← Methodology

Measurement / a visual guide

One question.
Two ways to examine it.

Alec has a yearly spending amount he wants to examine. His income, savings, obligations, and timing give that question a structure. Then two different kinds of evidence put it to the test.

Alec.

He wants to know how the life he imagines might fit the resources he has.

His example is fictional; the method behind it is real.

Conceptual illustration · No computed household dollar resultsPositions and distances in the drawing are not a money scale.Read the working detail ↓
Alec / yearly spendingOne structure · two kinds of evidence
Every stress case and model lens examines the whole household structure; a tile’s position does not assign it to one column. This drawing shows relationships, not Alec’s calculated dollar results.

Alec / his question

What could his life support?

Alec starts with an amount he would like to spend each year. It is a question he brings to the plan, not an answer the model hands him.

The starting lineHousehold input
Target SpendingAlec’s yearly spending amount to examine.

Alec / the Estate

Give that question a structure.

His income, Estate, obligations, and timeline make this his question. The same entered facts and assumptions remain in the drawing while two different examinations follow.

Held in the drawingInputs, not findings
IncomeWhen money enters.
EstateWhat resources are available.
LoadsWhat the structure must carry.
TimeWhen each part matters.

Instrument one / fixed stress plate

What if the years get difficult?

Alec’s same plan meets five severe cases. Four replay periods from U.S. market history; one is constructed. He can set how many of these tests his Floor must pass.

Five different pressuresFour historical · one constructed
  1. 1929 · Great DepressionHistorical replay / deep early loss and a long disruption.
  2. 1966 · Great InflationHistorical replay / years of pressure on buying power.
  3. 2000 · Dot-com bustHistorical replay / an equity decline with a drawn-out recovery.
  4. 2007 · Financial crisisHistorical replay / losses through a severe financial shock.
  5. Twelve-year stagflationConstructed stress / sustained weakness after inflation.
Five fixed cases / one requirementLoad Test

Alec sets how many of these cases a spending level must pass to qualify as the Floor. That number is a test requirement, not a probability.

A Studio run checks the cases against entered facts. This illustration does not assign Alec a pass or fail result.

The stress finding / Floor

The Floor has a reason.

Alec’s Floor is the highest yearly spending level that meets his selected fixed-stress requirement under the current model rules. It is a planning boundary, not a promise that every future will be safe.

Where the Floor comes fromFixed stresses, not Monte Carlo
Five case findingsChecked against the number Alec requires
FloorThe highest spending level meeting that test

Instrument two / modeled lens

What might other years look like?

Now Alec’s plan meets modeled market and economic sequences. He can use Datumae Blend, which combines four separate lens answers, or examine any one lens on its own. This is a different kind of evidence from the five fixed stresses.

Five Model Design choicesFour lens readings + Blend
Datumae BlendFour lenses, 10,000 modeled paths each. Their four answers are weighted equally; the 40,000 paths are not pooled into one result.

Or examine one 10,000-path lens on its own

  1. Parametric · its own upper-band answerMathematical return sequences under stated assumptions.
  2. Historical · its own upper-band answerObserved U.S. years sampled individually, not replayed in their old order.
  3. CAPE · its own upper-band answerA parametric reading adjusted for the starting valuation.
  4. Regime · its own upper-band answerFive-year blocks of real history, drawn mostly from years that began at market valuations like today’s.

One question / two readings

Where does Alec want to live?

The stress-tested Floor and modeled Ceiling give his spending question context. Alec can examine a higher or lower spending line and see how his plan behaves there. Neither boundary tells him what to choose.

Read the completed drawingThree different roles
FloorFrom Alec’s fixed-stress requirement
CeilingFrom his selected Model Design
Target SpendingAlec’s amount to examine. Moving the line here illustrates a question, not a calculated result or recommendation.

Working detail / current engine rules

How the Range is formed.

The drawing shows the different jobs each part performs. Here are the current model rules behind those readings. They describe modeled behavior, not a forecast.

01 / What “delivered” means

How much of the planned spending arrived?

For a future being tested, the engine adds up shortfalls below planned spending over all years and compares them with the total planned spending. The resulting share delivered is averaged across the examined futures at each spending level.

Planned spending includes the model’s one-person step-down where applicable. Healthcare is paid first and is not part of that planned-spending total. A share delivered is not a chance of success.

02 / Fixed-stress Floor

Five cases. A selected requirement.

The Floor is the highest annual spending amount at which at least the selected N of five fixed stress replays each deliver at least 99% of planned spending. The result is rounded down to $1,000. This calculation uses fixed replays, not Monte Carlo paths.

Four cases replay U.S. market history through the household’s stock, bond, and cash mix. The fifth is constructed stagflation. The household’s mix and relevant assumptions can change this finding; switching Model Design does not.

03 / Modeled paths and Ceiling

A different kind of reading.

A modeled path is one simulated sequence under stated assumptions, not a forecast. Each lens examines 10,000 paths and finds the spending level where average share delivered reaches 90% on the engine’s spending grid. This supplies that lens’s Ceiling answer.

Blend averages the four separate lens answers with 25% weight each, then rounds to $1,000. It does not pool their 40,000 paths into one distribution. Foundation and Keystone use separate 97% and 94% delivered thresholds; they are not extra public Range boundaries.

04 / The lenses are not the stresses

Five Model Design choices.

  • Datumae Blend: an equal weighting of four separate lens answers, each formed from 10,000 paths.
  • Parametric: log-normal paths using the household mix’s expected return, glide-path volatility, and a mild pull toward the long-run average.
  • Historical: real U.S. years from 1926–2025 sampled one at a time. It preserves the sampled year’s size, not its original order.
  • CAPE: parametric paths whose expected return responds to the starting valuation. No undated current valuation figure is assumed here.
  • Regime: five-year blocks of consecutive real years, drawn mostly from periods with comparable starting valuations.
05 / Spending can respond within a path

The guardrails matter.

Once working pay no longer carries the plan, the model compares the withdrawal rate with its starting rate. Above 120%, and above what the portfolio can sustain, modeled spending is cut 10%; below 80%, it rises 10%. Adjustments compound. The same rule runs in the Floor’s fixed replays.

That is one reason “share delivered” is more informative here than treating every path as a simple pass or fail. The rule describes the model’s behavior, not advice for when Alec should change real spending.

06 / The chosen line

Alec keeps the decision.

Target Spending is a household input. No Floor, Foundation, Keystone, or Ceiling rung is derived from the chosen amount. The engine assesses how Alec’s plan fares at his selection; the other marks give that assessment context.

Change the Estate, income, timing, mix, stress requirement, or another relevant assumption and the resulting Range can change. These modeled illustrations are not predictions, guarantees, investment recommendations, or personalized financial advice.

First looks are different. Standalone Sketch and Studio’s Retirement Baseline use a shared simplified calculation. Studio can center later growth paths on entered holdings, so matching methods do not guarantee identical outputs. Sketch calculates locally while you use it; signed-in saving stores work in a Datumae account, and continuing into Studio carries inputs forward.

About this page. Alec is fictional. The drawings show analytical roles and provenance, not his calculated results. These rules should be reviewed as the engine changes. For the wider seven-phase map, read the Method.

From a conceptual drawing to your own questions

The Range is a way to
see the trade-offs.

It cannot choose a life or a spending line for Alec—or for you. Studio is where your own facts, choices, and assumptions can be examined together.

Explore your structure in Studio Return to How the Range Is Formed