Portfolio Charts

Ask your investment, budget, and other money related questions here
Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: Portfolio Charts

Post by Tyler9000 »

Think of it this way -- the individual indices track actual index funds, and the return you see is the compound result at the end of the year (including dividends) that accounts for continuous activity. So even though the numbers on the site are annual, more precise data is already baked in. Because the site only rebalances annually, the overall portfolio numbers should be pretty accurate.

Unfortunately I don't have historical PE data for all of the asset classes, so it's hard to layer on too much of that kinda thing.

bryan
Posts: 1061
Joined: Sat Nov 29, 2014 2:01 am
Location: mostly Bay Area

Re: Portfolio Charts

Post by bryan »

Tyler9000 wrote:Yes, the data is all year-end. It's also assumed that any withdrawals and rebalancing are done on that date.
I've always wanted tools like firecalc or yours to randomize annual dates (for sampling/rebalance/withdrawal) or allow a specific annually repeating date to be entered. Seems like year end/beginning could be biased.

I myself randomly choose four rebalance dates once a year from possible weekdays (four random numbers between 1-262, basically). Though I am not good at resisting buying assets that fall sharply in the short term ("luckily" I usually have some cash sitting around not yet invested for those opportunities).

User avatar
jennypenny
Posts: 6851
Joined: Sun Jul 03, 2011 2:20 pm

Re: Portfolio Charts

Post by jennypenny »

Happy Birthday PC! ... CELEBRATING A YEAR OF PORTFOLIO CHARTS

It's a great site, Tyler.

BRUTE
Posts: 3797
Joined: Sat Dec 26, 2015 5:20 pm

Re: Portfolio Charts

Post by BRUTE »

weeeee they grow up so fast

Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: Portfolio Charts

Post by Tyler9000 »

Thanks! :) It really is sorta crazy how time flies.

BTW, special thanks to this forum for inspiring many of the ideas and providing great feedback that has helped me along the way. You guys & gals are awesome!

BRUTE
Posts: 3797
Joined: Sat Dec 26, 2015 5:20 pm

Re: Portfolio Charts

Post by BRUTE »

brute thanks Tyler9000 for bringing the concept of CAGR to his attention, which made brute feel like the entire financial industry has been lying to him for decades. and the idea that portfolio structure can have positive effects that none of the individual asset classes have. brute was firmly in the 100% stocks camp only a few months ago, and now is into both GB and GMP, thanks to Portfolio Charts and discussions with Tyler9000 here.

BlueNote
Posts: 501
Joined: Sat Jun 08, 2013 6:26 pm
Location: Toronto, Canada

Re: Portfolio Charts

Post by BlueNote »

@Tyler9000

Portfolio Charts is an awesome tool and I am glad you're still improving it.

I'm sure you've probably been asked for these features by now, but I'm going to point them out:

1) Different country level data (short bonds, long bonds, total bond market, real return bonds, stock market, REITS)

BTW there is a Canada centric backtesting tool here : http://www.ndir.com/cgi-bin/downside_adv.cgi, not as good as portfolio tools imho, but they do cite Canadian data sources near the bottom of the site so maybe it would be easy to get Canadian data, you could try asking the site owner :) .


2) A randomized simulation of returns based on the statistical characteristics of different asset classes. In my mind this sort of simulation would include well known statistical elements like stock market fat tails, momentum phenomena, value phenomena, asset co-relations, black swan possibilities etc. I think people would get an interesting perspective on backtesting if they tried something like that out. I'm imagining a tool that spits out a different answer, sometimes wildly different, each time it is run. However the results would be plausible given what we know about the historical data. It could be used for monte carlo simulations and It would be cool to see that sort of an engine output results into the same slick interface you created so we could compare the results against things like GB , minimize fat tails etc..

Dragline
Posts: 4436
Joined: Wed Aug 24, 2011 1:50 am

Re: Portfolio Charts

Post by Dragline »

jennypenny wrote:Happy Birthday PC! ... CELEBRATING A YEAR OF PORTFOLIO CHARTS

It's a great site, Tyler.
+Many

Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: Portfolio Charts

Post by Tyler9000 »

I'm just glad to see that people find it useful. :)

Data for multiple countries is still on my long-term wish list (thanks for the data tip). In the short-term I'm working on updating my existing asset choices a bit -- adding MCV and MCG and maybe eliminating a few redundant and lesser-used assets to reduce complexity that doesn't add value. BTW -- does anyone know of a good free source for corporate bond or international bond data since 1972?

Interesting idea regarding a simulation. I've actually toyed with the idea of randomizing fat tail events (crap -- I replaced a roof this year!) to model dirty, real-world things that spreadsheets often overlook. It just comes down to how to properly quantify realistic inputs and work around my personal aversion to most Monte Carlo methods that I don't think model real economies particularly well. But I do feel like there's a cool opportunity in there somewhere.

BlueNote
Posts: 501
Joined: Sat Jun 08, 2013 6:26 pm
Location: Toronto, Canada

Re: Portfolio Charts

Post by BlueNote »

Tyler9000 wrote: Interesting idea regarding a simulation. I've actually toyed with the idea of randomizing fat tail events (crap -- I replaced a roof this year!) to model dirty, real-world things that spreadsheets often overlook. It just comes down to how to properly quantify realistic inputs and work around my personal aversion to most Monte Carlo methods that I don't think model real economies particularly well. But I do feel like there's a cool opportunity in there somewhere.
Many people seem to place a lot of weight, too much in my opinion, on the relevance of historical data. As any accountant knows historical data is the most reliable type of financial data but when it comes to relevance it is often lacking and there are often other more relevant sources of data for making decisions. The fact is that even our best stock market data is very noisy and suffers from a high standard error page 25). I think a simulator could produce a more relevant result which could improve investment decision making by modelling investing outcomes from different scenarios. It would also be difficult to create something that accurately processes realistic inputs so I can definitely empathize with your aversion.

jacob
Site Admin
Posts: 15906
Joined: Fri Jun 28, 2013 8:38 pm
Location: USA, Zone 5b, Koppen Dfa, Elev. 620ft, Walkscore 77
Contact:

Re: Portfolio Charts

Post by jacob »

FWIW, in terms of quantitative investing, building a [good] simulator is exactly as difficult as building a good strategy. Think about it. If your simulator is able to simulate events accurately, you can easily construct a portfolio that outperforms when such events occur [as previously simulated] in reality. In particular, if you posses such a magical simulator, going from simulator-space to reality-space is just a question of translation. However, building a realistic simulator is hard.

For such planning, people do two things. One can either think in terms of scenarios and focus on accuracy or one can prefer precision and for that one looks to historical data. Historical data is more precise than any simulator. This is why historical data is widely used for quant investing. Of course this leads to most quantitative investors continuously preparing to fight the last war.---That with the general hope that statistically, all wars (past and future) are approximately equal in nature.

PS: Note the deliberate distinction between precision and accuracy.

stayhigh
Posts: 113
Joined: Sun Dec 06, 2015 4:20 pm

Re: Portfolio Charts

Post by stayhigh »

What's your idea for the next war?

Dragline
Posts: 4436
Joined: Wed Aug 24, 2011 1:50 am

Re: Portfolio Charts

Post by Dragline »

BlueNote wrote: I think a simulator could produce a more relevant result which could improve investment decision making by modelling investing outcomes from different scenarios. It would also be difficult to create something that accurately processes realistic inputs so I can definitely empathize with your aversion.
Actually, I think this would likely be impossible and/or useless for the same mathematical reasons the "three body problem" was shown to be unsolvable and you can't predict whether it will rain next year on this date. Also for the same reasons we can all predict that the stock market will crash and mass shootings will occur, but not when it will crash or who will be shot and where -- its what is known as an adaptive complex system.

BlueNote
Posts: 501
Joined: Sat Jun 08, 2013 6:26 pm
Location: Toronto, Canada

Re: Portfolio Charts

Post by BlueNote »

Dragline wrote:
BlueNote wrote: I think a simulator could produce a more relevant result which could improve investment decision making by modelling investing outcomes from different scenarios. It would also be difficult to create something that accurately processes realistic inputs so I can definitely empathize with your aversion.
Actually, I think this would likely be impossible and/or useless for the same mathematical reasons the "three body problem" was shown to be unsolvable and you can't predict whether it will rain next year on this date. Also for the same reasons we can all predict that the stock market will crash and mass shootings will occur, but not when it will crash or who will be shot and where -- its what is known as an adaptive complex system.
I think a simulator could provide useful information particularly if it were to roughly accurately model market reactions under various scenarios. I don't think that any simulator is going to tell you anything useful about when something will happen. However a good simulator should provide useful information about effects on asset values under different scenarios. So the simulator would be a complex model and we could feed it different inputs and get outputs that would be roughly right enough to be useful in reality. We could also adjust the settings/variables of the model itself so we can test how sensitive effects are to our assumptions. I would expect one of the main uses of such a tool would be to filter out strategies/allocations that backtest well but simulate poorly because the strategies success was simply a result of chance as opposed to being robust enough to do well in a vast majority of situations.

Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: Portfolio Charts

Post by Tyler9000 »

bryan wrote:
Lucky C wrote:
However, I don't think picking funds based on not being at all-time highs is a very good idea.
Would be nice if a tool that is a hybrid of some of the portfoliocharts.com stuff and robo-advisor tools. Basically allowing an input of 1) custom/highlighted asset allocation to test, 2) the asset allocation buy-in function (to what extant the investor is dollar cost averaging i.e. with an income and savings or excess cash) and buy picker (this week you would have bought 10 shares of GLD and 2 shares of TSM based on some metric, like asset is not in upper 95% of it's inflation adjusted price). And the resulting blog posts, paper on patterns seen (the more the lump-sum the better, better to start with buying the TSM portion, etc).

If @tyler9000 created such a tool with a nice UI like other robos and converted the excel-like work into something more "natural-to-the-web" like other robos, I think he could be acquired pretty quickly..
I cribbed this quote from another thread. I think Bryan just joined JennyPenny on my marketing team. ;)

I like your idea for something that helps people explore how to buy in. You mention robo-advisor tools -- do you have any good examples?

And an open question for the software crowd: What type of software and/or coding language is required for making "natural-to-the-web" tools? I'm starting to get to the point where I can see how that might be helpful. Even though I'm just old enough that my last coding class was in Fortran, I have enough of a natural affinity for thinking in code terms that it might be fun to play around with something a bit more... modern. :lol:

BRUTE
Posts: 3797
Joined: Sat Dec 26, 2015 5:20 pm

Re: Portfolio Charts

Post by BRUTE »

for recreating the current functionality, JavaScript. brute recommends React and D3 (a graph library). if Tyler9000 wants to add persistence for user accounts, saving portfolios, etc. he'll need a backend in a language like Ruby, Python, PHP, Java..

bryan
Posts: 1061
Joined: Sat Nov 29, 2014 2:01 am
Location: mostly Bay Area

Re: Portfolio Charts

Post by bryan »

Heh, yeah I'm a big fan of your site. If I was a webdev guy I would have already offered to join you in Austin (or for you to move to CA) and to work on your site together. Probably throwing together a proof of concept before making the offer. PF calcs and the like are quite interesting to me and I've hated mint.com for at least a couple years and thought it could be supplanted. The robos are pretty good but don't offer enough customization of plans or side tools.

Even for all the tools out there, none are satisfactory for allowing even just custom/saved spending plans over the next Y years (and offering the average spending curve or common expenditures you may not be considering, etc). Personal Capital comes the closest in their Retirement Planner calc (arbitrary spend X/year for Y years). I guess we all just work off averages or just take care of the big stuff (housing, healthcare, raising kids).

I don't think there is anything out there for "how to buy in." The best it gets is futureadvisor which tells you which trades to make across accounts to rebalance your portfolio (they will apparently make the trades for you if you subscribe... alt-universe MBA in me wants to determine if it would be more profitable for them to have a "buy-it-now" (make the trades for $X) button instead). The other investment robos require control of assets (and are intentionally oblivious about what assets you own outside of their control) to rebalance for you. All of the robos disallow custom target portfolios (I think they either target something like Swensen Portfolio (with some tilt?) for aggressive investors or something more basic for the conservative investors). Want a PP/GB and robo convenience/advice? Too bad, it's off to excel you go.

The "buy picker" as I described which suggests the security/asset to buy based on X, Y, or Z and some scheduling function doesn't exist anywhere afaik. In fact, haven't even seen any blog post/article on this single topic (many topics relate to it, like rising equity glidepath, lump sum v. DCA, CAPE, momentum investing, etc). Your site's approach to the human element (behavioural risk) also means you could incorporate things into the equation like needing some amount of cash in certain future windows of time.

There could be complimentary tools/products/features for the retired crowd like a "sell picker," similar to the above but optimized for a retiree's situation. I haven't explored what the robos offer for this (obviously at least you would consider rising equity glidepath and tax effect, consider TLH/TGH circumstance).

And of course the one-true-FIRE-calc/tool which has all of this stuff check-boxed in and pulled from other parts of the profile/site. Or simulate everything independently and give user some suggestions a la Mint and FutureAdvisor (want to shave 1 year off work/want to have 2% higher chance of not going broke before you die? consider doing a rising equity glidepath).

I really like how your site is filling in a gap of portfolio building. Seems natural to extend it to offer more portfolio comparison options.

Unfortunately, I haven't really had a calling to become a webdev, yet, even with so much potential for the skills to be useful. For now I am content enough with embedded systems.

From what I know, ES6/javascript, typescript, node.js, bootstrap, jquery, are some keywords for webdev. Elm and React maybe. For back-end the keyword languages would probably be python, go, and node.js but will probably end up with some php/phpQuery too. Elixir maybe.

JamesR
Posts: 947
Joined: Sun Apr 21, 2013 9:08 pm

Re: Portfolio Charts

Post by JamesR »

I'm actually quite interested in doing a non-excel-based version of portfoliocharts. I am a ruby web developer (I have some interest in elixir & elm/react too as well)

But before the website happens I'd probably just focus on implementing the core logic as a standalone ruby application, and then hook that up to the web side of things. The application would be nice to have as a sanity check against the excel version - are the results still the same?

Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: Portfolio Charts

Post by Tyler9000 »

Cool. If/when I get serious about rebuilding the tools from scratch I may just take you up on that offer. For now it's still in "hobby" mode and I sorta enjoy the learning process and building/tweaking myself.

henrik
Posts: 757
Joined: Fri Apr 13, 2012 5:58 pm
Location: EE

Re: Portfolio Charts

Post by henrik »


Post Reply