Author Topic: Anybody else get results that look wildly optimistic when using FireCalc?  (Read 5387 times)

Jrr85

  • Handlebar Stache
  • *****
  • Posts: 1200
When I do my own projections, and pick a 5 or 5.5% real return, I end up just barely being able to make retirement by retirement age, and will probably get to leave a good inheritance (say ~$1.5-$2M each to my kids). 

When I use the Total market portfolio with a 90% stock allocation, I get results ranging from $1.2M to $27M.  So $400k per kid nominal (~$80k real assuming 3% inflation over 54 years) to $9M per kid nominal ($1.8M real).  So the best case scenario is basically what I project, which I don't get.  How is the most optimistic result from historical returns not better than just picking a 5 or 5.5% real return? 

But then when I use the mixed portfolio of 10% microcap, 15% US Small, 10% US Small Value, 30% S&P 500, 25% large cap value, and 10% 1 month treasury, I get results that range from $30M to $92M.  (So roughly $10M per kid in worst case scenario, which is right at $2M current dollars assuming 3% inflation over 54 years).  So the worst case scenario is that I do what I project, and the best case scenario is that it's 3 times better than what I project.  Which that doesn't seem right.  Are they using unrepeatable returns for the value portions of the portfolio?  Or did small and micro cap outperform the S&P and total market by that much historically? 

Just curious whether Firecalc is reliable, or if it's only reliable for certain settings, what those settings should be.
« Last Edit: May 23, 2018, 10:00:12 AM by Jrr85 »

mathlete

  • Handlebar Stache
  • *****
  • Posts: 2070
I would look for a breakdown in the inputs somewhere along the line.

Short of that, maybe their small/microcap data had some survivorship bias in it?

jlcnuke

  • Pencil Stache
  • ****
  • Posts: 931
Well, a 5.5% real return is quite below historic norms, so you're comparing apples and oranges to start with...

Your well below norms assumptions happen to be similar to the worst historical values for your separate choices. That's about the only similarity in the comparisons.

Sent from my XT1635-01 using Tapatalk
« Last Edit: May 23, 2018, 07:15:47 PM by jlcnuke »

Jrr85

  • Handlebar Stache
  • *****
  • Posts: 1200
Well, a 5.5% real return is quite below historic norms, so you're comparing apples and oranges to start with...

Your well below norms assumptions happen to be similar to the worst historical values for your separate choices. That's about the only similarity in the comparisons.

Sent from my XT1635-01 using Tapatalk

I guess I wasn't clear bc of the title, but the 5.5% was as good as basically the best results using he total market portfolio, but then only as good as the worst results using the allocation stated. I would have thought 5.5% would have been somewhere in the midrange of both scenarios.

SwordGuy

  • Walrus Stache
  • *******
  • Posts: 8956
  • Location: Fayetteville, NC
Well, a 5.5% real return is quite below historic norms, so you're comparing apples and oranges to start with...

Your well below norms assumptions happen to be similar to the worst historical values for your separate choices. That's about the only similarity in the comparisons.

Sent from my XT1635-01 using Tapatalk

I guess I wasn't clear bc of the title, but the 5.5% was as good as basically the best results using he total market portfolio, but then only as good as the worst results using the allocation stated. I would have thought 5.5% would have been somewhere in the midrange of both scenarios.

My understanding is that historical average returns for the stock market are a bit over 10%, with real returns (i.e., subtracting out inflation) being 7%.




Bateaux

  • Handlebar Stache
  • *****
  • Posts: 2318
  • Location: Port Vincent
I really don't trust it.  Past performance is not a realistic predicter of the future. 

Mr Mark

  • Handlebar Stache
  • *****
  • Posts: 1229
  • Location: Planet Earth
  • Achieved Financial Independence summer 2014. RE'18
You need to be mindfull of the difference between averaging annual returns and the average annual rate of return.

The difference is due to compounding and volatility. Here's a simplified & extreme example:
I invest $100 over 4 years. The annual returns are:
Year 1: +50%
Year 2: -50%
Year3: +50%
Year 4: -20%

The question is: What is the average rate of return?


If I take a simple average of the annual returns, average return = +7.5% (50-50+50-20=30%/4= 7.5%)
So you'd perhaps expect after the 4 years the investment would be worth 100*(1.075^4) = $133.50

But over the 4 years my initial $100 investment is only worth $90!
Year 0 = $100
Year 1 = $150
Year 2 = $75
Year 3 = $112.5
Year 4 = $90

So the compounded average return was (90-100/100)/4= -2.5%! Oops.

So when taking the "average" of the annual returns I should take a geometric average, not arthimetic. For the example, the geometric average is -2.6%, much more accurately capturing the compounding effect.



marty998

  • Walrus Stache
  • *******
  • Posts: 7372
  • Location: Sydney, Oz
I'm 100% certain I'll be the one to get that $92M.

:D

Sand101

  • 5 O'Clock Shadow
  • *
  • Posts: 20
Anybody else get results that look wildly optimistic when using FireCalc?

I've waded through quite a number of these calculators and have gone so far as to write my own to make sure that what I got was believable. 

FC doesn't take taxes into account, which it notes, but is easy to forget.  If you account for that it's reasonable in it's answers.  I still tend to find it a bit optimistic in comparison to others, but that doesn't mean it's broken. 

pecunia

  • Magnum Stache
  • ******
  • Posts: 2840
Quote
I still tend to find it a bit optimistic in comparison to others, but that doesn't mean it's broken. 

What are some of the others out there that are less optimistic.  I figure I'm very close to having an adequate stash, but I like good guarantees.

sol

  • Walrus Stache
  • *******
  • Posts: 8433
  • Age: 47
  • Location: Pacific Northwest
What are some of the others out there that are less optimistic.  I figure I'm very close to having an adequate stash, but I like good guarantees.

If you're relying on website calculators to plan your retirement, you're probably not ready yet.

MDM

  • Senior Mustachian
  • ********
  • Posts: 11477
What are some of the others out there that are less optimistic.  I figure I'm very close to having an adequate stash, but I like good guarantees.
Best and/or Recommended Retirement Calculator - Bogleheads.org and links therein may be useful.

Cromacster

  • Handlebar Stache
  • *****
  • Posts: 1695
  • Location: Minnesnowta
Quote
I still tend to find it a bit optimistic in comparison to others, but that doesn't mean it's broken. 

What are some of the others out there that are less optimistic.  I figure I'm very close to having an adequate stash, but I like good guarantees.

You mean those guarantees that don’t even remotely exist in any way shape or form??


pecunia

  • Magnum Stache
  • ******
  • Posts: 2840
Sol:

Quote
If you're relying on website calculators to plan your retirement, you're probably not ready yet

You are probably right.  Fidelity guy told me I could, but my life would be better with a bigger stash, will have medical insurance and I can continue to learn about investing money before taking the big plunge.  This website does have a lot of good information and provides direction as well.

sol

  • Walrus Stache
  • *******
  • Posts: 8433
  • Age: 47
  • Location: Pacific Northwest
Sol:

Quote
If you're relying on website calculators to plan your retirement, you're probably not ready yet

You are probably right.  Fidelity guy told me I could, but my life would be better with a bigger stash, will have medical insurance and I can continue to learn about investing money before taking the big plunge.  This website does have a lot of good information and provides direction as well.

Sure, but I didn't mean that you don't sound financially ready, I meant that you don't sound mentally ready.  I think each person should have a pretty clear mental roadmap for how their future retirement is going to unfold, with options and decision points pre-planned.  It has provided a huge amount of piece of mind for me.

Step one was getting a VERY firm handle on our realistic expenses, including future healthcare and college costs, unplanned events like roofs and car replacement, and the occasional family emergency.  Step two was estimating our total future income from each source (roth ira pipeline, taxable accounts, rents or RE sales, pensions, social security).  Step three was building a spreadsheet that showed me where each month of income for the next 60 years was going to come from, based on tax treatment and availability, for a variety of future scenarios and market returns. 

After I had that done, and played with the spreadsheet enough, I convinced myself that I couldn't realistically see any failure scenarios anymore even if we did something stupid, and I believe that we are smart enough to avoid the stupidest future paths.  The most stupid one appears to be inflating your living expenses by more than about 5% per year if the stock market returns 0% or lower, without seeking out additional income. 

My situation is somewhat unique because we have pensions starting at age 62, which would cover all of our (kidless and mortgage free) expenses at that age when combined with social security.  So our stash really only needs to last ~20 years, which allows us to SWRs significantly higher than 4%.

Poeirenta

  • Bristles
  • ***
  • Posts: 345
@sol, would you be willing to share how you did all that (or maybe you have and I missed it)? I feel like that detailed scenario testing is what we should do next, but I'm at a loss for how to do it. I struggle with a numbers and math learning disability*, so templates are a huge help. We're about 5 years out from FIRE if firecalc and i-orp aren't lying to me. We also have pensions and a paid off house.

*it's ironic that I'm the FIRE manager in my household, right? I always doubt my numbers...

Seadog

  • Bristles
  • ***
  • Posts: 268
  • Age: 40
  • Location: Halifax, NS
You need to be mindfull of the difference between averaging annual returns and the average annual rate of return.

The difference is due to compounding and volatility. Here's a simplified & extreme example:
I invest $100 over 4 years. The annual returns are:
Year 1: +50%
Year 2: -50%
Year3: +50%
Year 4: -20%

The question is: What is the average rate of return?


If I take a simple average of the annual returns, average return = +7.5% (50-50+50-20=30%/4= 7.5%)
So you'd perhaps expect after the 4 years the investment would be worth 100*(1.075^4) = $133.50

But over the 4 years my initial $100 investment is only worth $90!
Year 0 = $100
Year 1 = $150
Year 2 = $75
Year 3 = $112.5
Year 4 = $90

So the compounded average return was (90-100/100)/4= -2.5%! Oops.

So when taking the "average" of the annual returns I should take a geometric average, not arthimetic. For the example, the geometric average is -2.6%, much more accurately capturing the compounding effect.

While I agree, I'm curious why this is even a thing. The "average arithmetic return" is a wholly meaningless number. Like income multiplied by shoe size.

sol

  • Walrus Stache
  • *******
  • Posts: 8433
  • Age: 47
  • Location: Pacific Northwest
@sol, would you be willing to share how you did all that

There is no template, sorry.  But I can describe it to you.

In excel, I made a column of every sequential month between retirement and death (in my case planning on age 100).
In the next columns over I made my age and my partner's age in that month.
In the next column over I put my checking account balance in that month.
In the next column over I put my estimated expenses in that month (adjusted up for inflation each month, and including my mortgage changes and then eventually empty nesting).

Then in the next columns I put the forecast balances in each of my retirement accounts, with an adjustable return rate so I can test what happens at 0% vs 10% market performance.  I keep separate columns for the estimated future value of my rental properties (90% of equity to account for sale costs) .

Each month, my checking account balance goes down by my expenses from the previous month, and up by the amount of any regular income I have (rent surplus, solar panel payments, etc).  When the checking account gets too low, I take money out of one of the investment or retirement accounts by depleting the investment account and adding that amount (minus taxes if applicable) to the checking account column.  First I draw from my taxable account, because it is restriction free.  Then I draw from my Roth IRA principal contributions, since those are also restriction free, then Roth IRA interest at age 59.5.  At age 60 my wife has a pension, which I add to my checking account balance each month.  My pension starts at age 62.  Then social security.  Our expenses go down a little when the kids move out, and a lot when our mortgage gets paid off.  If I have to draw from my 401k plan early, I play on paying the 10% penalty in addition to regular income taxes on that amount.

I graph our net worth over time from all sources under various scenarios, like 0% stock market returns and 5% inflation vs 8% returns and 2% inflation.  I can mathematically "sell" a house if necessary, removing it's monthly rent but pocketing the big check from the sale, and the future house value is adjustable (currently assuming 1.5% per year appreciation).

It's not a stochastic model, so it ignores market volatility.  If the market goes down 25% in one year, I'll have to manually update the spreadsheet's account balances but presumably it won't continue to go down 25% every year.

Then I've added some additional complications, like making annual Roth IRA pipeline rollovers and how to minimize our paper income for ACA subsidies or FAFSA applications during the relevant year by only withdrawing from taxable accounts during that window.

Basically, the whole thing is designed to give me a roadmap.  There are notes in the far right column of important dates, like SS eligibility or Roth IRA rollover maturation dates.  I should be able to just follow along over the years, updating balances as necessary, and always have a pretty clear view of what our financial future should look like.  If the future gets ugly, the spreadsheet should make it clear that it's time to adjust the plan or even (gasp!) go find more income somewhere.

Holocene

  • Bristles
  • ***
  • Posts: 369
When I use the Total market portfolio with a 90% stock allocation, I get results ranging from $1.2M to $27M.  So $400k per kid nominal (~$80k real assuming 3% inflation over 54 years) to $9M per kid nominal ($1.8M real).  So the best case scenario is basically what I project, which I don't get.  How is the most optimistic result from historical returns not better than just picking a 5 or 5.5% real return? 

By default, FIRECalc already incorporates inflation. On the Spending Models tab, it allows you to chose PPI, CPI, or a fixed percentage for inflation.  It defaults to CPI which is what I always leave it to.  The FireCalc results also say, "values are in terms of the dollars as of the beginning of the retirement period for each cycle."  So I'm pretty sure you're looking at $400k per kid real worst case and $9M per kid real best case.  You shouldn't adjust these numbers for inflation since FIRECalc already did that.

Jrr85

  • Handlebar Stache
  • *****
  • Posts: 1200
When I use the Total market portfolio with a 90% stock allocation, I get results ranging from $1.2M to $27M.  So $400k per kid nominal (~$80k real assuming 3% inflation over 54 years) to $9M per kid nominal ($1.8M real).  So the best case scenario is basically what I project, which I don't get.  How is the most optimistic result from historical returns not better than just picking a 5 or 5.5% real return? 

By default, FIRECalc already incorporates inflation. On the Spending Models tab, it allows you to chose PPI, CPI, or a fixed percentage for inflation.  It defaults to CPI which is what I always leave it to.  The FireCalc results also say, "values are in terms of the dollars as of the beginning of the retirement period for each cycle."  So I'm pretty sure you're looking at $400k per kid real worst case and $9M per kid real best case.  You shouldn't adjust these numbers for inflation since FIRECalc already did that.

That makes sense but then makes the results for the mixed portfolio even more out of line to the high side.  Unless one is corrected for inflation and one is not (which would be a weird way to do it), it seems like one or the other must be wrong.   

jlcnuke

  • Pencil Stache
  • ****
  • Posts: 931
When I use the Total market portfolio with a 90% stock allocation, I get results ranging from $1.2M to $27M.  So $400k per kid nominal (~$80k real assuming 3% inflation over 54 years) to $9M per kid nominal ($1.8M real).  So the best case scenario is basically what I project, which I don't get.  How is the most optimistic result from historical returns not better than just picking a 5 or 5.5% real return? 

By default, FIRECalc already incorporates inflation. On the Spending Models tab, it allows you to chose PPI, CPI, or a fixed percentage for inflation.  It defaults to CPI which is what I always leave it to.  The FireCalc results also say, "values are in terms of the dollars as of the beginning of the retirement period for each cycle."  So I'm pretty sure you're looking at $400k per kid real worst case and $9M per kid real best case.  You shouldn't adjust these numbers for inflation since FIRECalc already did that.

That makes sense but then makes the results for the mixed portfolio even more out of line to the high side.  Unless one is corrected for inflation and one is not (which would be a weird way to do it), it seems like one or the other must be wrong.   

As stated earlier, returns aren't "static", and "averaging out" doesn't actually work. That's why we have tools like FIRECalc that take your inputs and see how they would have done historically. Not once, but in every scenario they can with the data available. What you're calling "wildly optimistic" is what happened to people retiring into the Great Depression, or the late 70's crap market etc etc. Those times etc are your "low" numbers showing you would have been just fine with your equivalent stache back then.  If you personally believe that the Great Depression is likely to look like a cake-walk compared to the near-term future, then by-all-means assume these calculators are optimistic. If you think we're not likely to experience the worst returns in the history of this country in the near-term-future, then you're probably being overly pessimistic because you don't fully understand sequence of returns and how that concept impacts inflation adjusted portfolio return.

Jrr85

  • Handlebar Stache
  • *****
  • Posts: 1200
When I use the Total market portfolio with a 90% stock allocation, I get results ranging from $1.2M to $27M.  So $400k per kid nominal (~$80k real assuming 3% inflation over 54 years) to $9M per kid nominal ($1.8M real).  So the best case scenario is basically what I project, which I don't get.  How is the most optimistic result from historical returns not better than just picking a 5 or 5.5% real return? 

By default, FIRECalc already incorporates inflation. On the Spending Models tab, it allows you to chose PPI, CPI, or a fixed percentage for inflation.  It defaults to CPI which is what I always leave it to.  The FireCalc results also say, "values are in terms of the dollars as of the beginning of the retirement period for each cycle."  So I'm pretty sure you're looking at $400k per kid real worst case and $9M per kid real best case.  You shouldn't adjust these numbers for inflation since FIRECalc already did that.

That makes sense but then makes the results for the mixed portfolio even more out of line to the high side.  Unless one is corrected for inflation and one is not (which would be a weird way to do it), it seems like one or the other must be wrong.   

As stated earlier, returns aren't "static", and "averaging out" doesn't actually work. That's why we have tools like FIRECalc that take your inputs and see how they would have done historically. Not once, but in every scenario they can with the data available. What you're calling "wildly optimistic" is what happened to people retiring into the Great Depression, or the late 70's crap market etc etc. Those times etc are your "low" numbers showing you would have been just fine with your equivalent stache back then.  If you personally believe that the Great Depression is likely to look like a cake-walk compared to the near-term future, then by-all-means assume these calculators are optimistic. If you think we're not likely to experience the worst returns in the history of this country in the near-term-future, then you're probably being overly pessimistic because you don't fully understand sequence of returns and how that concept impacts inflation adjusted portfolio return.

You're not understanding.  I'm not questioning the results because they are particularly high or low.  I'm questioning them because I would not expect going with a mixed portfolio as described, compared to going with a total market portfolio as described, would explain the discrepancy in results (unless one is inflation adjusted and one is not).  It doesn't seem to pass the smell test that the best results with the total market portfolio would do roughly as well as the worst results using the mixed portfolio as described.  I would assume there is too much correlation between the total stock market and small caps, and that even if weighting towards small caps does on average provide better results, the outperformance would be so drastic because there would be some periods where small caps underperformed.

Maybe I don't appreciate how much small caps have outperformed in the past, but I would be surprised if a small cap weighted portfolio would outperform a total market portfolio over essentially every historical scenario. 

Telecaster

  • Magnum Stache
  • ******
  • Posts: 3551
  • Location: Seattle, WA

You're not understanding.  I'm not questioning the results because they are particularly high or low.  I'm questioning them because I would not expect going with a mixed portfolio as described, compared to going with a total market portfolio as described, would explain the discrepancy in results (unless one is inflation adjusted and one is not).  It doesn't seem to pass the smell test that the best results with the total market portfolio would do roughly as well as the worst results using the mixed portfolio as described.  I would assume there is too much correlation between the total stock market and small caps, and that even if weighting towards small caps does on average provide better results, the outperformance would be so drastic because there would be some periods where small caps underperformed.

Maybe I don't appreciate how much small caps have outperformed in the past, but I would be surprised if a small cap weighted portfolio would outperform a total market portfolio over essentially every historical scenario.

I was surprised at the disparity as well, but not entirely surprised.  Here's why:  With Total Stock Market you are betting on the whole market, but you are not betting equally on the whole market. While total stock market includes small caps, because the index is cap weighted the small caps almost don't matter.   You're betting mostly on large caps and small caps grow faster than large caps (but with more volatility), so for long term growth your money is really on the wrong horse.   Because total stock market and S&P 500 are both cap weighted they should track very closely, and they do.  The backtest here is a little short, but by stiring in some small and mid-caps, you can substantially improve returns:

https://www.portfoliovisualizer.com/backtest-asset-class-allocation?s=y&mode=2&startYear=1972&endYear=2018&initialAmount=100000&annualOperation=3&annualAdjustment=0&inflationAdjusted=true&annualPercentage=4.0&frequency=4&rebalanceType=1&benchmark=VFINX&portfolio1=Custom&portfolio2=Custom&portfolio3=Custom&TotalStockMarket1=50&TotalStockMarket2=60&TotalStockMarket3=60&MidCapBlend1=20&MidCapBlend2=40&MidCapBlend3=20&MidCapGrowth3=20&SmallCapBlend1=20&TreasuryNotes1=10

That's the flaw of "buy VTSAX and forget it."  There are worse strategies to be sure, but you're not really betting on the whole market.