after the first genuine decline, which is when the evidence arrives.