Aftcast is a way of simulating equity curves for different start years, usually from a large sample data ~100 years. Its kind of like start date sensitivity testing but in my case, I incorporated withdrawals from portfolio. What I am trying to see is how sustainable a portfolio would be if I were to withdrawal at different fixed rate percentages of initial portfolio starting value. (Withdrawals are also adjusted for inflation)

After adjusting portfolio return for inflation, I extracted the return and then looped over each day to calculate next period portfolio value (previous portfolio value * 1+return), and if the next period is the withdrawal period (every december of each year), i subtracted the fixed amount (0.06 if 6%). The equity curve...

enter image description here

Not very sustainable from first glance. The problem I am having is to generate an aftcast for this entire stream of return. By this I mean I will extract the daily return and then find the equity series for each rolling 40 year period. So 1900-1940 will give me one equity series, and 1901-1941 another....etc.

But the problem comes in here...

enter image description here

The returns are normal until around 1925 when there is great volatility in return (+-100%). This is caused by the relationship between withdrawal rate and portfolio value. As the portfolio value approaches zero in 1925, the fixed withdrawal rate becomes a big percentage of portfolio value. This gives aftcast scenarios way out of whack..

enter image description here

Sorry for the long post. Is there anyone out there that have experience with generating aftcast scenarios? Am I misinterpreting some aspect of calculating portfolio value after withdrawals as if I lower the withdrawal rate to 0.01, everything is normal..

enter image description here This is very urgent, any suggestions might help! Thanks so much.

  • $\begingroup$ I've merged all of your accounts, as per reused IP addresses. Make sure the result is what you want. I also notice that you and @JimOtar are both in Ontario. $\endgroup$ – chrisaycock Sep 17 '12 at 15:39
  • $\begingroup$ Yes, Jim invented Aftcast for portfolio sensitivity testing if I am not mistaken.. $\endgroup$ – user1234440 Sep 17 '12 at 20:10

When I first looked at this problem in the year 1999 using Monte Carlo, I realized that MC simulation do not work for modeling market events, especially when withdrawals are present; well, for several reasons (see my articles for that at my website). Thus, I developed my aftcast model in 2001 while writing my book "High Expectations and False Dreams". After suggestions of fellow advisors, I made it available to fellow/advisors and general public in 2004. It might be easier for to download it and use it instead of reinventing the wheel.

Yes, toward the end of the portfolio life, once the withdrawal rate exceeds a certain threshold, some parameters (asset mix, rebalancing, and so on) become irrelevant and you might get transient and -at times- unbounded results. Don't use percentages, it is irrelevant to the client. Use dollar amounts. That solves this problem, the portfolio disappeares smoothly, the client is broke, and the advisor has no more trailer fees from it. It is similar to cutting the proverbial chicken's head (Nicholas Taleeb's chicken in his "Black Swan" book) and then holding the chicken and waiting until the chicken dies, instead of measuring the blood pressure of that chicken while it dies trying to figure out if the chicken is indeed dying.

If your objective is to calculate the sustainable withdrawal amounts for various time horizons, asset mixes, equity benchmarks, real yields for IIB's and so on, use my calculator, it is a "single-button" answer. Enter $1 million in current retirement assets, and then push the "How much can I have?" button, it will run a few seconds and tell you the amounts for worst case (green zone) and unlucky case (bottom decile).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.