Thanks for trying out the planner and for the feedback. More comments inline...
2B said:
I am assuming the "median" withdrawal is what the original withdrawl is plus the "other" expenses (but inflation adjusted) times the % shown.
That's basically correct. However, since the simulation is run 10,000 times, the median withdrawal is the amount such that half of the trials had a lower number and half had a higher number for the given year.
It's important to keep in mind that unless the probability of success is 100%, some simulation trials ran out of money before the end of the plan. In those cases, the percent of expenses funded would hit 0%. That's why I report the median value rather than the average.
Overall, it's a pretty good application of the Guyton philosophy. Playing with the "other" costs lets me account for few expenses as I age -- ala Bernicke. CT should like the concept because it lets the "trailer on the trout stream" be the base with other expenses layered as desired.
I'm glad you made use of that feature. It was a bunch of work to do and now I know at least one person other than myself has used it. Incidentally, the additional inputs tab is also handy for simulating the effects of a market crash in year X of your plan. Just set the portfolio return to something like -20% and the std dev to zero for the year or years you want to simulate the crash in.
I am curious if it uses the "real" market returns like FIRECalc does. It's not intuitively obvious what variations drive the variables in the Monte Carlo analysis.
For simulating market returns, FRP only uses the given average return and standard deviation of returns (or multiple sets of them for different periods, if you choose). For each year in each of the 10,000 simulation iterations, the planner "draws" from a pseudo-random pool of returns that are normally distributed according to the average return and std dev given for that year.
Some critics point out that real market returns aren't normally distributed because they have fat tails or show something called kurtosis (or skew). I think this is valid criticism. However, you can roughly compensate for this by either lowering the return or increasing the standard deviation. The canned return/std dev pairs that I've provided to go with each investing style,are adjusted somewhat to account for this and for investment expenses. If you don't like what's there, you can just plug in your own return/std dev using the custom option.
Hope that answers your questions. If anyone wants more details on the inner workings of the planner, please don't hesitate to use the contact email listed on the web site. I obviously find this stuff pretty interesting and I've had a few interesting email conversations with users since the planner was launched.
Jim