Analysis and results
Read percentile paths
Understand P5 through P95 as ranks among simulated results, and use downside, median, and upside paths without mistaking them for promises.
What P5, P10, P50, P90, and P95 mean
A percentile is a rank among simulated results. P95 means that 95 percent of the results are at or below the displayed value and 5 percent are above it. It is an optimistic modeled outcome. P10 means 10 percent are at or below the value and 90 percent are above it, so it is a downside view. P50 is the median: half are below and half are above.
The direction matters. A higher balance is usually better, so P95 is strong and P10 is weak. For a measure where lower is better—such as tax, drawdown, or shortfall—the practical meaning may reverse. Always identify the measure, year, units, and dollar basis before interpreting the percentile label.
P95 does not mean YARCalc is 95 percent confident that the household will receive that amount. It says where that amount ranks inside this particular collection of simulations, using this saved plan and these model settings.
Choose percentiles that match the decision
Use a downside percentile such as P10 when asking whether essential spending or a commitment remains supportable in difficult modeled markets. Use P50 to understand the middle of the simulated range. Use P90 or P95 to understand upside and legacy potential, not to justify spending that only works in unusually strong outcomes.
No single percentile is the correct retirement plan. A useful decision normally considers the downside, the middle, probability of success, the first years in which depletion occurs, and what the household would actually do if results weaken.
Identify exactly what has been ranked
Before comparing percentiles, confirm the result measure and time. Ending-balance percentiles rank balances at the planning horizon. A percentile line on an annual chart ranks values separately in each displayed year. Success probability ranks paths by whether they met a success rule; it is not itself a balance percentile.
Also confirm whether the chart uses nominal or today dollars, which accounts or assets it includes, whether values are before or after tax, and which stochastic model, run count, seed, planning horizon, strategy, and lifespan settings produced the sample.
Use the percentile range in Stochastic Analysis
Begin with P10, P50, and P90. That gives a compact downside, middle, and upside view. If a tail decision is important, add P5 or P95, but recognize that extreme percentiles require more trials to estimate steadily than the median.
Compare the same percentile across scenarios or strategies only when the underlying analysis settings are the same. When two alternatives use a shared seed and paired paths, differences are easier to attribute to the plan change rather than to a different random sample.
For an irreversible decision, look for a conclusion that holds across a range rather than relying on one percentile. For example, a home purchase that works only at P90 has a different risk profile from one that preserves essential spending at P10.
Why a percentile line is not one simulated life
For each year on an annual percentile chart, YARCalc sorts all simulated values for that year and selects the requested rank. The simulation supplying P10 in one year may not be the simulation supplying P10 in the next year. The displayed line can therefore combine points from different underlying paths.
That construction is useful for showing how the distribution changes over time, but it should not be narrated as one household experiencing that exact sequence. To investigate sequence, depletion, or a specific stress history, use the path-level and timing information provided by the analysis rather than treating the percentile line as a literal biography.
P50 is the median, not necessarily the arithmetic average. A small number of very large outcomes can pull an average upward while leaving the median much lower. Percentiles avoid that distortion but do not describe the distance between every result.
Read width, movement, and thresholds together
The distance between P10 and P90 shows how widely results are dispersed. A widening band means uncertainty is accumulating. A falling downside line can matter even while the median rises, because the plan may be gaining upside and downside exposure at the same time.
Check whether a change improves the percentile that matters without creating an unacceptable tradeoff elsewhere. Delaying retirement might improve P10 balances but sacrifice desired time; a larger cash reserve might reduce early downside while lowering P90 growth; a spending guardrail might increase success by requiring cuts in weak paths.
When P95 or P5 changes sharply between otherwise identical runs, verify the trial count and repeatability. Tail estimates are based on fewer observations than the center of the distribution and naturally move more from sample to sample.
Percentiles describe the model, not the boundary of reality
P5 is not the worst event that can happen, and P95 is not a ceiling. Outcomes outside the displayed range exist within the sample, and real life can produce conditions absent from the model.
Percentiles inherit every assumption and limitation of the selected stochastic model: return distribution, correlations, inflation, taxes, spending behavior, lifespan, shocks, and the saved plan itself. Changing those assumptions creates a different distribution and therefore different percentile values.
Use percentiles to compare exposure and resilience, not to assign a precise probability to one dollar amount. A sound planning conclusion remains understandable when expressed as a range and paired with a response if the downside begins to develop.
Continue on the screen that uses this concept
Use this reference while reading the screen that produces the percentile paths.