Author Topic: cFiresim SEVERELY overestimates success rates for Mustachians  (Read 61719 times)

VoteCthulu

  • Bristles
  • ***
  • Posts: 409
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #100 on: July 19, 2017, 01:36:06 PM »
I think there's a lot of value warning people that if the market crashes or languishes in the 5 years after retirement, that you need to start implenting your backup plan(s).

That said,  I think the far larger threat to a portfolio is unexpected expenses. This could be medical, taxes, lawsuits, or anything that cant be completely predicted or insured against. If anyone knows of a FIRE blogger that's had their 4% SWR portfolio survive a large disruption like that, I'd be very interested in seeing how they managed.

Tyson

  • Magnum Stache
  • ******
  • Posts: 2512
  • Age: 49
  • Location: Denver, Colorado
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #101 on: July 19, 2017, 01:39:45 PM »
I think there's a lot of value warning people that if the market crashes or languishes in the 5 years after retirement, that you need to start implenting your backup plan(s).

That said,  I think the far larger threat to a portfolio is unexpected expenses. This could be medical, taxes, lawsuits, or anything that cant be completely predicted or insured against. If anyone knows of a FIRE blogger that's had their 4% SWR portfolio survive a large disruption like that, I'd be very interested in seeing how they managed.

Yes, I agree - I'm more worried about that as well.  What if my wife or child gets cancer?  Medical expenses are no joke.

DavidAnnArbor

  • Handlebar Stache
  • *****
  • Posts: 2279
  • Age: 55
  • Location: Ann Arbor, Michigan
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #102 on: July 19, 2017, 02:25:05 PM »
Your expenses could include a cushion of $10,000 year of unexpected things.
Some have talked about having a home equity line of credit to cover unexpected expenses.

VoteCthulu

  • Bristles
  • ***
  • Posts: 409
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #103 on: July 19, 2017, 08:35:25 PM »
Your expenses could include a cushion of $10,000 year of unexpected things.
Sure, but then it becomes the 3% (or 3.5% or 3.9% or whatever) rule.

Playing with Fire UK

  • Magnum Stache
  • ******
  • Posts: 3378
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #104 on: July 20, 2017, 01:10:24 AM »
Your expenses could include a cushion of $10,000 year of unexpected things.

You can also look at your budget and see what costs would naturally fall away with emergencies. For example, vacation spend would probably reduce with a health crisis or major home repair/upgrade.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #105 on: July 20, 2017, 01:58:01 AM »
I also worry that the very name of that thread "Stop Worrying about the 4% Rule" encourages people to accept it blindly.

I don't even accept gravity blindly, but I do accept it and encourage others to do the same.

It astounds me that anyone could read through the hundreds of pages of careful mathematical analysis this forum has done about safe withdrawal rates, and then accuse us of blind faith because of a thread title summarizing our findings.

Stop worrying about the gravitational constant.  Even if your assumptions turn out to be wrong, at least everyone else will fly off into space with you!

Leisured

  • Pencil Stache
  • ****
  • Posts: 697
  • Age: 76
  • Location: South east Australia, in country
  • Retired, and loving it.
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #106 on: July 20, 2017, 04:43:39 AM »

- Life Insurance.  My wife and I each have a 500,000 universal life policy on each other. 

Not to completely take over this thread, but why in the world do you have these policies, and what are they costing you?

With so many backup plans, and the fact that since you are on this site you are likely to FIRE at an early age, you are probably one with the least need for a whole life plan(with my overall opinion being that NOBODY needs whole life).

A FIRE couple needs NO life insurance because you are a liability in FIRE, not an asset. Your remaining stashe need only fund one person instead of the planned 2 people.

I would cash out a whole life plan and buy a term life plan until FIRE to get the surviving spouse to FIRE upon the others death.

I agreed with radram, who so far is the only one to comment on this point. I agree that term life insurance, as opposed to universal life insurance, is a good idea during the accumulation phase, but not thereafter.


Leisured

  • Pencil Stache
  • ****
  • Posts: 697
  • Age: 76
  • Location: South east Australia, in country
  • Retired, and loving it.
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #107 on: July 20, 2017, 04:58:07 AM »
An interesting and worthwhile topic.

I am Australian, and in my experience, large companies prefer to at least maintain dividends after a bad year. That is, if an increase in dividends is not justified, because of difficult market conditions, some companies try to maintain last year's dividend rate rather than reduce the dividend rate. Of course, if bad times persist for a few years, they will have to cut dividends. This means that dividend payments, in the short term, need not reflect a bad year for a retiree. There is more than stock prices.

Stock prices usually move upwards over time, so that one cannot easily detect a market peak before it happens, as many on this forum have pointed out. However, a suasion where dividend yields are unusually low, and price to earnings rates are unusually high, suggests that the market is over priced, so retiring in such market conditions would not be a good idea. Perhaps what is needed is an analysis of the success rates in preserving retirement assets, against the dividend yields in the year of retirement.


Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #108 on: July 20, 2017, 08:06:16 AM »
Stop worrying about the gravitational constant.  Even if your assumptions turn out to be wrong, at least everyone else will fly off into space with you!

Per usual Dragoncar's wisdom is well placed within smart-ass wit.  At a certain point no additional safety can help offset risk.  An S-curve situation occurs. The only real exclusion from this curve is the ultra-wealthy who have the ability to invest in assets to which your average millionaire does not have access.  Or at least enough wealth to afford some wild-card barbell type bets Taleb would encourage.

Why continually worry about outlying statistical situations that can neither be accurately predicted or mitigated against with a passive investment allocation?  It's better to invest in adaptability skills in those potential situations.  For the 4% rule to fail there has to be both a precipitous drop in asset value soon after draw-down begins AND long-term very low real returns. Even if this happens, someone with only half of FI assets remaining is better suited to take advantage of a universally bad situation, if they adapt to it.

spokey doke

  • Pencil Stache
  • ****
  • Posts: 514
  • Escaped from the ivory tower basement
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #109 on: July 20, 2017, 09:36:23 AM »

Why continually worry about outlying statistical situations that can neither be accurately predicted or mitigated against with a passive investment allocation?  It's better to invest in adaptability skills in those potential situations.  For the 4% rule to fail there has to be both a precipitous drop in asset value soon after draw-down begins AND long-term very low real returns. Even if this happens, someone with only half of FI assets remaining is better suited to take advantage of a universally bad situation, if they adapt to it.

I generally agree with this sentiment...("OMG!!! What if ____????!!! AHHHHH!!!!").

But you CAN address these fears...so if you have concerns, do some research and plug in numbers that would address them into cfiresim.  You can do that...add to your projected annual budget, and/or periodic additional expenses.  If that means working longer, well that is the price of being so conservative in your financial planning.

And then, if things are actually worse in a way that is unforeseen...well...life happens.  I actually think the prospects of getting some crazy disease that could eat up a bunch of money is a good reason to retire earlier and enjoy life while you can (same goes for some kind of financial apocalypse).  If these things come about, flexibility and badassity are two of your greatest assets in coping.

tooqk4u22

  • Magnum Stache
  • ******
  • Posts: 2530
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #110 on: July 20, 2017, 10:33:05 AM »

- Life Insurance.  My wife and I each have a 500,000 universal life policy on each other. 

Not to completely take over this thread, but why in the world do you have these policies, and what are they costing you?

With so many backup plans, and the fact that since you are on this site you are likely to FIRE at an early age, you are probably one with the least need for a whole life plan(with my overall opinion being that NOBODY needs whole life).

A FIRE couple needs NO life insurance because you are a liability in FIRE, not an asset. Your remaining stashe need only fund one person instead of the planned 2 people.

I would cash out a whole life plan and buy a term life plan until FIRE to get the surviving spouse to FIRE upon the others death.

I agreed with radram, who so far is the only one to comment on this point. I agree that term life insurance, as opposed to universal life insurance, is a good idea during the accumulation phase, but not thereafter.

In theory this is correct, in the real world it is far more situational.  I focus more on the finances and such than DW, and while if I passed she should be able to get by, my guess is that there will be an adjustment period and a certain lack of focus on things and more focus on grieving and getting my kids through it.  So for me it is important to have a chunk of cash to be able to fuck up with to get through the transition and into clearer thoughts and directions.  Also, what killed me may have left some large lingering bills so it might be good to be able to cover those without using the FIRE stash.

And no, two people going to one or in my scenario five people going to four does not translate to equivalent reduction in expenses, it might be less but not immediately.

tooqk4u22

  • Magnum Stache
  • ******
  • Posts: 2530
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #111 on: July 20, 2017, 10:37:03 AM »
I think there's a lot of value warning people that if the market crashes or languishes in the 5 years after retirement, that you need to start implenting your backup plan(s).

That said,  I think the far larger threat to a portfolio is unexpected expenses. This could be medical, taxes, lawsuits, or anything that cant be completely predicted or insured against. If anyone knows of a FIRE blogger that's had their 4% SWR portfolio survive a large disruption like that, I'd be very interested in seeing how they managed.

This is really what I am grappling with.  The 4% rule et al is pretty dynamic but relies on static expenses in real terms. My number includes factors for most large expenses (average amounts based on replacement for big items) but healthcare is doozy to figure out. 

Then there is the lifestyle/life-stage changes that happen and are mostly kid related but are not permanent and most likely <10 years.   

Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #112 on: July 20, 2017, 06:29:39 PM »

Why continually worry about outlying statistical situations that can neither be accurately predicted or mitigated against with a passive investment allocation?  It's better to invest in adaptability skills in those potential situations.  For the 4% rule to fail there has to be both a precipitous drop in asset value soon after draw-down begins AND long-term very low real returns. Even if this happens, someone with only half of FI assets remaining is better suited to take advantage of a universally bad situation, if they adapt to it.

I generally agree with this sentiment...("OMG!!! What if ____????!!! AHHHHH!!!!").

But you CAN address these fears...so if you have concerns, do some research and plug in numbers that would address them into cfiresim.  You can do that...add to your projected annual budget, and/or periodic additional expenses.  If that means working longer, well that is the price of being so conservative in your financial planning.

And then, if things are actually worse in a way that is unforeseen...well...life happens.  I actually think the prospects of getting some crazy disease that could eat up a bunch of money is a good reason to retire earlier and enjoy life while you can (same goes for some kind of financial apocalypse).  If these things come about, flexibility and badassity are two of your greatest assets in coping.

You can also better suit your allocation to macro and personal-micro (where are you in accumulation/draw-down) economic situations to some extent.  Tyler's site is great to back test ideas.  I would argue that is part of the adaptability component.

The general point being, there are significantly diminishing returns (by returns I mean anti-fragility or safety in FIRE) with dollars being saved after about 20-25X expenses. After that point, each year of expenses saved increases success rates (via historical back-testing) by very little in true S-curve fashion.  If 3.5%WR rates fails, its likely a 2.5% WR will fail too, because some horrible economic calamady has taken place which was outside the scope of backtesting.  That's a lot of extra years working for virtually no increase in FIRE safety margin. 

Better to save 25X and be prepared to adapt or just save 33X and refuse to adapt?
As far as return on time invested AND anti-fragility; the former is far superior to the latter.
« Last Edit: July 20, 2017, 06:32:53 PM by Classical_Liberal »

Shane

  • Handlebar Stache
  • *****
  • Posts: 1382
  • Location: PA
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #113 on: July 21, 2017, 01:19:49 AM »
I think there's a lot of value warning people that if the market crashes or languishes in the 5 years after retirement, that you need to start implenting your backup plan(s).

That said,  I think the far larger threat to a portfolio is unexpected expenses. This could be medical, taxes, lawsuits, or anything that cant be completely predicted or insured against. If anyone knows of a FIRE blogger that's had their 4% SWR portfolio survive a large disruption like that, I'd be very interested in seeing how they managed.

The fear that somehow unexpected medical expenses will derail early retirees' FIRE plans seems irrational to me.

At least whenever we're in the US, I plan on always having health insurance. If, for any reason, it becomes impossible or impractical to continue protecting our assets with medical insurance in the US, perhaps because of changes to the ACA, we will permanently leave the United States and move to where healthcare is cheap enough that it's possible to pay cash out of pocket. There are many countries around the world where early retirees can purchase good quality medial care at a tiny fraction of the cost in the US.

spokey doke

  • Pencil Stache
  • ****
  • Posts: 514
  • Escaped from the ivory tower basement
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #114 on: July 21, 2017, 09:11:00 AM »

The general point being, there are significantly diminishing returns (by returns I mean anti-fragility or safety in FIRE) with dollars being saved after about 20-25X expenses. After that point, each year of expenses saved increases success rates (via historical back-testing) by very little in true S-curve fashion.  If 3.5%WR rates fails, its likely a 2.5% WR will fail too, because some horrible economic calamady has taken place which was outside the scope of backtesting.  That's a lot of extra years working for virtually no increase in FIRE safety margin. 

Great point to consider...

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 28428
  • Age: -999
  • Location: Seattle, WA
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #115 on: July 22, 2017, 08:26:55 AM »
Title should be cfiresim severely overestimates for non-Mustachians.  For Mustachians, it barely overestimates, due to our high savings rates.
I am a former teacher who accumulated a bunch of real estate, retired at 29, spent some time traveling the world full time and am now settled with three kids.
If you want to know more about me, this Business Insider profile tells the story pretty well.
I (rarely) blog at AdventuringAlong.com. Check out the Now page to see what I'm up to currently.

dividendman

  • Handlebar Stache
  • *****
  • Posts: 1377
  • Age: 38
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #116 on: July 22, 2017, 10:17:48 AM »
Am I the only crazy one that thinks this way:

1) Save up enough with 4% rule that isn't completely bare-bones so there is some spending flexibility
2) FIRE as soon as you have that
3) Once FIRED allocate enough to bonds/CDs, etc that a 5-10 year shit in the market won't be the end of the world since you're drawing down mostly from bonds/CDs (i have 25% in BND and will go to 5% after 10 years, so most of the first 10 years of FIRE will come from bond liquidation and interest/dividends, unless the stock market goes on a big run, in which case it won't matter anyway)
4) Be happy that you're free and live without work! Woo hoo - this is essentially priceless (doing this in 4 weeks... wooooooooo)
5) Let's say the worst happens and you get fucked somehow and need to work again after 5 or 10 years... so what? This scenario is still better than not having taken time off to do the things you enjoy, IMO.

I don't see any better alternatives... maybe i'm cray cray.

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #117 on: July 22, 2017, 12:22:11 PM »
Title should be cfiresim severely overestimates for non-Mustachians.  For Mustachians, it barely overestimates, due to our high savings rates.

I don't know if non-Mustachians would think of using the 4% rule to determine their target number; it seems they'd just retire at 65 no matter what, without regard to market fluctuations.

This mostly affects low savings rate folks planning for an aggressive, high WR. So definitely not all Mustachians. But I could see someone with very low income reading this site being affected.


3) Once FIRED allocate enough to bonds/CDs, etc that a 5-10 year shit in the market won't be the end of the world since you're drawing down mostly from bonds/CDs (i have 25% in BND and will go to 5% after 10 years, so most of the first 10 years of FIRE will come from bond liquidation and interest/dividends, unless the stock market goes on a big run, in which case it won't matter anyway)

That sounds good in theory but a 50%/50% bond/stock allocation has lower success rate than 20/80 I think? Small difference either way.

msilenus

  • Pencil Stache
  • ****
  • Posts: 525
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #118 on: July 22, 2017, 03:35:20 PM »
Title should be cfiresim severely overestimates for non-Mustachians.  For Mustachians, it barely overestimates, due to our high savings rates.

Disagree with this characterization.  Check out the graph showing FIRE density again.  Both 10% and 50% were highly nonuniform.  It was at 90% where the effect on FIRE density was minor.  I believe that a 50% SR is much more typical of Mustachians than 90%.  90 is deep in ERE-land.

It would be interesting to have more data points, but I think it's suggestive that 50% looks so much more like 10% than it does 90%.

paddedhat

  • Handlebar Stache
  • *****
  • Posts: 2229
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #119 on: July 22, 2017, 04:08:18 PM »
I mean, I don't think anyone here will lay down and die on the street if the market crashes, so their "success" will always be 100%. Doesn't mean that their plan has worked though.

Don't be so sure. Family history claims that a great grandfather was fabulously wealthy prior to the great depression. After the market crash, he sent his five year old son off to school (my grandfather) and decided that being dead was better than being poor, so he shot himself in the head. Doubt I would repeat his performance if I lost my big pile of cheddar, but there are no shortage of folks who would, or will never recover financially, or emotionally, from a total loss of their savings/investments.

Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #120 on: July 22, 2017, 04:08:37 PM »
3) Once FIRED allocate enough to bonds/CDs, etc that a 5-10 year shit in the market won't be the end of the world since you're drawing down mostly from bonds/CDs (i have 25% in BND and will go to 5% after 10 years, so most of the first 10 years of FIRE will come from bond liquidation and interest/dividends, unless the stock market goes on a big run, in which case it won't matter anyway)

That sounds good in theory but a 50%/50% bond/stock allocation has lower success rate than 20/80 I think? Small difference either way.

It makes complete sense because sequence of return risk is isolated to the first decade of draw-down (maybe a little more depending on specifics).  This writer is sacrificing a few scenarios of dying very wealthy to limit the sequence risk.  It's called a reverse glide path strategy and is a very sound theory.

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #121 on: July 22, 2017, 04:19:15 PM »
3) Once FIRED allocate enough to bonds/CDs, etc that a 5-10 year shit in the market won't be the end of the world since you're drawing down mostly from bonds/CDs (i have 25% in BND and will go to 5% after 10 years, so most of the first 10 years of FIRE will come from bond liquidation and interest/dividends, unless the stock market goes on a big run, in which case it won't matter anyway)

That sounds good in theory but a 50%/50% bond/stock allocation has lower success rate than 20/80 I think? Small difference either way.

It makes complete sense because sequence of return risk is isolated to the first decade of draw-down (maybe a little more depending on specifics).  This writer is sacrificing a few scenarios of dying very wealthy to limit the sequence risk.  It's called a reverse glide path strategy and is a very sound theory.

Oh I read that wrong -- does anybody know the success rate of that strategy? I can't imagine it's much higher than constant 80/20...

Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #122 on: July 22, 2017, 04:52:44 PM »
Here is a Kitces article that touches on the subject.

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #123 on: July 22, 2017, 05:16:30 PM »
I just computed it.

For reference, here are the success rates for 50 year periods at 4% WR as a function of (constant) stock allocation:
50%: 79.8%
60%: 86.2%
70%: 90.4%
80%:  92.6%
90%:  95.7%
100%:  94.7%

Now, starting at 50% allocation and increasing to 100% at a given pace, I get:
over 5 years: 95.7%
over 10 years: 96.8%
over 20 years: 94.7%

so 50% -> 100% over 10 years seems to be the sweet spot. I doubt the difference is statistically significant given the small dataset though, because stopping at 90% allocation instead of 100% goes back to 95.7%.

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #124 on: July 22, 2017, 05:22:29 PM »
Oh, I see that Kitces has some results, but he seems to ramp up to the final allocation over the full 30 years, which seems too long for low stock allocations.

I see this in the comments:
Quote
- In terms of a shorter glidepath, we haven't yet tested extensively for varying the speed of the glidepath. I wouldn't be surprised at all if the optimal rising equity glidepath is at least SOMETHING faster than gliding for all 30 years (as realistically, the glidepath changes in the last 5-10 years have negligible effect anyway, because there just aren't many years remaining to compound). I doubt the optimal would be as fast as 3 years, though, as that's TOO fast and don't solve the REAL problem, which is not just bear markets but a mediocre DECADE (see http://www.kitces.com/blog/... ). A 3-year glidepath doesn't address the fact that the Dow hit 1,000 in 1966, hit 1,000 again in 1973, and hit 1,000 again in 1982. If you have a 3-year glidepath in 1966 or 1973, you averaged in "too fast" and still bore most of the volatility with little benefit. If I had to guess, we'll find that the optimal glidepath speed is probably more along the lines of gliding over the first 10-15 years and then leveling out. But again, that's still just a hypothesis we hope to test at this point. :)

so it seems his intuition that 10-15 years would be optimal was pretty close!
« Last Edit: July 22, 2017, 05:24:00 PM by gerardc »

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #125 on: July 22, 2017, 05:31:04 PM »
A natural follow-up to these simulations is coming up with the "ultimate withdrawal strategy", which I guess would be along the lines of start aggressively at a high WR%, be flexible, and ramp up stock allocation in the first decade. Unfortunately, historical data to backtest this is pretty sparse, so we'd probably be overfitting quickly, and finding it wouldn't improve things all that much, especially considering other risks that are not part of the model.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #126 on: July 22, 2017, 05:35:06 PM »
Title should be cfiresim severely overestimates for non-Mustachians.  For Mustachians, it barely overestimates, due to our high savings rates.

Disagree with this characterization.  Check out the graph showing FIRE density again.  Both 10% and 50% were highly nonuniform.  It was at 90% where the effect on FIRE density was minor.  I believe that a 50% SR is much more typical of Mustachians than 90%.  90 is deep in ERE-land.

It would be interesting to have more data points, but I think it's suggestive that 50% looks so much more like 10% than it does 90%.

I don't think fire density is relevant here beyond a curiosity.  Check out the graph showing Fire success again.  After 50% SR, it's pretty flat at 4% (not so great at higher WR)

TomTX

  • Magnum Stache
  • ******
  • Posts: 4493
  • Location: Texas
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #127 on: July 22, 2017, 06:31:42 PM »
I also worry that the very name of that thread "Stop Worrying about the 4% Rule" encourages people to accept it blindly.

I don't even accept gravity blindly, but I do accept it and encourage others to do the same.

It astounds me that anyone could read through the hundreds of pages of careful mathematical analysis this forum has done about safe withdrawal rates, and then accuse us of blind faith because of a thread title summarizing our findings.

Don't bother being astounded. It's been established he doesn't even bother to read the whole 4% thread, much less the many analysis threads that preceded and coincided with it.

The whole concept seems to be threatening his belief system. I suggest a mosey over to the Oatmeal for some background thoughts:

http://theoatmeal.com/comics/believe

BTDretire

  • Magnum Stache
  • ******
  • Posts: 3077
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #128 on: July 22, 2017, 06:51:02 PM »
I just computed it.

For reference, here are the success rates for 50 year periods at 4% WR as a function of (constant) stock allocation:
50%: 79.8%
60%: 86.2%
70%: 90.4%
80%:  92.6%
90%:  95.7%
100%:  94.7%

Now, starting at 50% allocation and increasing to 100% at a given pace, I get:
over 5 years: 95.7%
over 10 years: 96.8%
over 20 years: 94.7%

so 50% -> 100% over 10 years seems to be the sweet spot. I doubt the difference is statistically significant given the small data set though, because stopping at 90% allocation instead of 100% goes back to 95.7%.
  Those are very interesting numbers. Even though I have been reading MMM for about 2 years and have seen the simulations that a higher percentage of stock results in a longer success rate, I have 30 years behind me of "when you get close to retirement shift your assets over to 50/50 or 60/40."
 I'm 62 and it's get serious about asset allocation time. Logically I should have zero concerns about my stache, and should just leave it in the market to grow for the two kids, because we have about 34 x spending. I'll get $20k in SS in three years and another $18K in 7 years for the wife. We can easily live on $50k so that leaves a $12k shortfall. With about $64k coming with the 4% rule.
  All this argument does is convince me logically I can just leave it in the stock market.
But then, "when you get close to retirement shift your assets over to 50/50 or 60/40."
and, "when you have Critical Mass you need to adjust so you never lose it"
is cackling in my ear.
 I know that if a 2000 or 2008 happens again, I'll could easily see $400k or $500k disappear in a year.
Yes I will be fine, but still that would stink.
 I got through a $240k loss in 2008 and blossomed from 2010 until now.
 So that down time may only be 3 to 5 tears, I mean years.
Just some things running loose in my mind.

 Edit to add, I had not read the oatmeal believe comic before I posted my two lines of thought, one long ingrained and one new information, that I'm trying to incorporate. While not all that deep it runs in the same
company.
 btw, RE: the believe comic, I thought, were they dead or alive, did they cut the roots off and what kind of glue did they have to hold them together. Probably shows a lack of empathy, or a more engineer/science background.
« Last Edit: July 22, 2017, 07:19:48 PM by BTDretire »

dividendman

  • Handlebar Stache
  • *****
  • Posts: 1377
  • Age: 38
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #129 on: July 22, 2017, 07:59:47 PM »
I also worry that the very name of that thread "Stop Worrying about the 4% Rule" encourages people to accept it blindly.

I don't even accept gravity blindly, but I do accept it and encourage others to do the same.

It astounds me that anyone could read through the hundreds of pages of careful mathematical analysis this forum has done about safe withdrawal rates, and then accuse us of blind faith because of a thread title summarizing our findings.

Don't bother being astounded. It's been established he doesn't even bother to read the whole 4% thread, much less the many analysis threads that preceded and coincided with it.

The whole concept seems to be threatening his belief system. I suggest a mosey over to the Oatmeal for some background thoughts:

http://theoatmeal.com/comics/believe

The Oatmeal delivers again. Great piece.

Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #130 on: July 23, 2017, 12:21:38 AM »
I just computed it.

For reference, here are the success rates for 50 year periods at 4% WR as a function of (constant) stock allocation:
50%: 79.8%
60%: 86.2%
70%: 90.4%
80%:  92.6%
90%:  95.7%
100%:  94.7%

Now, starting at 50% allocation and increasing to 100% at a given pace, I get:
over 5 years: 95.7%
over 10 years: 96.8%
over 20 years: 94.7%

so 50% -> 100% over 10 years seems to be the sweet spot. I doubt the difference is statistically significant given the small dataset though, because stopping at 90% allocation instead of 100% goes back to 95.7%.

Thanks for the runs!

I wonder if the impact becomes more pronounced at higher WR's... like 4.5 or 5? (im too lazy at the moment, lol) 

Also an important subject for some, how much left at the end for legacy?  IOW how much legacy potential is sacrificed for slightly higher success rates?

A natural follow-up to these simulations is coming up with the "ultimate withdrawal strategy", which I guess would be along the lines of start aggressively at a high WR%, be flexible, and ramp up stock allocation in the first decade. Unfortunately, historical data to backtest this is pretty sparse, so we'd probably be overfitting quickly, and finding it wouldn't improve things all that much, especially considering other risks that are not part of the model.

Then we can get into AA debates like this one... It's fun to try an find the absolute "best" history has to offer, but I think your analysis is correct; there are too many other factors to consider, including unknown, unknowns vis a vis Talebesque black swans.  At some point we just need to stop, accept some rather foundational concepts (4% rule and heavy stock allocation), then make some intelligent adaptations to account for changing market forces and personal risk tolerances over time.

Our efforts are better placed learning to adapt to the future and/or our current personal situation vs finding what would have worked best in the past for a theoretical person.


dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #132 on: July 23, 2017, 02:19:49 AM »

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #133 on: July 23, 2017, 02:25:36 AM »
Thanks for the runs!

I wonder if the impact becomes more pronounced at higher WR's... like 4.5 or 5? (im too lazy at the moment, lol) 

Also an important subject for some, how much left at the end for legacy?  IOW how much legacy potential is sacrificed for slightly higher success rates?

Success rates for 50 year periods at 5% WR as a function of (constant) stock allocation:
50%: 39.4%
60%: 47.9%
70%: 58.5%
80%:  66.0%
90%:  72.3%
100%:  72.3%

50% -> 100% allocation ramp up
over 5 years: 73.4%
over 10 years: 69.1%
over 20 years: 63.8%

Conclusion: for 5% WR a high stock allocation is even more crucial (constant 80/20 is much worse), so makes sense that optimal ramp up period of 5 years is shorter. Then again, very noisy numbers, we're talking a handful of failed years difference between the strategies.


I didn't give a shit about any of the "facts" presented.  Maybe I have no beliefs.

Haha, same :D I think some facts were supposed to provoke a strong political reaction, but whatever.

TomTX

  • Magnum Stache
  • ******
  • Posts: 4493
  • Location: Texas
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #134 on: July 23, 2017, 07:14:46 AM »
Thanks for the runs!

I wonder if the impact becomes more pronounced at higher WR's... like 4.5 or 5? (im too lazy at the moment, lol) 

Also an important subject for some, how much left at the end for legacy?  IOW how much legacy potential is sacrificed for slightly higher success rates?

Success rates for 50 year periods at 5% WR as a function of (constant) stock allocation:
50%: 39.4%
60%: 47.9%
70%: 58.5%
80%:  66.0%
90%:  72.3%
100%:  72.3%

50% -> 100% allocation ramp up
over 5 years: 73.4%
over 10 years: 69.1%
over 20 years: 63.8%

Conclusion: for 5% WR a high stock allocation is even more crucial (constant 80/20 is much worse), so makes sense that optimal ramp up period of 5 years is shorter. Then again, very noisy numbers, we're talking a handful of failed years difference between the strategies.


Your data runs are very helpful. Would you mind re-running the glide path with different initial allocations? ie - initial 40, 50, 60, 70, 80, 90% stocks. The withdrawal rates at 4, 4.5 and 5% are probably most interesting.

msilenus

  • Pencil Stache
  • ****
  • Posts: 525
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #135 on: July 23, 2017, 12:04:34 PM »
We did a less rigorous version of this exercise on another thread (w/o the SR component, obviously.)  We found an apparent sweet spot at 70% stocks-> 100% over ten years.

https://forum.mrmoneymustache.com/investor-alley/changing-asset-allocation-as-fire-approaches/msg1241674/#msg1241674

gerardc

  • Pencil Stache
  • ****
  • Posts: 767
  • Age: 37
  • Location: SF bay area
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #136 on: July 23, 2017, 12:37:31 PM »
Your data runs are very helpful. Would you mind re-running the glide path with different initial allocations? ie - initial 40, 50, 60, 70, 80, 90% stocks. The withdrawal rates at 4, 4.5 and 5% are probably most interesting.

Here you go.

4.0% WR
ramp up over (years) \ initial stock allocation (%)405060708090100
395.796.897.997.996.895.794.7
593.695.797.997.996.895.794.7
794.796.896.897.996.895.794.7
1093.696.896.897.996.895.794.7
1594.794.797.997.996.895.794.7
inf62.879.886.290.492.695.794.7

4.5% WR
ramp up over (years) \ initial stock allocation (%)405060708090100
381.981.983.083.083.080.983.0
578.781.981.981.983.079.883.0
777.780.981.980.983.079.883.0
1077.780.980.980.981.979.883.0
1572.378.779.879.880.979.883.0
inf41.554.367.074.579.879.883.0

5.0% WR
ramp up over (years) \ initial stock allocation (%)405060708090100
371.371.371.372.372.373.472.3
570.273.472.372.373.474.572.3
769.171.372.371.374.574.572.3
1067.069.169.169.173.474.572.3
1561.763.866.067.073.474.572.3
inf31.939.447.958.566.072.372.3

5.5% WR
ramp up over (years) \ initial stock allocation (%)405060708090100
358.558.557.459.659.658.557.4
558.559.659.659.659.657.457.4
759.658.557.458.559.657.457.4
1056.456.457.456.458.557.457.4
1551.152.153.256.456.457.457.4
inf17.030.939.441.547.955.357.4

6.0% WR
ramp up over (years) \ initial stock allocation (%)405060708090100
345.747.947.948.946.850.052.1
545.747.948.946.848.951.152.1
745.746.846.846.847.948.952.1
1043.644.744.744.745.747.952.1
1539.442.642.643.644.747.952.1
inf12.818.130.936.240.445.752.1


runewell

  • Bristles
  • ***
  • Posts: 417
  • Age: 49
  • actuary
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #137 on: July 24, 2017, 10:38:29 AM »

Why continually worry about outlying statistical situations that can neither be accurately predicted or mitigated against with a passive investment allocation?  It's better to invest in adaptability skills in those potential situations.  For the 4% rule to fail there has to be both a precipitous drop in asset value soon after draw-down begins AND long-term very low real returns. Even if this happens, someone with only half of FI assets remaining is better suited to take advantage of a universally bad situation, if they adapt to it.

Vanguard says that P/E is a reliable indicator of future returns.  No formula is going to give you the day or week when the market corrects or tumbles, but it appears there are warning signs and strategies that can help. 

The 4% rule failed after the great depression, and when enough history plays out, it might also fail during some of the severe downturns of the 2000's.  Guess what?  CAPE market valuations are back up in lime with the great depression started (although I would argue it's not quite as bad now as it was then).  Don't give yourself a false sense of security that the idea from a single research paper puts the question of the safety of a 4% withdrawal rate to rest, at least with these high market valuations.

runewell

  • Bristles
  • ***
  • Posts: 417
  • Age: 49
  • actuary
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #138 on: July 24, 2017, 10:45:45 AM »
If 3.5%WR rates fails, its likely a 2.5% WR will fail too, because some horrible economic calamady has taken place which was outside the scope of backtesting.  That's a lot of extra years working for virtually no increase in FIRE safety margin. 

The 3% withdrawal rate was pretty rock-solid in the study.  2% must be super-mega-rock-solid.  It seems fairly unlikely for either to fail, in which case nobody really knows what a scenario looks like where 3% fails and 2% doesn't.  The above statement thought is complete and utter speculation and I wouldn't trust it.

runewell

  • Bristles
  • ***
  • Posts: 417
  • Age: 49
  • actuary
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #139 on: July 24, 2017, 10:46:42 AM »
The 4% rule et al is pretty dynamic but relies on static expenses in real terms.

What do you mean by "dynamic" here?

tooqk4u22

  • Magnum Stache
  • ******
  • Posts: 2530
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #140 on: July 24, 2017, 12:04:33 PM »
The 4% rule et al is pretty dynamic but relies on static expenses in real terms.

What do you mean by "dynamic" here?

The input, not the output, as it includes ever changing information of periods of time to get to a single number. 

Classical_Liberal

  • Handlebar Stache
  • *****
  • Posts: 1171
  • Age: 45
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #141 on: July 24, 2017, 04:45:56 PM »
If 3.5%WR rates fails, its likely a 2.5% WR will fail too, because some horrible economic calamady has taken place which was outside the scope of backtesting.  That's a lot of extra years working for virtually no increase in FIRE safety margin. 

The 3% withdrawal rate was pretty rock-solid in the study.  2% must be super-mega-rock-solid.  It seems fairly unlikely for either to fail, in which case nobody really knows what a scenario looks like where 3% fails and 2% doesn't.  The above statement thought is complete and utter speculation and I wouldn't trust it.

You've made it quite clear you don't trust anything around here.  I will rephrase the statement. 

If historical back-testing shows a 100% success rate for a 3.5% WR, then some sequence of economic events occurs which causes a failure of 3.5%WR.  It becomes clear the new sequence of events is outside the scope of historical back-testing, hence this data is irrelevant.  No WR (2.5 or less) is technically "safe" based on back-testing because we are in virgin territory.

In historical back-testing there is a VERY clear S-curve in portfolio success rates, the 4% rule is clearly on the far right of that curve.  Each year of additional savings has significantly diminishing returns to success rates. One is better off mitigating risk in others ways vs saving piles of extra money that remains at risk.

If you are interested in risk mitigation for unpredictable events I suggest you read Taleb.  Then come back and comment.

 

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #142 on: July 24, 2017, 05:08:47 PM »
If 3.5%WR rates fails, its likely a 2.5% WR will fail too, because some horrible economic calamady has taken place which was outside the scope of backtesting.  That's a lot of extra years working for virtually no increase in FIRE safety margin. 

The 3% withdrawal rate was pretty rock-solid in the study.  2% must be super-mega-rock-solid.  It seems fairly unlikely for either to fail, in which case nobody really knows what a scenario looks like where 3% fails and 2% doesn't.  The above statement thought is complete and utter speculation and I wouldn't trust it.

You've made it quite clear you don't trust anything around here.  I will rephrase the statement. 

If historical back-testing shows a 100% success rate for a 3.5% WR, then some sequence of economic events occurs which causes a failure of 3.5%WR.  It becomes clear the new sequence of events is outside the scope of historical back-testing, hence this data is irrelevant.  No WR (2.5 or less) is technically "safe" based on back-testing because we are in virgin territory.

In historical back-testing there is a VERY clear S-curve in portfolio success rates, the 4% rule is clearly on the far right of that curve.  Each year of additional savings has significantly diminishing returns to success rates. One is better off mitigating risk in others ways vs saving piles of extra money that remains at risk.

If you are interested in risk mitigation for unpredictable events I suggest you read Taleb.  Then come back and comment.

What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.

Tyson

  • Magnum Stache
  • ******
  • Posts: 2512
  • Age: 49
  • Location: Denver, Colorado
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #143 on: July 24, 2017, 05:29:10 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE. 

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #144 on: July 24, 2017, 09:03:54 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE.

Sure, but you are likely leaving the majority of your wealth behind, be it 25 or 50x expenses.

Tyson

  • Magnum Stache
  • ******
  • Posts: 2512
  • Age: 49
  • Location: Denver, Colorado
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #145 on: July 24, 2017, 09:10:34 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE.

Sure, but you are likely leaving the majority of your wealth behind, be it 25 or 50x expenses.

Just don't wait till SHTF.  Especially nowadays it's pretty easy to move wealth around.  It might be a concern in a 2nd or 3rd world country, but not likely to be a concern in the US or in most of Europe.  Now, if you live someplace like Greece, you should have already moved!  Take your money and get out.  A place like Greece (and Germany before WWII) is clearly in trouble and you should not ignore those warning signs. 

Besides, if a country does nationalize (like Germany or China), gold won't help - it'll be taken from you like everything else.  Whether you stay or go, physical assets can be seized just as easily as any other asset. 

maizefolk

  • Walrus Stache
  • *******
  • Posts: 5665
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #146 on: July 24, 2017, 09:16:29 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE.

Sure, but you are likely leaving the majority of your wealth behind, be it 25 or 50x expenses.

Assuming you wait to leave until after capital controls are put into place. Otherwise just transfer your money out.

If you cannot do that, how likely are you to be able to get your physical gold out of the country without having it seized? An average sized stash (say $600k) would be about 35 lbs of gold at current prices, which would be hard to hide in your carry on luggage and in a situation with capital controls I imagine you wouldn't just be able to declare it and take it with you no questions asked.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9536
  • Registered member
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #147 on: July 24, 2017, 10:01:25 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE.

Sure, but you are likely leaving the majority of your wealth behind, be it 25 or 50x expenses.

Assuming you wait to leave until after capital controls are put into place. Otherwise just transfer your money out.

If you cannot do that, how likely are you to be able to get your physical gold out of the country without having it seized? An average sized stash (say $600k) would be about 35 lbs of gold at current prices, which would be hard to hide in your carry on luggage and in a situation with capital controls I imagine you wouldn't just be able to declare it and take it with you no questions asked.

I don't personally keep gold around for SHTF scenarios, but in theory it's great for bribes, chartering travel, whatever is necessary.

I like yous guys attitude.  Just leave before SHTF or capital controls are in place.  Likewise, invest in 100% stocks and just sell before the crash.  Buy in at the bottom.  Simple, right?

These days, it's easy to move money around, as long as the electronic systems are working.  I'm not a conspiracy theorist, but I wouldn't be surprised if the government had the ability to shut down or reverse electronic money transfer at will.  You do keep your money as 1's and 0's in the bank, right?
« Last Edit: July 24, 2017, 10:04:07 PM by dragoncar »

maizefolk

  • Walrus Stache
  • *******
  • Posts: 5665
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #148 on: July 24, 2017, 10:10:52 PM »
Fortunately there is no equivalent of the efficient market hypothesis for the emergence of totalitarian regimes.

Tyson

  • Magnum Stache
  • ******
  • Posts: 2512
  • Age: 49
  • Location: Denver, Colorado
Re: cFiresim SEVERELY overestimates success rates for Mustachians
« Reply #149 on: July 24, 2017, 11:03:14 PM »
What was the SWR for Jews in nazi Germany?  East Berliners when the wall went up?  Chinese nationalists during the communist revolution?  2 vs 3% didn't make a difference and is part of the reason the permanent portfolio technically holds physical gold.
Best strategy in those situations is to already be wealthy and MOVE.

Sure, but you are likely leaving the majority of your wealth behind, be it 25 or 50x expenses.

Assuming you wait to leave until after capital controls are put into place. Otherwise just transfer your money out.

If you cannot do that, how likely are you to be able to get your physical gold out of the country without having it seized? An average sized stash (say $600k) would be about 35 lbs of gold at current prices, which would be hard to hide in your carry on luggage and in a situation with capital controls I imagine you wouldn't just be able to declare it and take it with you no questions asked.

I don't personally keep gold around for SHTF scenarios, but in theory it's great for bribes, chartering travel, whatever is necessary.

I like yous guys attitude.  Just leave before SHTF or capital controls are in place.  Likewise, invest in 100% stocks and just sell before the crash.  Buy in at the bottom.  Simple, right?

These days, it's easy to move money around, as long as the electronic systems are working.  I'm not a conspiracy theorist, but I wouldn't be surprised if the government had the ability to shut down or reverse electronic money transfer at will.  You do keep your money as 1's and 0's in the bank, right?

Or just keep your money in a bank in Switzerland.