Personal Investments • Re: How to "correctly" use backtesting

I could provide examples, but only if any of the commentary above made sense to you

Thank you very much for your detailed response. I'm kind of familiar with those tools in the links, but I really don't understand the assumptions and inner workings of the tools. I certainly would appreciate if you can provide some examples in Excel or other tools.

So, I tried to run Monte Carlo simulation based on historical mean & standard deviation (both in Excel and R programming language).

This was my "logic" (which apparently doesn't make sense based on some feedback I received from another post):

1. Determine the annual mean & standard deviation of my portfolio. Because of when my funds started, I couldn't get data from major market drops, such as early 2000s, 2007s, etc., so I'm not sure how good my stats are.

2. Determine the mean and standard deviation of inflation (general and medical) based on the available data from, say, 1968 to today so the data would include the very high inflation rates of the 70's to mid 80's.

3. Perform 1000 runs for each year (assumed to be 30 years) in retirement as follows:

RUN 1
Year 1
• Starting balance = current balance
• Based on mean & std, randomly find a rate of return
• Based on mean & std, randomly find general & medical inflation rates
• Use pension+social security (if applicable) + RMD (if applicable) to calculate federal & state taxes
• Calculate various expenses based on randomly determined inflation rates
• Return on investment = (starting balance - RMD if applicable)*rate of return
• Balance at the end of year 1 = Starting balance + return on investment + pension+social security – RMD - expenses - taxes
Year 2
• The same as year 1, except the starting balance is the balance at the end of year 1
Year 3-30
• The same as above by using the balance at the end of the previous year as the starting balance
• Store the balances at the end of each year for RUN 1

RUN 2
The same as RUN 1, each time resetting the starting balance to current balance

RUNs 3-1000
The same as above

At the end, I've a matrix of the ending balances with 1000 rows (one for each run) and 30 columns (one for each year)

Calculate the 10th, 50th, 90th percentile for each year

Plot the results vs age to see if I would run of money at the end of 30 years.

Does this logic make sense?

Thanks again for your response.

Statistics: Posted by Cincy_1988 — Sat Jan 25, 2025 6:37 pm

---