FIRE calculator with weighted start intervals, based on FIRE chance that year?

[ender]

I’ve been mulling over something recently.

The Trinity study assumes each year for a FIRE target date is equally likely.  But this is a really flawed assumption.

So for example, saying 95% success rate - it means if you took each historical 30 year period, you’d successfully achieve a 30 year FIRE in 95% of them.

But what the Trinity study fails to consider is that each 30 year period is not at all equally likely for an aspiring early retiree to achieve FI. And are in fact wildly different.

As an example, 2000 is probably 10x or 100x more likely as a start date than 2001 or 2002, because market growth in the years prior to 2000 would create more FI folks than 2001/2002 would have.

What this means though is when the Trinity study includes intervals like those starting in 1999/2000/2001/2002 (or historical periods with similar behavior; obviously there hasn’t been 30 years yet) by treating them all the same from an outcome perspective, it ignores the fact that those four years aren’t even close to as likely for folks to achieve FI in and thus start a 30 year period.

Practically it makes the 30 year intervals much different from a weighting perspective in such a significant way that I’m actually increasingly less confident the Trinity study is useful at all. In fact I'm unaware of ANY model for FIRE that doesn't have this fundamentally flawed assumption.

All this to say, does anyone know of an article/study/calculator using a more appropriate set of FIRE intervals? I’m envisioning something where each start interval is “weighted” in some way that would reflect how likely it would be for someone to actually achieve FI and be in a position to FIRE that year.

Bonus points if you can input yearly savings to make the model account for how much you can “overpower” market losses with raw savings.

[secondcor521]

Your weighting idea assumes people retire as soon as they have the capital to do so.

This is often true but not always, and maybe not even true even nearly always.

Some people will hang on a few more years to pad the nest further, either because they have uncertain future expenses (teenagers? college? weddings? change in lifestyle) or because they suddenly worry about the 4% rule or because they raise their lifestyle goals.  Some will continue working because they have FI/FU freedom and the job isn't that bad (me).  Some will hang on for a milestone year, like 20 years of service or retiree medical (latter only true for early-50's FIREes).  For some a combination of these will mean many more years of work past the 25X date.

Another person has pointed out that prospective FIREes usually have high savings rates, which flattens the distribution out even further because FIRE dates become at least somewhat more a function of savings and less a function of market returns.

On the flip side, some might be beyond exasperated in the midst of a downturn - "So close, and then this #$*& market pullback happened!" and find ways to cut expenses or make more generous assumptions or include Social Security and retire into the teeth of a downturn.  Especially if they were mentally exhausted from work.

There's also the binary nature hidden inside your questions which impacts the answer.  Rarely do people hit 25x on a given date, quit work, then never work again.  People go back to work, people have gigs, people slowly taper, people consult with their former employers.  In these cases which are less black-and-white but more real world likely, a FIRE start date is more of a fuzzy concept.

We're all still trying to predict the unclear future, which is still unknowable regardless of how you weight the past.  Even if a FIREee has a 100% certain result based on start-year weighting of historical stock market and inflation data, they could get hit by a bus the day after they retire (technically a FIREcalc success), or have unexpected expenses like 47 grandchildren, or live 20 years past that 30 year window, or hit a not-yet-seen worse-than-ever-before sequence of stock/bond/inflation data.

I do agree with you that there is some effect.  However, since at high percentage historical success rates like 95% most years are successes anyway and any weighting effect is mild, I think the overall effect of any reasonable weighting scheme would be mild overall, and probably outweighed by the very rarely mentioned life expectancy effect (getting hit by a bus is a success -- see https://retireearlyhomepage.com/swrlife.html for the AFAIK original article) and the frequently mentioned "FIREee resiliency and adaptability" argument.

[ender]

I do agree with you that there is some effect.  However, since at high percentage historical success rates like 95% most years are successes anyway and any weighting effect is mild, I think the overall effect of any reasonable weighting scheme would be mild overall, and probably outweighed by the very rarely mentioned life expectancy effect (getting hit by a bus is a success -- see https://retireearlyhomepage.com/swrlife.html for the AFAIK original article) and the frequently mentioned "FIREee resiliency and adaptability" argument.

I suspect the impact is far greater based on how the historical models are included.

For example, consider the 1920 - 1950 time period.

FIRE tools all consider each year in that and the subsequent 30 years as equally likely. But it's significantly more likely someone would have FIRE'ed in 1928 than in 1932. Or than in 1935.

But the modeling all treats each start year as equally likely. Imo the relevant of datapoints for series starting in 1932-1935 being relevant to FIRE calculators is effectively 0%. Because historically it would never have been a time someone considered FIRE (with the exception of someone who saved a lot of $ in the years prior to overcome the market drops prior).

[RWD]

Bonus points if you can input yearly savings to make the model account for how much you can “overpower” market losses with raw savings.

I think this is a key variable and not optional to properly model this. Someone with a very low savings rate will be likely hit their number with help from a sudden bull market (very affected by this bias). Someone with a high savings rate will be more likely to hit their FIRE number in some arbitrary year.

[ender]

This is an interesting analysis on the subject which I think is the only one I've really seen - https://earlyretirementnow.com/2017/12/13/the-ultimate-guide-to-safe-withdrawal-rates-part-22-endogenous-retirement-timing/comment-page-1/

[secondcor521]

I do agree with you that there is some effect.  However, since at high percentage historical success rates like 95% most years are successes anyway and any weighting effect is mild, I think the overall effect of any reasonable weighting scheme would be mild overall, and probably outweighed by the very rarely mentioned life expectancy effect (getting hit by a bus is a success -- see https://retireearlyhomepage.com/swrlife.html for the AFAIK original article) and the frequently mentioned "FIREee resiliency and adaptability" argument.

I suspect the impact is far greater based on how the historical models are included.

For example, consider the 1920 - 1950 time period.

FIRE tools all consider each year in that and the subsequent 30 years as equally likely. But it's significantly more likely someone would have FIRE'ed in 1928 than in 1932. Or than in 1935.

But the modeling all treats each start year as equally likely. Imo the relevant of datapoints for series starting in 1932-1935 being relevant to FIRE calculators is effectively 0%. Because historically it would never have been a time someone considered FIRE (with the exception of someone who saved a lot of $ in the years prior to overcome the market drops prior).

Rats.  You appear to have ignored the vast majority of my counterarguments and repeated your original one.

Let's take just one sentence of the one paragraph of mine you quoted and use your numbers.

www.firecalc.com with default inputs results in a 95.1% success rate on a 30 year basis because 6 out of 123 failed.
 Alternately 117/123 succeeded.  Let's say you choose to completely ignore 1932-1935, which is four start years.  That would change the math from 117/123 to 113/119.  Which is 94.9%.  Throw out another four arbitrary start years of your choosing to handle the 1966ish start years and you have 109/115, which is 94.8%.

I suppose you could try to argue that you need to use 113/123 and 109/123, but that argument would not make sense to me.  The percentage success is based on what proportion of time a start year fails - a retiree not starting in 1932-1935 can't logically include those as start years because they didn't start in those years...right?

But even if you had a good argument for that, 109/123 is still 88.6%.  Not great, but not really bad odds considering the rest of the counterarguments.  And that's tossing out eight starting years affected by the two worst known periods in FIRE history.

And that's ignoring the rest of my counterarguments.

I still think it barely moves the needle.

[secondcor521]

I'll give a different, less argumentative answer too.

If I were to do something like this, I'd probably find as many "20xx cohort" threads as I could here and at other forums.  I'd then total up approximately how many members were in each cohort.  I'd then find the degree to which the stock market performance in that year and the number of members correlated with each other.  I'd then need to look up how to use that degree of correlation to weight all the start years based on their stock market performance and the weighting factor, because I don't know that answer from a statistics point of view.  I'd then rerun FIREcalc with those weightings.

(Paging @maizefolk who probably knows the statistics methods.)

[ender]

I do agree with you that there is some effect.  However, since at high percentage historical success rates like 95% most years are successes anyway and any weighting effect is mild, I think the overall effect of any reasonable weighting scheme would be mild overall, and probably outweighed by the very rarely mentioned life expectancy effect (getting hit by a bus is a success -- see https://retireearlyhomepage.com/swrlife.html for the AFAIK original article) and the frequently mentioned "FIREee resiliency and adaptability" argument.

I suspect the impact is far greater based on how the historical models are included.

For example, consider the 1920 - 1950 time period.

FIRE tools all consider each year in that and the subsequent 30 years as equally likely. But it's significantly more likely someone would have FIRE'ed in 1928 than in 1932. Or than in 1935.

But the modeling all treats each start year as equally likely. Imo the relevant of datapoints for series starting in 1932-1935 being relevant to FIRE calculators is effectively 0%. Because historically it would never have been a time someone considered FIRE (with the exception of someone who saved a lot of $ in the years prior to overcome the market drops prior).

Rats.  You appear to have ignored the vast majority of my counterarguments and repeated your original one.

Let's take just one sentence of the one paragraph of mine you quoted and use your numbers.

www.firecalc.com with default inputs results in a 95.1% success rate on a 30 year basis because 6 out of 123 failed.
 Alternately 117/123 succeeded.  Let's say you choose to completely ignore 1932-1935, which is four start years.  That would change the math from 117/123 to 113/119.  Which is 94.9%.  Throw out another four arbitrary start years of your choosing to handle the 1966ish start years and you have 109/115, which is 94.8%.

I suppose you could try to argue that you need to use 113/123 and 109/123, but that argument would not make sense to me.  The percentage success is based on what proportion of time a start year fails - a retiree not starting in 1932-1935 can't logically include those as start years because they didn't start in those years...right?

But even if you had a good argument for that, 109/123 is still 88.6%.  Not great, but not really bad odds considering the rest of the counterarguments.  And that's tossing out eight starting years affected by the two worst known periods in FIRE history.

And that's ignoring the rest of my counterarguments.

I still think it barely moves the needle.

I ignored it since it's reductionistic and missing the point I'm getting at.

Even this analysis is effectively saying "almost all years are equally as likely for someone to FIRE in" which I guess... if that's your underlying assumption, then what I'm saying in this thread is completely different.

It's not just a year or two that I'm talking about. I'm talking about a non-trivial percentage of years included in the overall study and the entire framing and analyses people use it for.

And I guess we just disagree if moving from 95.1% to 88.6% is not meaningful.

Either way though luckily the article I was able to find and linked above does the detailed analysis that I was curious about. Though I wish he also showed the % of 30/40/50 year FIRE scenarios that switch from pass/fail if you account for the endogenous retirement dates.

[secondcor521]

I do agree with you that there is some effect.  However, since at high percentage historical success rates like 95% most years are successes anyway and any weighting effect is mild, I think the overall effect of any reasonable weighting scheme would be mild overall, and probably outweighed by the very rarely mentioned life expectancy effect (getting hit by a bus is a success -- see https://retireearlyhomepage.com/swrlife.html for the AFAIK original article) and the frequently mentioned "FIREee resiliency and adaptability" argument.

I suspect the impact is far greater based on how the historical models are included.

For example, consider the 1920 - 1950 time period.

FIRE tools all consider each year in that and the subsequent 30 years as equally likely. But it's significantly more likely someone would have FIRE'ed in 1928 than in 1932. Or than in 1935.

But the modeling all treats each start year as equally likely. Imo the relevant of datapoints for series starting in 1932-1935 being relevant to FIRE calculators is effectively 0%. Because historically it would never have been a time someone considered FIRE (with the exception of someone who saved a lot of $ in the years prior to overcome the market drops prior).

Rats.  You appear to have ignored the vast majority of my counterarguments and repeated your original one.

Let's take just one sentence of the one paragraph of mine you quoted and use your numbers.

www.firecalc.com with default inputs results in a 95.1% success rate on a 30 year basis because 6 out of 123 failed.
 Alternately 117/123 succeeded.  Let's say you choose to completely ignore 1932-1935, which is four start years.  That would change the math from 117/123 to 113/119.  Which is 94.9%.  Throw out another four arbitrary start years of your choosing to handle the 1966ish start years and you have 109/115, which is 94.8%.

I suppose you could try to argue that you need to use 113/123 and 109/123, but that argument would not make sense to me.  The percentage success is based on what proportion of time a start year fails - a retiree not starting in 1932-1935 can't logically include those as start years because they didn't start in those years...right?

But even if you had a good argument for that, 109/123 is still 88.6%.  Not great, but not really bad odds considering the rest of the counterarguments.  And that's tossing out eight starting years affected by the two worst known periods in FIRE history.

And that's ignoring the rest of my counterarguments.

I still think it barely moves the needle.

I ignored it since it's reductionistic and missing the point I'm getting at.

Even this analysis is effectively saying "almost all years are equally as likely for someone to FIRE in" which I guess... if that's your underlying assumption, then what I'm saying in this thread is completely different.

It's not just a year or two that I'm talking about. I'm talking about a non-trivial percentage of years included in the overall study and the entire framing and analyses people use it for.

And I guess we just disagree if moving from 95.1% to 88.6% is not meaningful.

Either way though luckily the article I was able to find and linked above does the detailed analysis that I was curious about. Though I wish he also showed the % of 30/40/50 year FIRE scenarios that switch from pass/fail if you account for the endogenous retirement dates.

It seems we're completely talking past each other.  I'm not sure why.

Anyway, you seem to have found at least a partial answer to your question.  Maybe you could send ERN a request to address your request in the last sentence of your reply above ("endogenous").

Good luck to you.

[Ron Scott]

This may be interesting, but it doesn’t account for the major problem with the 4% Rule. The Rule was developed with investment return data during a unique period in US and world history and is therefore is biased by a restriction in range.The conditions present in those times reflect specific global challenges and successes, and will almost certainly not repeat themselves in the same way.

The disclaimer contains all you need to know and is not just fine print: “Past performances is not indicative of future results.”

[Laura33]

So fundamentally, this is about SORR.  The ERN article was a good explanation of how this happens, but it's logical: if you're targeting a particular NW figure for retirement, you're most likely to hit that target in a bull market, but the further you are into a bull market, the more likely the market is to drop shortly after you retire, which means it's more likely your investments will be depleted in the early years, which increases SORR. 

ERN does a good job of trying to quantify that risk, and it turns out to be relatively small.  This is also entirely logical, because the market tends to go up a lot more often than it goes down -- we see long rides up, followed by steep drop-offs over a short period of time, followed by maybe some time sort of wiggling around flat, but soon morphing into another long ride up.  So unless you are unlucky enough to retire literally right at the peak of the market, your investments will continue to increase in value immediately after you retire -- so even if you planned on a flat 4% withdrawal rate, you won't actually need that full 4% to cover your expenses for that first month or year or five years, thus allowing your portfolio to continue to grow more than projected for that period.  And that, in turn, provides some degree of cushion when the market does go down.

IMO, it's not particularly valuable to spend all the time and energy trying to put a precise figure on what the SORR would be in any particular future year/month/week.  I mean, we can never predict the future, we can't possibly know how long any given bull market will continue.  Rather than put all your eggs into the "I am going to achieve perfect quantification" basket, the more practical approach is to just assume that something bad is going to happen (because it always does) and take some steps to protect yourself against the downside risk that is always there -- things like immediate annuities to provide a guaranteed income stream, or a CD/bond ladder to ensure you have enough cash available to ride out a market drop, or just having a lot of flex in the budget to adjust your spending to 4% of the current value of your assets after a drop. 

I think it is just a matter of different perspectives in the accumulation phase vs. the nearing-and-post-retirement phase.  When you're accumulating, your goal is to get there as fast as possible, so you can afford to invest aggressively -- it's all about increasing your upside potential.  And the 4% rule provides a perfectly reasonable target for that phase.  When you get closer to the actual decision, however, your priorities need to shift in part to minimizing your downside risk.  If you take the same "100% stocks, maximize my upside" approach into retirement, that's the thing that's going to set you up for long-term failure, not the fact that your retirement projections were based on the 4% rule. 

[ender]

This may be interesting, but it doesn’t account for the major problem with the 4% Rule. The Rule was developed with investment return data during a unique period in US and world history and is therefore is biased by a restriction in range.The conditions present in those times reflect specific global challenges and successes, and will almost certainly not repeat themselves in the same way.

The 4% rule has a few problems, but what you said isn't even close to a "major" problem with it.

So fundamentally, this is about SORR.  The ERN article was a good explanation of how this happens, but it's logical: if you're targeting a particular NW figure for retirement, you're most likely to hit that target in a bull market, but the further you are into a bull market, the more likely the market is to drop shortly after you retire, which means it's more likely your investments will be depleted in the early years, which increases SORR. 

...

IMO, it's not particularly valuable to spend all the time and energy trying to put a precise figure on what the SORR would be in any particular future year/month/week.  I mean, we can never predict the future, we can't possibly know how long any given bull market will continue.  Rather than put all your eggs into the "I am going to achieve perfect quantification" basket, the more practical approach is to just assume that something bad is going to happen (because it always does) and take some steps to protect yourself against the downside risk that is always there -- things like immediate annuities to provide a guaranteed income stream, or a CD/bond ladder to ensure you have enough cash available to ride out a market drop, or just having a lot of flex in the budget to adjust your spending to 4% of the current value of your assets after a drop. 

Yeah. I am surprised the increase in risk is small as well, at least if the ERN article figures are accurate.

ERN does a good job of trying to quantify that risk, and it turns out to be relatively small.  This is also entirely logical, because the market tends to go up a lot more often than it goes down -- we see long rides up, followed by steep drop-offs over a short period of time, followed by maybe some time sort of wiggling around flat, but soon morphing into another long ride up.  So unless you are unlucky enough to retire literally right at the peak of the market, your investments will continue to increase in value immediately after you retire -- so even if you planned on a flat 4% withdrawal rate, you won't actually need that full 4% to cover your expenses for that first month or year or five years, thus allowing your portfolio to continue to grow more than projected for that period.  And that, in turn, provides some degree of cushion when the market does go down.

This is one of the things I've thought of and I really like the way ENR visualized this:



It makes you realize while some of the red dots are unlucky and at the peak, in a lot of cases the interval behind them is pretty short. Other than the 1920, 1970, and 2000 decades though.

I think it is just a matter of different perspectives in the accumulation phase vs. the nearing-and-post-retirement phase.  When you're accumulating, your goal is to get there as fast as possible, so you can afford to invest aggressively -- it's all about increasing your upside potential.  And the 4% rule provides a perfectly reasonable target for that phase.  When you get closer to the actual decision, however, your priorities need to shift in part to minimizing your downside risk.  If you take the same "100% stocks, maximize my upside" approach into retirement, that's the thing that's going to set you up for long-term failure, not the fact that your retirement projections were based on the 4% rule. 

I wonder too how some of the crazy outlier scenarios for FIRE in calculators influence this - if you put 100% stocks in, you get crazy results like if you FIRE'd in 1932 and you end up with stupidly high portfolios (or really any year after a market crash). This definitely results in a perception of overly high expected long term net worth from 30 year FIRE periods.

I think the real takeaway is to recognize just how meaningful SoR actually is here and do something once you hit your number to minimize it for say the first 5 years.

[maizefolk]

Have folks here had a chance to check out the thread from (no longer active as far as I know) forum member gerardc on this topic? Complete with python code to implement his statistical model.



https://forum.mrmoneymustache.com/welcome-to-the-forum/cfiresim-severely-overestimates-success-rates-for-mustachians/msg1625045/#msg1625045

There is an effect from the concentration of the years in which people hit 25x expenses (or 33.3x or whatever) near market peaks.

For people saving only 10-20% of their income (where market returns make a big different in ones timeline to retirement) the failure rate of the 4% rule is perhaps 2x as high looking at which years the most people would have saved enough to retire in vs weighting each potential starting year equally.


But the effect size is reasonably modest so long as you're saving a larger percentage of your income. At  that point the number of years until FIRE is really driven by your savings rate, not market returns.

[bluecollarmusician]

This may be interesting, but it doesn’t account for the major problem with the 4% Rule. The Rule was developed with investment return data during a unique period in US and world history and is therefore is biased by a restriction in range.The conditions present in those times reflect specific global challenges and successes, and will almost certainly not repeat themselves in the same way.

The disclaimer contains all you need to know and is not just fine print: “Past performances is not indicative of future results.”

Hi, @Ron Scott - there are plenty more "problems" with the Trinity Study to use as the basis for an "infallible" "4% rule:

(to steal... https://www.mrmoneymustache.com/2012/05/29/how-much-do-i-need-for-retirement/)-

The trinity study is based on a prosperity anomaly: the United States during its boom years. You can’t project good times like that into the future, because we’re just about to enter the Doom Years!

Economic growth and stock appreciation was all based on cheap fossil fuels. How will this all look after Peak Oil hits us!?

You can’t take a one-size-fits-all rule and apply it to something as varied as an economy and an individual’s life! My health care costs could go up! Hyperinflation could strike!

Even at a 4% withdrawal rate, there’s still a chance of portfolio failure. That means I’ll be flat broke and out on the street in my old age. I recommend doubling your savings, and going for a 2% SWR instead because there’s never been a failure in that scenario!


But it also assumes you will:


never earn any more money through part-time work or self-employment projects

never collect a single dollar from social security or any other pension plan

never adjust spending to account for economic reality like a huge recession

never substitute goods to compensate for inflation or price fluctuation (vacation in a closer place one year during  an oil price spike, or switch to almond milk in the event of a dairy milk embargo).

never collect any inheritance from the passing of parents or other family members

never do what most old people tend to do according to studies – spend less as they age


It is by no means a rule, but a rule of thumb for people as they prepare for relying on capital as a primary source of income. Perhaps for any given individual it would be 3%, 4%, 5%; although historically going as far back as the middle ages 3% above "inflation" has been consistently achievable, which would suggest the "safe" withdrawal rate will likely be somewhere above that if you don't mind drawing down your assets over time.  I agree with the assertion

“Past performances is not indicative of future results.”

.

But it is unlikely that the next 30-50 years will be "worse" than the last thousand.

To the op, @ender  I think this is a great thought experiment (and I really like ERN's thoughts on it as well).  Based on seeing the #'s I agree with you that I think it is much more likely for someone to hit their "#" and therefore more likely to retire after a run-up.  But I agree with @secondcor521 's assessment- with a proper asset allocation, and a solid plan an ER person should be able to overcome SORR.  The trinity study was 60/40 (I believe) Any talk of retiring 100 equities is not understanding the Study or it's implications.  Someone planning to retire on a high % of stocks (especially after a big run-up) is taking on a lot of risk.  Another way of thinking about it might be: if you plan to retire the second you hit 25x's you will be far more likely to be retiring at the top of the market.  And based on the Trinity study you will likely still be.   If you have flexible spending, ability to earn more money, and a good support system that seems like a great margin of safety.

[bluecollarmusician]

I wonder too how some of the crazy outlier scenarios for FIRE in calculators influence this - if you put 100% stocks in, you get crazy results like if you FIRE'd in 1932 and you end up with stupidly high portfolios (or really any year after a market crash). This definitely results in a perception of overly high expected long term net worth from 30 year FIRE periods.

I think the real takeaway is to recognize just how meaningful SoR actually is here and do something once you hit your number to minimize it for say the first 5 years.

Not sure I understand what you mean... whose perceptions? I don't think anyone seriously planning on retirement thinks that "betting it all on red" is the best way to retire securely.  It's the equivalent of saying you only need to save one dollar for retirement just because you need to buy that lotto ticket.

The key difference here is the Trinity Study was designed to account for the worst case historical scenarios and find what would NOT have failed.  The 4% "rule" (though I used that term loosely) is based not on maximizing returns, but minimizing failure rates.  To your original point: You might be more likely to pull the plug at the top, but fortunately, if you do it right, it shouldn't really matter.

[ender]

To the op, @ender  I think this is a great thought experiment (and I really like ERN's thoughts on it as well).  Based on seeing the #'s I agree with you that I think it is much more likely for someone to hit their "#" and therefore more likely to retire after a run-up.  But I agree with @secondcor521 's assessment- with a proper asset allocation, and a solid plan an ER person should be able to overcome SORR.  The trinity study was 60/40 (I believe) Any talk of retiring 100 equities is not understanding the Study or it's implications.  Someone planning to retire on a high % of stocks (especially after a big run-up) is taking on a lot of risk.  Another way of thinking about it might be: if you plan to retire the second you hit 25x's you will be far more likely to be retiring at the top of the market.  And based on the Trinity study you will likely still be.   If you have flexible spending, ability to earn more money, and a good support system that seems like a great margin of safety.

Just to be clear I do not think this "sinks" or otherwise ruins the 4% rule, which I think has a whole ton of other safeties built into it such that I still think 4% is a reasonable rule of thumb even if it's "only" a 85% pass rate historically.

@maizefolk that thread and the linked one by Sol are really interesting. I hadn't considered the impact that savings rate also has but that's another way to consider the impact of powering through market downturns/etc with additional savings I reference in the OP.

The funny thing is I commented in that thread but clearly didn't understand the implications until more recently :)

[2sk22]

I'll give a different, less argumentative answer too.

If I were to do something like this, I'd probably find as many "20xx cohort" threads as I could here and at other forums.  I'd then total up approximately how many members were in each cohort.  I'd then find the degree to which the stock market performance in that year and the number of members correlated with each other.  I'd then need to look up how to use that degree of correlation to weight all the start years based on their stock market performance and the weighting factor, because I don't know that answer from a statistics point of view.  I'd then rerun FIREcalc with those weightings.

(Paging @maizefolk who probably knows the statistics methods.)

But many individual stories of the decisions of early retirement have nothing to do with the market conditions! I retired at the end of 2020 because I was sick of work not because our net worth had reached a specific number. I could have retired much earlier but worked until it became a chore rather than a pleasure. Also, the fact that my wife still works also fed into the decision as she continues to earn a fabulous salary.

[secondcor521]

I'll give a different, less argumentative answer too.

If I were to do something like this, I'd probably find as many "20xx cohort" threads as I could here and at other forums.  I'd then total up approximately how many members were in each cohort.  I'd then find the degree to which the stock market performance in that year and the number of members correlated with each other.  I'd then need to look up how to use that degree of correlation to weight all the start years based on their stock market performance and the weighting factor, because I don't know that answer from a statistics point of view.  I'd then rerun FIREcalc with those weightings.

(Paging @maizefolk who probably knows the statistics methods.)

But many individual stories of the decisions of early retirement have nothing to do with the market conditions! I retired at the end of 2020 because I was sick of work not because our net worth had reached a specific number. I could have retired much earlier but worked until it became a chore rather than a pleasure. Also, the fact that my wife still works also fed into the decision as she continues to earn a fabulous salary.

Totally agree.  I made that general point in the second post on this thread, although I didn't give your type of experience as a specific example:

Some people will hang on a few more years to pad the nest further, either because they have uncertain future expenses (teenagers? college? weddings? change in lifestyle) or because they suddenly worry about the 4% rule or because they raise their lifestyle goals.  Some will continue working because they have FI/FU freedom and the job isn't that bad (me).  Some will hang on for a milestone year, like 20 years of service or retiree medical (latter only true for early-50's FIREes).  For some a combination of these will mean many more years of work past the 25X date.

[Ron Scott]

This may be interesting, but it doesn’t account for the major problem with the 4% Rule. The Rule was developed with investment return data during a unique period in US and world history and is therefore is biased by a restriction in range.The conditions present in those times reflect specific global challenges and successes, and will almost certainly not repeat themselves in the same way.

The disclaimer contains all you need to know and is not just fine print: “Past performances is not indicative of future results.”

Hi, @Ron Scott - there are plenty more "problems" with the Trinity Study to use as the basis for an "infallible" "4% rule:

(to steal... https://www.mrmoneymustache.com/2012/05/29/how-much-do-i-need-for-retirement/)-

The trinity study is based on a prosperity anomaly: the United States during its boom years. You can’t project good times like that into the future, because we’re just about to enter the Doom Years!

Economic growth and stock appreciation was all based on cheap fossil fuels. How will this all look after Peak Oil hits us!?

You can’t take a one-size-fits-all rule and apply it to something as varied as an economy and an individual’s life! My health care costs could go up! Hyperinflation could strike!

Even at a 4% withdrawal rate, there’s still a chance of portfolio failure. That means I’ll be flat broke and out on the street in my old age. I recommend doubling your savings, and going for a 2% SWR instead because there’s never been a failure in that scenario!


But it also assumes you will:


never earn any more money through part-time work or self-employment projects

never collect a single dollar from social security or any other pension plan

never adjust spending to account for economic reality like a huge recession

never substitute goods to compensate for inflation or price fluctuation (vacation in a closer place one year during  an oil price spike, or switch to almond milk in the event of a dairy milk embargo).

never collect any inheritance from the passing of parents or other family members

never do what most old people tend to do according to studies – spend less as they age


It is by no means a rule, but a rule of thumb for people as they prepare for relying on capital as a primary source of income. Perhaps for any given individual it would be 3%, 4%, 5%; although historically going as far back as the middle ages 3% above "inflation" has been consistently achievable, which would suggest the "safe" withdrawal rate will likely be somewhere above that if you don't mind drawing down your assets over time.  I agree with the assertion

“Past performances is not indicative of future results.”

.

But it is unlikely that the next 30-50 years will be "worse" than the last thousand.

The 4% rule has not been tested for 1000 years. The rule was developed in 1994 and was based on historical data from part of the 20th century.

It has nothing to do with Social Security, as it speaks to the percentage of your invested assets that can be drawn down and spent each year during retirement. And if you run out of money when you’re 80 years old, you are certainly free to go back to work.

You can rationalize all day long about how safe the 4% rule is, but the truth is that neither of us knows what the future will bring and we have no idea whatsoever whether it will be better or worse for investments.

And while we’re talking about what took place in the Middle Ages LOL lets not forget the general history of “retirement”.

Retirement became common in the second half of the 20th century as life expectancy increased and the economy grew. Before that, most people worked until they died. In the early 1900s, the average life expectancy was around 47 years, so people didn't have much time to retire.

“Retirement” is possible given modern life expectancy, the unique economic growth in the 20th century given the Industrial Revolution and advances in productivity, and government programs like SS.

[bluecollarmusician]

Hi, @Ron Scott

The 4% rule has not been tested for 1000 years. The rule was developed in 1994 and was based on historical data from part of the 20th century.

I am not suggesting it has :-).  I am familiar with both the Trinity Study as well as extensive work by Bill Bengen and his SAFEMAX withdrawal rate. (Btw, here's some interesting updated info on that... it's a good read, it includes updated info on suggested starting withdrawal rates by accounting for CSI....https://www.fa-mag.com/news/choosing-the-highest--safe--withdrawal-rate-at-retirement-57731.html?section=3)

To quote myself:

It is by no means a rule, but a rule of thumb for people as they prepare for relying on capital as a primary source of income. Perhaps for any given individual it would be 3%, 4%, 5%

It has nothing to do with Social Security, as it speaks to the percentage of your invested assets that can be drawn down and spent each year during retirement. And if you run out of money when you’re 80 years old, you are certainly free to go back to work.

I am not sure I understand your point.  You are correct, the Trinity Study does not reference SS or or the ability to go back to work.  This is a feature not a flaw of the argument that 4% is likely a robust (and conservative) starting place for a withdrawal.  Perhaps instead frame it all as "additional income you may get that would allow you to not draw on your assets." 

You can rationalize all day long about how safe the 4% rule is, but the truth is that neither of us knows what the future will bring and we have no idea whatsoever whether it will be better or worse for investments.

Here is a point where I hear you, I understand you, and agree with you 100%- all plans are merely starting points.  There is no point where one can blindly put your head in the sand and ignore realities.

[bluecollarmusician]

Retirement became common in the second half of the 20th century as life expectancy increased and the economy grew. Before that, most people worked until they died. In the early 1900s, the average life expectancy was around 47 years, so people didn't have much time to retire.

“Retirement” is possible given modern life expectancy, the unique economic growth in the 20th century given the Industrial Revolution and advances in productivity, and government programs like SS.

I agree with you that the modern notion of retirement is new; and is already outdated since DB pension have become extinct during my working lifetime.  But the idea deriving income from capital is not new.  People have been doing it for centuries.  Suggest this is the way to shift mindset- you are not "retiring" instead you are learning to live like a rentier. 

[maizefolk]

In the early 1900s, the average life expectancy was around 47 years, so people didn't have much time to retire.

This is a very common misconception/misunderstanding of life expectancy at birth numbers which, until about a century ago, were dominated by high rates of infant and childhood mortality (only two thirds of human beings survived even five years past their birth).

Here's an example showing the ages at which people born in Sweden in 1800 and 1900 died as well as a projection of the ages at which people born in Sweden in 2000 will die. Even when life expectancy was much lower than it was today, people who made it past adolescence had a reasonable chance of living into their 70s.

[ender]

Anyone looking to make a FIRE calculator and/or consider results of things such as the Trinity study in their analysis for "can I FIRE?" has already assumed that you can use historical data in some capacity to predict future results.

If you disagree with that fundamental assumption that we can use historical data to predict FIRE results, it's an entirely different conversation than what I'm bringing up here. There's a many page long thread about the 4% rule which addresses a whole host of "why I'm a special snowflake and the 4% rule doesn't apply to me" considerations that is far more appropriate for that conversation.

[vand]

Yes, this has been discussed before on these pages. I don't personally see it as a "problem" or "failure" of the 4% rule as I do of the failure of most people who want simple definitive answers when and where they are not appropriate... Extrapolating B, C & D from an initial premise of A without considering the important nuances.