Portfolio Charts: Question about Safe Withdrawal Rate

[SeattleCPA]

One of my favorite investing website resources is @Tyler 's www.portfoliocharts.com website. Wow. Amazing!

But I have a question surely someone here knows answer to or maybe even Tyler will answer.

First some background though: Portfolio Charts does many cool things but one of those things is estimate safe withdrawal rates for different popular asset allocations based on their historical performance. The worst historical case represents a safe withdrawal rate and shows up as an orange line as displayed below:

[img=https://portfoliocharts.files.wordpress.com/2017/03/wr-classic-60-40.jpg][/img]

BTW, those other blue lines show all the other historical scenarios. Some are shorter and some are longer because of course, the three or four decade scenario that started in 1990 or 2000 has not yet completed.

But here's my question: While the orange line worst case scenario plots out a safe withdrawal, its distance from the second worst case scenario is inconsistent. In other words, for some portfolios, the worst case scenario rubs right up against the second worst case scenario. This makes sense. (See the Permanent Portfolio chart or the Golden Butterfly portfolio chart.) But for other portfolios it looks as if there's a little, well, padding between the second to worst case scenario and the very worst case scenario. (See Swensen portfolio chart, for example.)

This page below includes an animation that demonstrates this for three asset allocations near the bottom of the discussion, for example:

https://portfoliocharts.com/2017/03/21/how-to-predict-withdrawal-rates-without-a-crystal-ball/

My question (sorry for long-winded lead up)... does anybody know if there's a little more to the orange line than a simple tracing of the worst historical case? E.g., is their "padding" based on the portfolio's standard deviation or something like that?

[Tyler]

My question (sorry for long-winded lead up)... does anybody know if there's a little more to the orange line than a simple tracing of the worst historical case? E.g., is their "padding" based on the portfolio's standard deviation or something like that?

Good question.

I use the same "Safemax" method promoted by William Bengen (author of the original SWR paper) where the orange line traces the real-life worst case scenario.  (I'm additionally able to extend that line for more recent start dates using methods covered in your linked article, but I think that's outside of your question).  However, some withdrawal studies use different definitions of how to derive the orange line you see in the charts.  For example, the most common interpretation of the Trinity study effectively draws the line by excluding the worst 5% of the scenarios, and some people use their success rate tables to get more and more aggressive with assumptions about how many past failures they want to dismiss as outliers.  I personally prefer to err on the side of caution and point out the very real worst cases, but obviously reasonable people may have different approaches to the same problem.

As for the variations of individual "padding", some portfolios are definitely more consistent than others and the spread of results is an indicator of that consistency across all start dates and economic conditions.  It's very similar to the spread of returns explanation outlined here.  And due to the way withdrawal rates work, that consistency is often an indicator of a portfolio with surprisingly high withdrawal rates.  Basically, the more dependable a portfolio the less likely you are are hit a start date where the WRs veer way off path to set a new low.

[SeattleCPA]

My question (sorry for long-winded lead up)... does anybody know if there's a little more to the orange line than a simple tracing of the worst historical case? E.g., is their "padding" based on the portfolio's standard deviation or something like that?

Good question.

I use the same "Safemax" method promoted by William Bengen (author of the original SWR paper) where the orange line traces the real-life worst case scenario.  (I'm additionally able to extend that line for more recent start dates using methods covered in your linked article, but I think that's outside of your question).  However, some withdrawal studies use different definitions of how to derive the orange line you see in the charts.  For example, the most common interpretation of the Trinity study effectively draws the line by excluding the worst 5% of the scenarios, and some people use their success rate tables to get more and more aggressive with assumptions about how many past failures they want to dismiss as outliers.  I personally prefer to err on the side of caution and point out the very real worst cases, but obviously reasonable people may have different approaches to the same problem.

As for the variations of individual "padding", some portfolios are definitely more consistent than others and the spread of results is an indicator of that consistency across all start dates and economic conditions.  It's very similar to the spread of returns explanation outlined here.  And due to the way withdrawal rates work, that consistency is often an indicator of a portfolio with surprisingly high withdrawal rates.  Basically, the more dependable a portfolio the less likely you are are hit a start date where the WRs veer way off path to set a new low.

Got it. Thank you Tyler.

BTW I really encourage anyone who hasn't already done this to spend some serious time reviewing the Portfolio Charts tools. Lots of really insightful stuff there. Tons and tons...

[Bicycle_B]

The Accuracy Chart is changing how I think about portfolios. I think it's finally convincing me to diversify a little more, and to be more systemic about doing so.

Will report back in a couple months or so.

[SeattleCPA]

The Accuracy Chart is changing how I think about portfolios. I think it's finally convincing me to diversify a little more, and to be more systemic about doing so.

Will report back in a couple months or so.

Me too. I'm glad it's there. Note though Tyler's description here, Target Practice: The Target Accuracy tool is great for planning but sort of visually understates the risk by lopping off the bottom and top 15%.

In my myth of the long run stock market chart blog post from last summer, the same sort of chart shows a pretty "wide" mouth (see below). Getting to a $500K by saving $10K a year might happen as quickly as 15 or so years... or take as long as, gosh, 36-38 years? And 15 years of saving $10K a year might get you nearly $500K or less than $100K.

[Bicycle_B]

Me too. I'm glad it's there. Note though Tyler's description here, Target Practice: The Target Accuracy tool is great for planning but sort of visually understates the risk by lopping off the bottom and top 15%.

Hmm, maybe I am misreading the chart.  Here are the details; if you can see where I got off track, let me know.

I read the tool as though the CAGR percentages refer to the middle 70%, which does lop off the highest and lowest 15%.  But in the chart itself, the middle band is shown in darker color, with little squares on the right hand y axis marking the upper and lower bound of the middle 70%. In other words, the little squares mark the 15 year result of the stated CAGRs. However, the chart also shows the actual best case and worst case in the data. These are marked by solid lines. The highest 15% and lowest 15% are not excluded, they are displayed on the chart in light color.

If it were up to me, I'd add the second box of CAGRs. I'd label the existing box "Likely CAGR". The second box would show the highest and lowest CAGR. I'd label that one something like "Highest/Lowest CAGR". Like you, SeattleCPA, I instinctively feel that the max boundaries in the data set are really important. Once I delude my myself into thinking that I understand the chart, I immediately (and worriedly) begin estimating the bottom boundary. 

That said, Tyler generally manages the trade-off between clarity and info overload really really well. I usually want more data that most people can stand.

The "Worst, Best and Average" chart posted upthread is thought provoking too.

[SeattleCPA]

Me too. I'm glad it's there. Note though Tyler's description here, Target Practice: The Target Accuracy tool is great for planning but sort of visually understates the risk by lopping off the bottom and top 15%.

Hmm, maybe I am misreading the chart.  Here are the details; if you can see where I got off track, let me know.

I read the tool as though the CAGR percentages refer to the middle 70%, which does lop off the highest and lowest 15%.  But in the chart itself, the middle band is shown in darker color, with little squares on the right hand y axis marking the upper and lower bound of the middle 70%. In other words, the little squares mark the 15 year result of the stated CAGRs. However, the chart also shows the actual best case and worst case in the data. These are marked by solid lines. The highest 15% and lowest 15% are not excluded, they are displayed on the chart in light color.

If it were up to me, I'd add the second box of CAGRs. I'd label the existing box "Likely CAGR". The second box would show the highest and lowest CAGR. I'd label that one something like "Highest/Lowest CAGR". Like you, SeattleCPA, I instinctively feel that the max boundaries in the data set are really important. Once I delude my myself into thinking that I understand the chart, I immediately (and worriedly) begin estimating the bottom boundary. 

That said, Tyler generally manages the trade-off between clarity and info overload really really well. I usually want more data that most people can stand.

The "Worst, Best and Average" chart posted upthread is thought provoking too.

I think you are right... and I was wrong. Sorry.

Good thing (for me)... at least I didn't make this error in public forum. :-)