Author Topic: Are tools like FIRECalc useable also if investing not only in the US market?  (Read 3124 times)

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
So, do I need to do any major adaptions for tools like FIRECalc if I do not only invest in the US market ? This is what I not fully understand...I've only found a historical comparison of the DowJones(+230% last 20 years) and DAX(~330% last 20 years), but of course that differs much depending on the timespan and DowJones exists much longer....

My portfolio(bonds/stocks [index fonds] is roughly 20% US, 45 % World (here ~50% is also again U.S, else Europe(most), then Asia ), 5 % Asia and 30 % Europe(mainly Germany).
Since my home currency is Euro, I'm also not sure how to consider exchange rate fluctuations in the forecast.

Thanks for any comments here...

PapaBear

  • 5 O'Clock Shadow
  • *
  • Posts: 82
If you want to use FIRE calc, there is the option to use a random portfolio with a given return and given volatility. It is the last option on the portfolio tab:
Quote
A portfolio with random performance, with a mean total portfolio return of  xx% and variability (standard deviation) of xx %. Assume an inflation rate of  xx %.

Getting the right values is a bit tricky:
You can go to portfoliovisualizer.com and replicate your porfolio in one of the backtesting tool there to identify the CAGR and standard deviation (Stdev) of your portfolio:
Either https://www.portfoliovisualizer.com/backtest-asset-class-allocation for a generic comparison, or https://www.portfoliovisualizer.com/backtest-portfolio for a specific comparison with your funds. For the specific comparison, you will need to search for US ticker symbols of equivalent funds - the ETF selector on http://etfdb.com/screener/ or the US ishares website is your best bet to find these quickly. The average inflation rate in your country should be available at Eurostat or the statistics office of your country.

Please note that on portfoliovisualizer.com the analyzed timeframe is limited by the available data for the index - thus it makes sense to pick the ticker symbols of the oldest ETF available for the given index. The generic comparison will have a longer timeline, but even this timeline is not as long as the US only timelines embedded in FIREcalc.

EDIT: Please make sure to edit the expense ratio as well - FIREcalc has a default of 0.18% but quite a few European ETFs have higher expense ratios. You can use the weighted TER of the ETFs in your portfolio.
« Last Edit: July 11, 2017, 06:24:34 PM by PapaBear »

Bicycle_B

  • Handlebar Stache
  • *****
  • Posts: 1825
  • Mustachian-ish in Live Music Capital of the World
Portfoliocharts.com allows you to model years to retirement based on various weightings between different regions, based on historical data.  I think this does not include fees; past returns no guarantee of future results, obviously.  Germany is one of the five locations for which data has been provided.

https://portfoliocharts.com/
https://portfoliocharts.com/calculators/
https://portfoliocharts.com/portfolio/financial-independence/

Non-expert thoughts on choosing national vs international investments (on these, I defer to wiser commenters to follow):
1. It probably makes sense to have some overweighting in your home currency, since you spend there, but not too much; and again, some overweighting in your home country, because your housing expenses are there, but not too much.  Goal would be to adjust to rising prices if your country or currency suffers them, but also to support you if local economy softens. 
2. One baseline would be to compare with the % of the global stock market.  So if euro zone stocks are 20% of world's stock market capitalization, perhaps have 40%, similar to what you have.  Likewise if Deutsche Boerse is 2.5% of world, then perhaps 10% in Deutsche Boerse.
3. Over time, maintaining a fixed balance between regions through annual rebalancing should provide some advantage.

http://www.visualcapitalist.com/all-of-the-worlds-stock-exchanges-by-size/
« Last Edit: July 12, 2017, 12:03:33 PM by Bicycle_B »

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
Thx for your inputs, portfoliocharts is very helpful!

Do you think I should specially adapt the success rate concerning the exchange rate (especially Dollar->Euro) ? There is historical data available  here: http://www.boerse.de/performance/Euro-Dollar/EU0009652759

I could increase the standard derivation but then it gets very 'random'...do y see a better approach to simulate currency issues ?

Bicycle_B

  • Handlebar Stache
  • *****
  • Posts: 1825
  • Mustachian-ish in Live Music Capital of the World
Very good question.  I have no idea!!!

More accurately, I have ideas but insufficient expertise to advise anyone.  Again, I share thoughts while waiting for some wiser commenter to intervene.

Historical variance is good for proving things that can happen.  Maybe not so good for defining the boundaries of how far variance can go.  (Consider the book Black Swan by Nassim Taleb.)  Maybe an additional strategy to prepare would be consider scenarios where a major event occurs, and seek asset allocations that would account for them as well as normal times.  For example, if the euro collapses, that would be outside of historical experience in the PortfolioCharts data, but presumably USA portion of your allocation would be helpful at such a time.  Likewise a rebalancing strategy would allow you to harvest value if you could execute it calmly. 

One advantage of PortfolioCharts is that it provides data showing how different types of investments balance out a portfolio allocation, not just balancing stocks from different regions.  For example, I found that according to the data, it appears that a surprisingly large amount of cash can be included with little or no harm to return as long as it is counterbalanced by a similar amount of commodities.  I don't mean to recommend this allocation specifically, but rather am mentioning it as an example of different asset classes where maintaining a rebalanced allocation across asset classes could provide relatively stable returns in a wide variety of scenarios. 

Fwiw, it appears in the PortfolioCharts data that the US environment is one of the few where stocks have consistently outperformed other asset classes.  In more global environments, historically an all-stock allocation does not appear to be consistently best.  But those data are only a few decades long.  History has more examples but they are tough to quantify.  This leads me to seek a somewhat diversified portfolio - perhaps the most diversified portfolio that does not calculate a reduced return, on the principle that diversification can address unexpected events in addition to expected ones - but I am learning too.
 
Please research carefully rather than just take my word for it!
« Last Edit: July 12, 2017, 01:56:33 PM by Bicycle_B »

Tyler

  • Handlebar Stache
  • *****
  • Posts: 1141
Thx for your inputs, portfoliocharts is very helpful!

Do you think I should specially adapt the success rate concerning the exchange rate (especially Dollar->Euro) ? There is historical data available  here: http://www.boerse.de/performance/Euro-Dollar/EU0009652759

I could increase the standard derivation but then it gets very 'random'...do y see a better approach to simulate currency issues ?

If you're using PortfolioCharts, the returns are already converted to accurately model performance in the country you select.  So if you choose Germany, all numbers are in German currency (Euro) and account for German inflation.  Here's some more info on how it works:  https://portfoliocharts.com/2017/04/29/portfolio-charts-is-going-global/
« Last Edit: July 12, 2017, 04:26:22 PM by Tyler »

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
@Tyler: Thx, also very helpful. Is my understanding correct that https://portfoliocharts.com/portfolio/long-term-returns/ is already inflation-adjusted, that is in FIREcalc I set the inflation rate to 0.0 (portfolio section) ?

PapaBear

  • 5 O'Clock Shadow
  • *
  • Posts: 82
Portfoliocharts shows the real CAGR, that is including inflation.
If you use the "consistent annual market growth" option in FIREcalc, you should leave the inflation assumption at 0.0%.

// However, please note that a scenario with consistent annual market growth is VERY unlikely, as it completely excludes sequence of returns risk. In my opinion, the "random performance" option is somewhat closer to reality.

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
Yup, I've done that...do you think 10% standard-derivation is reasonable?

dougules

  • Handlebar Stache
  • *****
  • Posts: 1598
  • Location: AL
I think a related question would be if the market is efficient within the US, is it also efficient in moving assets between countries.

Tyler

  • Handlebar Stache
  • *****
  • Posts: 1141
@Tyler: Thx, also very helpful. Is my understanding correct that https://portfoliocharts.com/portfolio/long-term-returns/ is already inflation-adjusted, that is in FIREcalc I set the inflation rate to 0.0 (portfolio section) ?

Yes, all numbers on Portfolio Charts are inflation-adjusted using the actual historical CPI changes (not averages). 

If you're plugging your own custom numbers into FIREcalc, I would use the "random performance" option.  Use the Annual Returns calculator to find the Average and Standard Deviation for your portfolio in your country and set the inflation to zero (because the PC number already accounts for it).  Or you could just use my Withdrawal Rates and Retirement Spending calculators and let them do all the work.  ;)  The results will be more historically accurate that way.
« Last Edit: July 14, 2017, 02:48:58 PM by Tyler »

Bicycle_B

  • Handlebar Stache
  • *****
  • Posts: 1825
  • Mustachian-ish in Live Music Capital of the World
@Hawkeye_de, not sure if you're aware, but Tyler is pretty familiar with portfoliocharts.com... he's the guy who created it! 

Just thought it'd be fun to know while you sort all this out.

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
Yeah, in the meantime I've realized it ;)

@Tyler: your stuff is very helpful and the figures are similar to other (German) sources (+- 0.5%).

The only thing left for me is the pretty high standard derivation of 14,5%, thus I get in fireCalc figures success rates between 78% and 100%.

Since I do not strive for a very high success rate (>90), since my plan is to be financial independent enough to do projects(which I like), which hopefully return some money, this is not super crucial to me...

Nevertheless: Wouldn't it make more sense that such tools like cFireCalc simulate the variability deterministically? I mean you could run the stuff now x times and get the average but this is probably statistically not correct, because x needs to be very high.
« Last Edit: July 15, 2017, 01:53:19 AM by hawkeye_de »

Tyler

  • Handlebar Stache
  • *****
  • Posts: 1141
Nevertheless: Wouldn't it make more sense that such tools like cFireCalc simulate the variability deterministically? I mean you could run the stuff now x times and get the average but this is probably statistically not correct, because x needs to be very high.

A model like the PortfolioCharts withdrawal rates tool or the default FIREcalc settings is deterministic (you get the same results every time) while Monte Carlo simulations like the "random performance" option in FIREcalc (where the only inputs are an average and a standard deviation) are stochastic (the results change based on random variations in the inputs). PortfolioCharts outputs may resemble Monte Carlo outputs in that they show many different outcomes, but they're simply displaying the historically accurate uncertainty based on varying start years rather than generating hypothetical returns from a random number generator.

I've never really cared for Monte Carlo simulations because the underlying normally distributed statistical model does not describe the real world very well - you can run it a million times and it still may not capture known historical results in some situations. The details would bore most people, but "The Misbehavior of Markets" by Benoit Mandelbrot explains this very well even in terms a layman can understand. Long story short, it's garbage in, garbage out.

All that said, I still like that FIREcalc feature and have used it plenty myself. :)  It's always helpful to study problems from different perspectives. Just be sure to understand the assumptions and limits of each method so that you can put the results into proper context.
« Last Edit: July 15, 2017, 09:39:15 AM by Tyler »

hawkeye_de

  • 5 O'Clock Shadow
  • *
  • Posts: 15
Yup, makes sense.

I guess an improvement would be a 'customized' Monte Carlo simulation, that is it is run by constraints - for instance the probability of several (big) market crashes in one decade is more or less 0 - so such a model could eliminate such results and then it could also give the Min/Max boundaries...but I guess that's what your tools are also display.

I need FIRECalc to add my expected private and government pensions...


Tyler

  • Handlebar Stache
  • *****
  • Posts: 1141

I need FIRECalc to add my expected private and government pensions...


Ah -- makes sense.  Thanks for mentioning how you use it, as that's how I get ideas for future improvements.  :)