Portfolio Charts

[dualstow]

It would be fairly trivial to write a program that tried every combination in, say, 1% increments to find the maximum number of dark green cells, or the greatest number of dark green cells with the minimum number of some other color.  But what would this mean?

The optimum allocation last year,

Well of course, of course. These charts make us think, and  they don't have to lead to a magical optimum to give us insight. You have noticed the facetious comments and smiley faces in my & Tyler's recent exchange about dark green cells.

Having said that, I'd be very curious if an all mid-to-dark green chart were to come to light, and I'd want to know what was behind it. For example, some charts make you think about what happened in the 1970s, as has been mentioned earlier in this thread.

The chart that I created with the numbers above have fairly-light-to-white lines slicing through them for three years: 1987, 1998 and 1999. That wouldn't mean anything to someone who hasn't read the relatively recent history of the stock market and other assets. To others...

And it's interesting to think that someone could invest on January 1st of 1987 with said portfolio, stay the course, and not realize even after 28 years that they would have done so much better having started at a different point in time. I can't even think of an analogy for that terrible thought. (A good argument for rebalancing, continuing to save and continuing to reinvest, right?)

And yet, when I think about people who never stay the course due to frustration and performance chasing, I always imagine that the result would look something like starting at the top left cell of pretty much any of these charts and reading DOWNWARD, sticking to the leftmost column. With more red dots than the face of an 8th grader, that leftmost column makes the "1987 anomaly course" horizontal light green line look like a dream. That may or may not be an accurate depiction, but it's a powerful visual one can show a family member who's always changing course.

So then, what would a field of dark green mean? It depends on the person to whom you're showing it.

[Libertarian666]

If you let me pick investments after the fact, I can obviously come up with fabulous results.

In fact, I just got a solicitation of sorts for a variable annuity that has fabulous results... obtained by backtesting.

Needless to say, I didn't bite.

[dualstow]

Needless to say, I didn't bite.

Yes, sticking with your gold-heavy portfolio. You sure showed them. 

[Tyler]

It would be fairly trivial to write a program that tried every combination in, say, 1% increments to find the maximum number of dark green cells, or the greatest number of dark green cells with the minimum number of some other color.  But what would this mean?

The optimum allocation last year, or over the last 5 years, or over the last 30 years is almost certainly not the optimum next year, or over the next 5 years, or over the next 30 years.

This relates to Harry Browne's rule #4: No one can predict the future.

Everyone knows the past.  If determining the optimum portfolio allocation were as simple as maximizing some backward looking metrics there would certainly already be some smart fund manager who would be doing this.

Of course this is correct, and I hope my previous words were read with the correct facetious tone.

I've actually had an optimizer tool for several months that can do is, and recently figured out a way to make it simple enough to easily publish online with no special code. But I've been holding back until I figure out something to do with it that doesn't steer people in a deceptive direction. The hard part is not figuring out HOW to optimize historical data, but WHAT to optimize that is actually helpful for someone today.

So if simply maximizing returns or minimizing volatility over a certain historical timeframe isn't useful, what is?  Since I can also study multiple historical periods simultaneously (like the pixel chart) I feel like there's probably some information in there that people may find legitimately useful. Taking suggestions!

[Pointedstick]

I'm really coming to believe that the best investment portfolio you can hold is one that you are emotionally comfortable with, and can implement on a platform that lets you automate contributions and purchases and forget about it entirely.

[Mark Leavy]

So if simply maximizing returns or minimizing volatility over a certain historical timeframe isn't useful, what is?  Since I can also study multiple historical periods simultaneously (like the pixel chart) I feel like there's probably some information in there that people may find legitimately useful. Taking suggestions!

What I've found useful is to optimize for maximum CAGR while not exceeding a maximum draw down.  For small DD limits, this forces a portfolio to be consistently smooth across all time periods.  I've somehow convinced myself that if the portfolio content is relatively simple (i.e. not hundreds of assets) and that the "action rules" are also limited and simple, then a portfolio with smooth steady growth across all historical timeframes is probably a good all weather asset mix - and the backtesting is showing something stronger than just data fitting.

Another way that I look at optimizing back testing is to assume that this shows me the "Best Case" scenario for that asset mix.  I would assume that any portfolio following this mix and set of rules would always produce worse future results than the optimized backtested scenario.

I use the optimized backtested results as an UPPER LIMIT of portfolio performance and not the expected performance.

Expected future portfolio performance would be something between 0 and the upper limit.

That's my theory anyway...

[Tyler]

I'm really coming to believe that the best investment portfolio you can hold is one that you are emotionally comfortable with

I totally agree.  Lacking a crystal ball, one of my meta goals is to provide information that allows people to feel emotionally comfortable with their investing decisions. 

So maybe we can start there.  What makes you emotionally comfortable with investments?  What fears make you explore other options, and what information might quell those fears?  I realize this will differ for everyone, so I'm interested in all perspectives.

[rickb]

So if simply maximizing returns or minimizing volatility over a certain historical timeframe isn't useful, what is?  Since I can also study multiple historical periods simultaneously (like the pixel chart) I feel like there's probably some information in there that people may find legitimately useful. Taking suggestions!

I think what might be useful is showing how much variation there is in "optimal" allocations when you optimize over different backtested time periods.  For example, if you take 20 year periods starting in 1972, and optimize for each one, how different are the optimal allocations - and what do the out of sample results look like?  There's certainly no harm in showing optimized allocations over your entire data set, but IMO it needs to come with a very strong disclaimer. 

[Libertarian666]

I'm really coming to believe that the best investment portfolio you can hold is one that you are emotionally comfortable with

I totally agree.  Lacking a crystal ball, one of my meta goals is to provide information that allows people to feel emotionally comfortable with their investing decisions. 

So maybe we can start there.  What makes you emotionally comfortable with investments?  What fears make you explore other options, and what information might quell those fears?  I realize this will differ for everyone, so I'm interested in all perspectives.

Since you are asking everyone, what makes me comfortable with my investments is knowing that short of another Chixulub event, I won't be wiped out. CAGR is secondary, and volatility tertiary. Although I also consider tax effects, at this point they are almost insignificant in my situation.

[lordmetroid]

I am most at ease when I am invested in the best perceived option, still I hawk my investments everyday to reposition my investments when I notices a downtrend.

[Mountaineer]

I'm really coming to believe that the best investment portfolio you can hold is one that you are emotionally comfortable with

I totally agree.  Lacking a crystal ball, one of my meta goals is to provide information that allows people to feel emotionally comfortable with their investing decisions. 

So maybe we can start there.  What makes you emotionally comfortable with investments?  What fears make you explore other options, and what information might quell those fears?  I realize this will differ for everyone, so I'm interested in all perspectives.

Comfortable:
* Investing policy statement that my wife and I review quarterly with well documented reasons for all changes.  Basically, measure several times before sawing and understanding it is more important to keep what we have than gamble on a percent or two higher CAGR.  Enough is enough, that is being satisfied with what we have and understanding how we will adapt if major source of income losses occur.  I am somewhat of a Larry Swedroe fan who speaks of the emotional stuff as the stomach acid test. 
* Simplicity, i.e. minimum number of funds necessary to meet my investing goals.  Reason: make it easy for my wife if I die first.
* Stable, reputable, company (e.g. Vanguard and my local bank) that has been in business for several decades.
* Low cost of investing.
* Understanding that is almost impossible to predict the future, thus, invest similar to the Boglehead method by buying the markets.
* Investments that require very little monitoring - e.g. a few minutes per quarter of my time.
* Realizing there is far more to life than obsessing over money, or for that matter, obsessing over much of anything.

Fears:
None if I stick to what makes me comfortable.  Fears based on variables that are outside my control are only speculative and not actionable - in my opinion.

Information:
I would need a very strong logical case made before I adopted something other than "buying the global markets".  Buying the markets is a forward looking strategy that recognizes there is no reliable crystal ball, whereas most other strategies seem to be looking in the rear view mirror.

That was off the top of my head without a lot of thought.  I'm interested in what others say on the topic.

... Mountaineer 

[Tyler]

What I've found useful is to optimize for maximum CAGR while not exceeding a maximum draw down.  For small DD limits, this forces a portfolio to be consistently smooth across all time periods. 

I like that.  It's sorta like finding the efficient frontier, but defining it in terms that actually mean something to real people.  I'm already working on an idea along those lines. 

Max drawdown is a good filter.  I also like something like worst 10-year CAGR (the max time a reasonable investor is likely to stick with something that isn't meeting expectations).  If I can find a few more specific pain points, I can maybe build a helpful calculator to allow the user to look at the problem from a few different perspectives.

I think what might be useful is showing how much variation there is in "optimal" allocations when you optimize over different backtested time periods.  For example, if you take 20 year periods starting in 1972, and optimize for each one, how different are the optimal allocations - and what do the out of sample results look like?  There's certainly no harm in showing optimized allocations over your entire data set, but IMO it needs to come with a very strong disclaimer. 

Good idea!  FWIW, I also like to look at different allocation percentages not so much from an "ideal allocation" standpoint, but more from the perspective of offering a different way of looking at things for the "just put it all in stocks because returns are all that matter" crowd.  I understand it's a fine line and the messaging is very important. 

[Reub]

That was just beautiful!  Almost had to wipe away a tear from my eye.

[Dieter]

Maybe this has been mentioned, but in the pre-defined portfolios, such as http://portfoliocharts.com/portfolio/merriman-ultimate/, the data includes CAGR, Average Return, and Standard Deviation, while the Pixel Chart doesn't (if the URL copying works -- https://onedrive.live.com/view.aspx?cid ... xWoeuQxMsA&) -- was trying to compare the Four Fund version/TrevH Merriman Portfolio with the Merriman portfolio.)

Would be nice to have the same info.

[Tyler]

That's true.  I had to be selective because of space constraints. 

You can find CAGR and SD in the Funnel chart, though.  http://portfoliocharts.com/portfolio/funnel/

[dualstow]

Tyler, I apologize if this has already been asked.  Can't remember.
Any plans to add corporate bonds, like VCIT?
Perhaps the data doesn't go back far enough? Or, it's not popular?

[ochotona]

I propose one mega SWR chart with all of the lazy portfolios on one chart.

With the bars lined up from tallest to shortest, left-to-right. Sort based on the 30 year horizon (they might not sort the same way for all decades). Different colors for each portfolio, too.

[Tyler]

Tyler, I apologize if this has already been asked.  Can't remember.
Any plans to add corporate bonds, like VCIT?
Perhaps the data doesn't go back far enough? Or, it's not popular?

That's probably the single asset I'd most like to add.  While it's not readily available in the Simba spreadsheet for some reason, it's also an asset class where I know good long-term data exists because some retirement studies quote it.  I think long term corporate bonds are included in the Ibbotson SBBI classic yearbook, so at some point maybe I'll break down and either buy a copy or transcribe what I need from the library.  If anyone here has easy access to corporate bond data going back to at least 1972 (finance major?) please send me a PM.  Or other interesting asset classes, for that matter. 

[Pointedstick]

I think MG does. He might be willing to share the next time he pops by.

[dualstow]

I think MG does. He might be willing to share the next time he pops by.

Whose turn is it to put electrolytes and omega three fatty acids on his ashes, that He may arise once again?

[Mountaineer]

I think MG does. He might be willing to share the next time he pops by.

Whose turn is it to put electrolytes and omega three fatty acids on his ashes, that He may arise once again?

Perhaps he moved to Phoenix and has not yet been found by the internet.

... M

[Tyler]

I propose one mega SWR chart with all of the lazy portfolios on one chart.

With the bars lined up from tallest to shortest, left-to-right. Sort based on the 30 year horizon (they might not sort the same way for all decades). Different colors for each portfolio, too.

Ah -- now that's a good idea.

I've resisted publishing a cold optimization calculator because making a "recommendation" without also curating the results for people who do not fully understand the nuance of the calculations could harm more people than it helps.  Just because the results for a given set of assets are "optimized" doesn't mean the derived portfolio is a wise choice.  So rather than providing a calculator to find the efficient frontier for Nigerian bonds, lottery tickets, and payday loans, perhaps simply creating some nice comparison/filtering tools to help people select one of many known good lazy portfolios that's right for them is a more helpful goal. 

[Dieter]

I propose one mega SWR chart with all of the lazy portfolios on one chart.

With the bars lined up from tallest to shortest, left-to-right. Sort based on the 30 year horizon (they might not sort the same way for all decades). Different colors for each portfolio, too.

Ah -- now that's a good idea.

I've resisted publishing a cold optimization calculator because making a "recommendation" without also curating the results for people who do not fully understand the nuance of the calculations could harm more people than it helps.  Just because the results for a given set of assets are "optimized" doesn't mean the derived portfolio is a wise choice.  So rather than providing a calculator to find the efficient frontier for Nigerian bonds, lottery tickets, and payday loans, perhaps simply creating some nice comparison/filtering tools to help people select one of many known good lazy portfolios that's right for them is a more helpful goal.

+1

[lordmetroid]

Could you expand the commodities section.
I am particularly interested in oil because it is the life blood of modern civilization and silver as people suggest it to be gold on crack.

[lordmetroid]

I also came to think about an interesting experiment. Would be interesting to see what asset allocation the mathematics would produce.
Could you calculate the historically best portfolios by optimization? Perhaps three categories would be nice: CAGR, volatility and CAGR+volatility.