Expected longevity is NOT static! This first appeared in blog post that describes the graphs in more details https://blog.betterfinancialeducation.com/sustainable-retirement/what-are-the-three-paradigms-of-retirement-planning/ Objective of the strategic use of longevity percentiles is to roll, i.e., model the aging effect, into lower and lower probabilities of outliving the corresponding end table age. The modeling strategy makes it more and more unlikely that the retiree will outlive their money, barring catastrophic spending, while still being able to optimize spending during those years they’re still alive at ANY AGE. What this eliminates … guessing or picking some arbitrary end age, typically age 95. What age is it then when retirees approach that age 95, and when would they change age 95 to some other age? Arbitrary results in the simulation using a single factor for the Monte Carlo time function on top of the single factor simulation of just one portfolio. What about other time periods and other allocation characteristics for other models (or even portfolio statistical characteristics of the not-yet-client “inefficient” allocation?). Why not just let statistics determine the table end age to derive the time period for the stochastic calculations? Below show how the ending table age slowly moves older and older as the retiree ages. Ending table age changes from age 93 to 107! It’s NOT a static age 95! Why does it matter? You can see the effect on the drawdown % (DR%) in the data cloud data and the fact that WHEN (at what ages) an allocation adjustment may be suggested by the data and methodology itself (both effects not seen under today’s paradigm). Oh, most people think that the table end age is when money runs out! Remember the use of 10% iteration failure rate? That translates into 90% of the iterations outlast that table end age! The LAST page illustrates how longevity and aging affects the decreasing rolling time periods which affects simulations too!