starting year for sampling

[palomalou]

The starting year of stock market /economy sampling in Firecalc makes a huge difference in the results. What starting date gives the most valid result? If I put 1871, I get 68 %; 1955, 37%; 1945, 56%. Thanks for your thoughts.

 

[Midpack]

I understand how the results would change as your reducing the data/cycles used, but what's the purpose of entering a starting year other than the default 1871? Unless you know of bad or unrepresentative data in early years, doesn't more data provide more meaningful results? You entered 1955 for example, that's only 27 cycles (based on 30 years) - many of us will be retired for 30-40 years.

 

[palomalou]

Thanks, Midpack. I just wondered whether I would get very different results, and wow, did I ever! No scientific reason to dispute the 1871, just wondering/fretting.

 

[youbet]

Thanks, Midpack. I just wondered whether I would get very different results, and wow, did I ever! No scientific reason to dispute the 1871, just wondering/fretting.

To track down starting years that result in outliers in your outcomes, drop a spreadsheet and run a histogram of the end year you're interested in. Also, just sort the column of that same year and note which years give you the worse outcomes.

The FireCalc run I do which I feel best tests my current situation (complicated by upcoming portfolio adjustments funding 529b's for the grandkids, setting up a trust for a special needs grandson, helping to fund MIL's future NH care and blaah, blaah, blaah) has a dip in year 5. The run recovers and is 100% successful with significant margin, but, still, I was interested in what beginning year ( and the econ condx of those times) resulted in a dip in year 5 given my parameters. I just requested the excel sheet and sorted year 5 from highest to lowest amounts and then noted which beginning year corresponded to the low dip.

There's a lot of info in the Excel sheets you can drop when doing the FireCalc runs.

And you're spot on that the starting year is a major factor in the outcome variability. Frankly, there isn't all that much you can do about it except to accept that even the best laid, conservative plans with lots of flexibility may be a wild roller coaster ride. It's just the way it is.

BTW, you're just kind of beating around the bush by changing the default 1871 starting year. Go in and actually look at the outcomes for all the starting years. It's not hard to do and very interesting, especially if you plot histo's and gasp at the fact that a set of parameters can result in going broke or dieing very, very rich depending on the years you're living your plan.

 

[Midpack]

The FireCalc run I do which I feel best tests my current situation (complicated by upcoming portfolio adjustments funding 529b's for the grandkids, setting up a trust for a special needs grandson, helping to fund MIL's future NH care and blaah, blaah, blaah) has a dip in year 5. The run recovers and is 100% successful with significant margin, but, still, I was interested in what beginning year ( and the econ condx of those times) resulted in a dip in year 5 given my parameters. I just requested the excel sheet and sorted year 5 from highest to lowest amounts and then noted which beginning year corresponded to the low dip.

I've also played with the spreadsheet results and it can be helpful and interesting indeed.

I'm confused by the exercise you describe above though. Picking out a FIRECALC start year(s) with a dip in returns at year five isn't the same as your known unusual expenses in year five. A dip in returns is not the same as unusual expenses, sequence of returns would actually compound the dip if you'd retired that year. Maybe you've already done it, but I'd be more inclined to enter the expenses you describe above as "off chart spending" or "lump sum changes (subtract)" in FIRECALC on those tabs. Those results may be even more representative of what you might have expected.

Apologies if I misunderstood your exercise or you've already tried off chart and or lump sum subtraction for comparison. FWIW...

 

[youbet]

I've also played with the spreadsheet results and it can be helpful and interesting indeed.

I'm confused by the exercise you describe above though. Picking out a FIRECALC start year(s) with a dip in returns at year five isn't the same as your known unusual expenses in year five.

Correct.

A dip in returns is not the same as unusual expenses, sequence of returns would actually compound the dip if you'd retired that year.

Correct again

Maybe you've already done it, but I'd be more inclined to enter the expenses you describe above as "off chart spending" or "lump sum changes (subtract)" in FIRECALC on those tabs.

That's what I did. Then I looked at the FireCalc output graph, noted that one starting year had an interesting dip in portfolio value at year #5 (not a dip in "returns" as you stated). I dropped the Excel data and sorted the year 5 outputs, just out of curiosity, to see what starting year resulted in the portfolio value dip in year #5. Interestingly, it was 1937. It's of no particular consequense other than contemplating how the parameters of my run interact with the FireCalc data beginning in 1937.

Just giving a simple example of how you can look at, sort, graph, run descriptive statistics, etc., on FireCalc output data. It's never led me to any significant actionable conclusions, but it's interesting. The biggest thing I've learned from repeatedly doing this is that ending portfolio values vary wildly depending on the years you're FIRE'd and there is little you can do about it. Learn to love variation!