How Different Monte Carlo Models Perform In The Real World: Assessing Quality Of Predictiveness In Retirement Income Forecasting Models

It’s natural to assume retirement planning forecasts are reasonably good -but are they really? Check out the latest Kitces article co-authored by Justin Fitzpatrick and Derek Tharp to learn how some forecasting methods are better than others (and none is great).

Takeaways:

• Despite the fact that many people use what we deem as “traditional Monte Carlo” out of a desire to be more conservative than historical projections, the use of traditional Monte Carlo historically would have
  actually led to recommending spending levels that are less conservative than historical simulation, at least among the probability of success range most commonly used (e.g., 70% to 97%).
• Among commonly used success levels, historical simulation outperformed in terms of accuracy, and regime-based Monte Carlo performed better in terms of generating the more conservative estimates many advisors think they are getting with traditional Monte Carlo.
• Use the best you can find and update and adjust over time.

Read full article >>

Justin Fitzpatrick, PhD, CFA, CFP

Justin Fitzpatrick is President and Co-Founder of Income Lab, retirement income planning software used by thousands of financial advisors. He developed the guardrails-based approach to retirement income distribution after a decade in financial services at Jackson and seven years in academia at MIT, Harvard, and UCLA. His research on adjustment-based planning has been published on Kitces.com, ThinkAdvisor, AdvisorPerspectives, and FinancialPlanning Magazine.

View all articles by Justin LinkedIn