siamond wrote:

A long time ago siamond posted the above. Over the weekend I tried to replicate his analysis. (We all get to define "fun" in our own ways, right?

)

My results are quite a bit different than his and came out in favor of Prime Harvesting. Since I make dozens of mistakes every day, it is entirely possible I have made yet another mistake. After the results, I'll include links to much of my raw data so people can point out mistakes and/or try to reconcile the differences between my results and siamond's. (I get tired of typing Prime Harvesting and "a 60/40 portfolio annually rebalanced" so I will sometimes call them PH and REB.)

On every metric I look at, PH seems to do noticeably better.

Here are withdrawals. PH has a median annual withdrawal of $54,332 versus REB's $50.754; 7% higher.

PH has a "certainty equivalent withdrawal" that is $1,148 higher than REB; further evidence that PH seems stronger. This makes the WER and HREFF-3 slightly higher. The Risk Quotient (from Milevsky) favors REB, though not as much as I expected it might, given how often PH deals you an asset allocation with a lot of stocks.

Here are portfolio values. PH does better on mean, median, standard deviation, and "average of 25% lowest" but the differences aren't big enough to mention.

PH has a lower maximum withdrawal than REB. That goes along with significantly lower kurtosis ("fat tails"). That suggests that PH has a lower upside but is more consistent about delivering good (instead of great) withdrawals.

Here's a histogram of all annual withdrawals between the two, which shows this a bit more clearly. Notice the large green hump starting around $60,000 (good) and the corresponding blue hump on the left (bad).

So...what were the assumptions that went into generating all of this? I tried to keep it on par with those siamond outlined but there may be inadvertent differences.

- I used the data from simba's spreadsheet.

- I used VPW to determine the size of the withdrawal. I used a 50 year depletion horizon. VPW adjusts its internal rate of return correctly for Prime Harvesting's variable asset allocation.

- The scenarios ran for 40 years.

- Due to the 40 year horizon, the data retirements begin in 1871-1975.

- i'm showing stats based on every year, not just the final year.

(Note, I included 10/90, 20/80, 30/70, 40/60, 50/50, 60/40, 70/30, 80/20, 90/10, 100/0, Prime Harvesting, and Alternate Prime Harvesting in the rollups but not the detailed year-by-year sheet.)

Here are the rolled up stats for withdrawals:

https://docs.google.com/spreadsheets/d/ ... =631965724
Here are the rolled up stats for portfolio values:

https://docs.google.com/spreadsheets/d/ ... 1038299973
Here is the raw data (showing every withdrawal of every retiree every year, along with their portfolio balance). This is probably the best place to start looking at why my results differ from siamond's:

https://docs.google.com/spreadsheets/d/ ... =569790987
Here are some other derived stats (certainty equivalent withdrawals, WER, HREFF-3, risk quotient):

https://docs.google.com/spreadsheets/d/ ... 1019841369