The 4% Rule – A Castle in the Air

Ask your investment, budget, and other money related questions here
zbigi
Posts: 978
Joined: Fri Oct 30, 2020 2:04 pm

Re: The 4% Rule – A Castle in the Air

Post by zbigi »

jacob wrote:
Mon Sep 26, 2022 7:10 am
That makes perfect sense, thanks. The law capping the amount of treasury stock a company may hold pretty much kills my scenario.

WFJ
Posts: 416
Joined: Sat Apr 24, 2021 11:32 am

Re: The 4% Rule – A Castle in the Air

Post by WFJ »

I can only respond to questions about hypothesis testing, data and analysis. Statistical methods are more of an art than a science and debate is always preferred in all domains of science compared to consensus, opinions and expert decrees.

My statical analysis tools do not predict the future in any domain, only reveals the probability of outcomes.

The 4% reminds me of the problem of a call center/data center managing flow or a dam engineer managing water flow where there is a fixed outflow with variable inputs where some slack is required, which has a cost.

zbigi
Posts: 978
Joined: Fri Oct 30, 2020 2:04 pm

Re: The 4% Rule – A Castle in the Air

Post by zbigi »

WFJ wrote:
Tue Sep 27, 2022 2:39 pm
I can only respond to questions about hypothesis testing, data and analysis. Statistical methods are more of an art than a science and debate is always preferred in all domains of science compared to consensus, opinions and expert decrees.

My statical analysis tools do not predict the future in any domain, only reveals the probability of outcomes.

The 4% reminds me of the problem of a call center/data center managing flow or a dam engineer managing water flow where there is a fixed outflow with variable inputs where some slack is required, which has a cost.
In trading, backtesting a trading algorithm (running it against historical data and seeing if it makes or loses money) is not considered a convincing proof that the algorithm is a viable money-maker. That is because, the future is often very different than the past (plus, the algorithm can be overfit: https://en.wikipedia.org/wiki/Overfitting).

classical_Liberal
Posts: 2283
Joined: Sun Mar 20, 2016 6:05 am

Re: The 4% Rule – A Castle in the Air

Post by classical_Liberal »

mathiverse wrote:
Sun Aug 28, 2022 1:02 pm
c_L, now that you've got a few more years under your belt. Do you still worry about growing out of your current spending levels? How has your thinking changed?
My thinking has changed because its become much harder to track and understand spending.

For example, last year I spent about 26K. But I also bought a van for about 9K, and its current value is about 12K. So did I make 3K or spent 9K on that transaction considering I still own the van?

I guess I could say that my needs at the low spending level have not really changed. If I made it my goal to have a low spend year, I could drop my spending significantly lower than I ever have before (12K ish). That would not be a burden at all, I can think of many awesome things to do in a year on that budget because I "invested" in resilience. But, that's not the priority anymore. I'm more willing to spend money to make progress in areas of my life that matter to me than I was when I wrote the reference quote. Sometimes those changes create value added and an increase in physical property value (like above). Sometimes the spending creates value added in my knowledge, skill, or happiness.

IOW, I have not grown tired of the low spend lifestyle. However, sometimes I spend more to invest in things that just don't show up on a spending spreadsheet. Yet these things create a resiliency that would allow me to spend much less if it became a priority again.

ertyu
Posts: 2893
Joined: Sun Nov 13, 2016 2:31 am

Re: The 4% Rule – A Castle in the Air

Post by ertyu »

jacob wrote:
Thu Sep 26, 2019 7:47 am
@daylen - The assumption/assertion of stationarity is what turns Trinity from a description of the past into a prediction of the future. The disagreement is between whether the future is like or no worse than the past because world wars, depression, etc. (stationarity holds) or it is because declining productivity and population growth trends (stationarity does not hold). These issues are essentially apples and oranges which is why/where the debate above is stuck. A way forward would be to detrend the past data and replace it with a model of the declining megatrends so that one has a dataset of world wars+depression+... with declining macrofactors and then run the MC again.
was backreading, came across this. to the best of your guys' knowledge, has anyone done this and what were the results?

zbigi
Posts: 978
Joined: Fri Oct 30, 2020 2:04 pm

Re: The 4% Rule – A Castle in the Air

Post by zbigi »

jacob wrote: ↑Thu Sep 26, 2019 2:47 pm
@daylen - The assumption/assertion of stationarity is what turns Trinity from a description of the past into a prediction of the future. The disagreement is between whether the future is like or no worse than the past because world wars, depression, etc. (stationarity holds) or it is because declining productivity and population growth trends (stationarity does not hold). These issues are essentially apples and oranges which is why/where the debate above is stuck. A way forward would be to detrend the past data and replace it with a model of the declining megatrends so that one has a dataset of world wars+depression+... with declining macrofactors and then run the MC again.
OMG, I think I completety ignored the population part. It's obvious that, in periods when world population was going through exponential growth (doubling every couple of decades, as it happened in the past two centuries), businesses were going to grow like crazy as well (as the customer base just keeps growing). Now, when the exponential population growth is more or less over, this huge factor will be missing. The growth in company profits/valuations will have to come from increased efficiencies alone.

ertyu
Posts: 2893
Joined: Sun Nov 13, 2016 2:31 am

Re: The 4% Rule – A Castle in the Air

Post by ertyu »


zbigi
Posts: 978
Joined: Fri Oct 30, 2020 2:04 pm

Re: The 4% Rule – A Castle in the Air

Post by zbigi »

ertyu wrote:
Thu Sep 29, 2022 5:54 am
Your Home Country Is Inseparable From Your Withdrawal Rate at Portfolio Charts
Thanks @ertyu, that was a good link.

7Wannabe5
Posts: 9372
Joined: Fri Oct 18, 2013 9:03 am

Re: The 4% Rule – A Castle in the Air

Post by 7Wannabe5 »

zbigi wrote:Now, when the exponential population growth is more or less over, this huge factor will be missing. The growth in company profits/valuations will have to come from increased efficiencies alone.
Not when median global income/spend per capita is still just a bit over half a jacob. IOW, "increased efficiencies alone" won't be the situation until every former peasant farmer's grand-daughter spends her honeymoon in Paris (or becomes adherent of ERE.) Of course, this is ignoring the resource collapse vs extreme innovation necessary for current median income global resident to achieve "American" dream lifestyle with median per capita income/spending of approximately $36,000.

mathiverse
Posts: 788
Joined: Fri Feb 01, 2019 8:40 pm

Re: The 4% Rule – A Castle in the Air

Post by mathiverse »

classical_Liberal wrote:
Wed Sep 28, 2022 9:26 pm
Thank you for the insight!

zbigi
Posts: 978
Joined: Fri Oct 30, 2020 2:04 pm

Re: The 4% Rule – A Castle in the Air

Post by zbigi »

7Wannabe5 wrote:
Thu Sep 29, 2022 8:39 am
Not when median global income/spend per capita is still just a bit over half a jacob. IOW, "increased efficiencies alone" won't be the situation until every former peasant farmer's grand-daughter spends her honeymoon in Paris (or becomes adherent of ERE.) Of course, this is ignoring the resource collapse vs extreme innovation necessary for current median income global resident to achieve "American" dream lifestyle with median per capita income/spending of approximately $36,000.
I don't think we disagree. GDP per capita will grow, thanks to aforementioned gradual increase in efficiencies, which means people's standard of living will increase over time (ignoring collapse arguments and scenarios for a second). The peasant's granddaughter might just be able to go to Paris for her honeymoon thanks to that. However, the resulting growth of companies sales and profits (which is what counts for the SWR), will be slower than when it was driven by both per-capita growth *and* overall population growth.

BTW, this interesting interview touches this and similar subjects (the guy goes deeper into analysing demographic cohorts, which makes the picture much more bleak): https://www.youtube.com/watch?v=wRT7P-VKM0k
Last edited by zbigi on Thu Sep 29, 2022 10:17 am, edited 1 time in total.

sky
Posts: 1726
Joined: Tue Jan 04, 2011 2:20 am

Re: The 4% Rule – A Castle in the Air

Post by sky »

Zeihan is a prophet of the collapse of globalism, leading to mass hunger in Asia and a return to fighting over resources in Europe. The only optimism is that North American countries (and a few more areas) have enough resources to continue on with economic growth after decade or so of problems related to the end of globalisation.

chenda
Posts: 3289
Joined: Wed Jun 29, 2011 1:17 pm
Location: Nether Wallop

Re: The 4% Rule – A Castle in the Air

Post by chenda »

sky wrote:
Thu Sep 29, 2022 11:21 am
(and a few more areas)
Where are these ?

sky
Posts: 1726
Joined: Tue Jan 04, 2011 2:20 am

Re: The 4% Rule – A Castle in the Air

Post by sky »

Argentina, Turkey, Australia, France, Nigeria

Based on adequate resource availability and ability to self produce food.

prudentelo
Posts: 173
Joined: Sat Jan 22, 2022 8:55 am

Re: The 4% Rule – A Castle in the Air

Post by prudentelo »

WFJ wrote:
Tue Sep 20, 2022 12:53 pm
The reason why a 50-year time period fails exponentially more than a 30-year time period is the probability of having a run of negative returns is exponentially higher (not linear) in a 50-year time period than a 30-year time period. This is why a 1% WR for 50 years fails about as many times as a 4% WR for 30 years. I suspect even 1% WR will fail in a 100-to-500-year time period due to this dynamic (similar to above video analysis).
Robert Shiller provides the data for (from the memory) 130 year and several people implemented this in monte carlo simulations and simply do not see dynamic you describe, instead seeing logarithmic decay of SWR to 3.x % regardless length of retirement (30yr is bit different to 50yr but 50 yr is close to infinity yr).

Result you describe seems to be only found by you. Would you be willing to write up full argument with supporting data?

WFJ
Posts: 416
Joined: Sat Apr 24, 2021 11:32 am

Re: The 4% Rule – A Castle in the Air

Post by WFJ »

Probably not short at all... It's a statistical argument revealing the probability of runs being significantly higher the longer the duration of the time period. The Taleb video (below) "Why you always lose a favorable bet" is the best I've seen that semi-explains this concept where even very favorable bets lose as the time period increases.

In my basic simulations, what fails all retirement SWR for a 50-year time period (except below 1%) is a run of 4 out of 5 years, 5 out of 7 years or longer of non-positive returns (not the WWII-Famine-Peak Oil-Climate Change-down 90% hyperboles). Runs in nature are exponentially more common than human minds (without training) can estimate. Logic reveals the probability of longer runs in a 50-year time period is exponentially higher than runs in a 30-year time period.

A good ice breaker exercise that demonstrates this (which is fun for stats people) is in a group of 20+ people, break them into 4-10 groups and instruct half the groups to do a real coin flip 20 times and the other half of the groups to simulate 20-coin flips and record the results on a board. The presenter leaves the room, let's the groups do the real coin flips and simulated coin flips, then the presenter returns and will identify which groups were real and which groups were simulated either by magic, mind reading, or stats. The real coin flips always contain significantly longer runs than the simulated ones (assuming there isn't a group of mostly statisticians). I've done this 20+ times and the rule always holds as humans are unable to recognize the high likelihood of runs in even a 20-trial run. 50 trials is even more pronounced where there will be more and longer runs when compared to a 20-trial run (or 30 trial run in the 4% rule paper). One can do this experiment themselves, you can do a few real 30 coinflips trials and a few with 50 trials and the runs in the 50 trials will be longer and more frequent than in the 30 trials.

All SWR estimates are playing a "run avoidance game". I've been gambling almost all my life, placed my first bet at a horse track with my father around 6-7 years old, went to the dog tracks with family/friends, played blackjack, poker, sports betting nearly all my life and have been to Nevada 30+ times for gambling and have a lot of formal stats training. Recognizing odds comes quite naturally and can spot weaknesses by instinct well before I can model it. 60/40 for 50 years is almost a coinflip exercise where failures can happen at any time.

https://www.youtube.com/watch?v=91IOwS0gf3g

WFJ
Posts: 416
Joined: Sat Apr 24, 2021 11:32 am

Re: The 4% Rule – A Castle in the Air

Post by WFJ »

zbigi wrote:
Tue Sep 27, 2022 3:06 pm
In trading, backtesting a trading algorithm (running it against historical data and seeing if it makes or loses money) is not considered a convincing proof that the algorithm is a viable money-maker. That is because, the future is often very different than the past (plus, the algorithm can be overfit: https://en.wikipedia.org/wiki/Overfitting).
It is difficult to get a man to understand something when his salary depends upon his not understanding it.

Upton Sinclair (but also other authors)

Not just in trading, but all big data exercises. It is CRAZY easy to find whatever one wants to with large data sets over longer time periods with a few suspect data manipulation techniques and many control variables (stepwise regression is a common tool).

The best book I've read that identified this danger is "Cult of Statistical Significance" that plagues "research" today.

This most recently reared its ugly head in the COVID nightmare with junk stats coming from IHME and has been obvious with Population Bomb, Peak Oil, Global Warming, Climate Change, and Sharknado. When someone is paid to produce research finding imminent doom, they will produce research finding imminent doom, it's that easy to manipulate most "research scientists" for a small stipend. The IPSS, International Panel of Sharknado Scientists, will always find an increase in probability of doom if that is what donors want them to find or increases donations from ignorant people or governments.

In my current position, the basics are the entire job is based on the ability to go along with "2+2=Khat" and don't have a job if one finds "2+2= 4, four or IV". I've been FIRE for years and can afford to not go along with "2+2=Khat" fraud, but most if not all of my colleagues can't and are forced to either turn the other way or swallow their pride and go along with the obvious fraud. I filed a whistleblower, but nobody cares as benefits are concentrated and victims are diffuse and lagged.

Some of the stats coming out of IHME violated the first few chapters of "Stats for Dummies" but were used as some justification of lockdowns and other command and control measures. When I looked up who was in IHME, it was mostly "Political Science/Anthropology Econ" profs who had marginal or terrible stats training, but were paid to find "COVID was going to kill millions" and just put out what they were paid to do. I assume there were a few honest people who stood up and identified the obvious fraud, were just voted off the island and lost their stipends or everyone just understood this was a fraud from the beginning.

The link used to show who was making these paid for findings with manipulated stats, but now requires a login. When no academic will put their name to a finding, it's 100% false.
https://www.healthdata.org/about/team

IlliniDave
Posts: 3845
Joined: Wed Apr 02, 2014 7:46 pm

Re: The 4% Rule – A Castle in the Air

Post by IlliniDave »

WFJ wrote:
Fri Sep 30, 2022 4:04 pm
... This most recently reared its ugly head in
I had a rare moment of good internet yesterday evening and had a chance to listen to my favorite podcasters for the first time since early June. Coincidentally, they spent a lot of time in their most recent episode discussing the same issue and the rickety "science" that underpinned questionable (especially in hindsight) policy decisions and the general disinterest in follow-up when things didn't work out the way it was supposed to. They are both scientists but went so far as to say we are in a new scientific Dark Age. Dunno if I agree with that, but they made a good case.

While caveat emptor should be the guiding principle when looking at data investment managers and others that stand to profit produce, the prior point about fooling ourselves is a good caution as well. For the record I think the Trinity Study is a fine study for what it is. It's much more important in my view to understand its limitations than to memorize its conclusions.

User avatar
Slevin
Posts: 626
Joined: Tue Sep 01, 2015 7:44 pm
Location: Sonoma County

Re: The 4% Rule – A Castle in the Air

Post by Slevin »

WFJ wrote:
Fri Sep 30, 2022 4:04 pm


Some of the stats coming out of IHME violated the first few chapters of "Stats for Dummies" but were used as some justification of lockdowns and other command and control measures. When I looked up who was in IHME, it was mostly "Political Science/Anthropology Econ" profs who had marginal or terrible stats training, but were paid to find "COVID was going to kill millions" and just put out what they were paid to do. I assume there were a few honest people who stood up and identified the obvious fraud, were just voted off the island and lost their stipends or everyone just understood this was a fraud from the beginning.

The link used to show who was making these paid for findings with manipulated stats, but now requires a login. When no academic will put their name to a finding, it's 100% false.
https://www.healthdata.org/about/team
So, even if there was a violation of some stats rules, looking at the latest numbers from the WHO, US death toll from COVID is at almost 1.05 Million. Global death tolls are around 6.5 Million, or getting fairly close to .1% of the global population. So, COVID has killed millions, even locally. So while they may have had statistical errors, they seem to have come out to a fairly correct conclusion (and I assume they were running numbers on a "no response" sort of scenario, which isn't what did happen). This in fact seems to be an exact counter argument for the point you are trying to make. Or am I misinterpreting something?
Last edited by Slevin on Sat Oct 01, 2022 8:56 am, edited 1 time in total.

OutOfTheBlue
Posts: 295
Joined: Thu Mar 03, 2022 9:59 am

Re: The 4% Rule – A Castle in the Air

Post by OutOfTheBlue »

Although 6.5 millions is more like close to 0.1% of the global population, not 1%.

Still, millions.

Post Reply