The Flexible Retirement Planner

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
calqueuelater
Posts: 33
Joined: Fri Feb 03, 2012 9:58 pm

The Flexible Retirement Planner

Post by calqueuelater »

Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
The Sensible Steward
Dinosaur Dad
Posts: 233
Joined: Wed Feb 22, 2017 5:05 pm
Location: Connecticut

Re: The Flexible Retirement Planner

Post by Dinosaur Dad »

willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
willthrill81: I have played with a bunch of retirement calculators including those at Fidelity, I-orp, Personal Capital, New Retirement, etc. Are you saying that this drawback is inherent in ALL Monte Carlo simulations (i.e., the underlying math)? Is there a specific piece of research you've found on this? Pretty significant issue for those of us using this output in planning.
"Take calculated risks - that is quite different from being rash." | General George S. Patton
Dinosaur Dad
Posts: 233
Joined: Wed Feb 22, 2017 5:05 pm
Location: Connecticut

Re: The Flexible Retirement Planner

Post by Dinosaur Dad »

Dinosaur Dad wrote: Wed Jan 17, 2018 10:57 am
willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
willthrill81: I have played with a bunch of retirement calculators including those at Fidelity, I-orp, Personal Capital, New Retirement, etc. Are you saying that this drawback is inherent in ALL Monte Carlo simulations (i.e., the underlying math)? Is there a specific piece of research you've found on this? Pretty significant issue for those of us using this output in planning.
OK duh I see you linked the article. Apologies.
"Take calculated risks - that is quite different from being rash." | General George S. Patton
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

Dinosaur Dad wrote: Wed Jan 17, 2018 10:57 am
willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
willthrill81: I have played with a bunch of retirement calculators including those at Fidelity, I-orp, Personal Capital, New Retirement, etc. Are you saying that this drawback is inherent in ALL Monte Carlo simulations (i.e., the underlying math)? Is there a specific piece of research you've found on this? Pretty significant issue for those of us using this output in planning.
I think there are a handful of MC simulators out there that do incorporate mean reversion, but I cannot recall them off the top of my head.
The Sensible Steward
User avatar
corn18
Posts: 2867
Joined: Fri May 22, 2015 6:24 am

Re: The Flexible Retirement Planner

Post by corn18 »

I have been using FRP for 3 years now. It's a tool that I like and it is very flexible. I find it useful for modeling many different scenarios to stress test my retirement plan. (SS, pension, pay off house, higher inflation, different returns, varying savings rates, etc...) It is the only model I have that accounts for everything in one place, which is nice. I don't use Firecalc as I feel it is redundant to FRP. As I get closer to retirement, I refine the inputs and monitor the outputs.

I also have a spreadsheet I use that is less complex.

And I have a simple 25 x expenses number.

The mosaic of the three inputs reveals that all of them are close enough to validate my approach to retirement savings.
Consistently sets low goals and fails to achieve them.
2015
Posts: 2906
Joined: Mon Feb 10, 2014 1:32 pm

Re: The Flexible Retirement Planner

Post by 2015 »

I'm not sure everyone agrees with the Kitces post. See this thread:

viewtopic.php?f=10&t=222695&sid=ef201ab ... 132cfc0c29
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

2015 wrote: Wed Jan 17, 2018 11:09 am I'm not sure everyone agrees with the Kitces post. See this thread:

viewtopic.php?f=10&t=222695&sid=ef201ab ... 132cfc0c29
When it comes to investing, 100% of the people don't agree on pretty much anything.

Below is an argument I have in favor of the Kitces post.
willthrill81 wrote: Wed Jul 05, 2017 6:39 pmThere is a substantial difference between historical performance and most MC simulations. 6.3% of the MC simulations being worse than 100% of the historical periods is a major statistical anomaly. Given that history represents past reality, many, including me, give it more credence than most MC simulations. Of course it's possible that the worst years the stock market will ever see are in front of us, but we don't know that with any certainty. Some of the MC trials look like end-of-the-world scenarios (i.e. 30 nearly straight years of declining stocks) that would be virtually impossible to guard against.
Further, mean reversion is a well documented phenomenon across time and geography and should be incorporated into any serious projection of future potential returns.
The Sensible Steward
2015
Posts: 2906
Joined: Mon Feb 10, 2014 1:32 pm

Re: The Flexible Retirement Planner

Post by 2015 »

willthrill81 wrote: Wed Jan 17, 2018 11:11 am
2015 wrote: Wed Jan 17, 2018 11:09 am I'm not sure everyone agrees with the Kitces post. See this thread:

viewtopic.php?f=10&t=222695&sid=ef201ab ... 132cfc0c29
When it comes to investing, 100% of the people don't agree on pretty much anything.

...
willthrill81 wrote: Wed Jul 05, 2017 6:39 pm
Well I agree 100% with your statement above! :D. My thinking mirrors that of Rob S in the comments section of the Kitces post:
A very interesting study. I think the moral here is that we shouldn’t expect these models to predict the range of outcomes with much accuracy. We are taking historical economic data that is influenced by countless sources including tough decisions by the Federal Reserve and compressing that into two numbers (mean and variance). The author has some useful guidance on how the Monte Carlo outcomes appear to deviate from the historical record that is being modeled. Every simulation result should be interpreted with a grain of salt in the context of how reliable the model is.

There is a tendency for modelers to fall in love with their models and expect more than is reasonable from them. The statistician George Box put it best: “All models are wrong, but some are useful.”
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
I have used it for about three years and like it very much. I love the ability to change parameters at a granular level and to change things like return and inflation rates over specific periods of time (like having one return rate during accumulation phase and another one during retirement, reflecting a more conservative asset allocation.)

It's a powerful tool, but again it's only as good as its assumptions (which are yours). As for whether, in general, MC simulators are too conservative...that may be so. But it's much better than the opposite. I think it is a very useful guide to give one a range of possibilities. One of the things I like best is the "Goal Seek" section, which models how much you need to save to retire with a certain future withdrawal rate; what rate of return you need to meet a specific spending/portfolio target, and so on.
User avatar
Topic Author
calqueuelater
Posts: 33
Joined: Fri Feb 03, 2012 9:58 pm

Re: The Flexible Retirement Planner

Post by calqueuelater »

Thanks to all for their replies, particularly Admiral and corn18. They are helpful and appreciated.
User avatar
JoeRetire
Posts: 15381
Joined: Tue Jan 16, 2018 1:44 pm

Re: The Flexible Retirement Planner

Post by JoeRetire »

2015 wrote: Wed Jan 17, 2018 11:09 am I'm not sure everyone agrees...
This is pretty much always correct.
This isn't just my wallet. It's an organizer, a memory and an old friend.
User avatar
WoodSpinner
Posts: 3504
Joined: Mon Feb 27, 2017 12:15 pm

Re: The Flexible Retirement Planner

Post by WoodSpinner »

calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
OP,

I have been using it for about a year and have built some very detailed models for my retirement planning. Personally, I think it works really well and gives you a great amount of control.

Of course, it’s up to you to figure out the input, but the tool has worked well for me.

Did you have any specific questions?

WoodSpinner 8-)
WoodSpinner
User avatar
Topic Author
calqueuelater
Posts: 33
Joined: Fri Feb 03, 2012 9:58 pm

Re: The Flexible Retirement Planner

Post by calqueuelater »

WoodSpinner wrote: Wed Jan 17, 2018 2:43 pm
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
OP,

I have been using it for about a year and have built some very detailed models for my retirement planning. Personally, I think it works really well and gives you a great amount of control.

Of course, it’s up to you to figure out the input, but the tool has worked well for me.

Did you have any specific questions?

WoodSpinner 8-)
No, no specific questions. Have enjoyed using it and the different ways to model and experiment with scenarios. Was really just wanting to confirm that it was a quality product and a reliable tool (given appropriate inputs). Thanks again to all.
2015
Posts: 2906
Joined: Mon Feb 10, 2014 1:32 pm

Re: The Flexible Retirement Planner

Post by 2015 »

JoeRetire wrote: Wed Jan 17, 2018 2:35 pm
2015 wrote: Wed Jan 17, 2018 11:09 am I'm not sure everyone agrees...
This is pretty much always correct.
Yep. I was simply--and not too effectively--attempting to point out an earlier, interesting thread on the noted Kitces article. One of my heuristics is to always attempt to combat insidious and ubiquitous confirmation bias. The earlier BH thread provides some interesting perspectives if one is seeking additional POV's.
Diogenes
Posts: 602
Joined: Sat Mar 03, 2012 2:58 pm

Re: The Flexible Retirement Planner

Post by Diogenes »

Just tried to use it (downloaded version) on a Mac. It prompted some unusual security certificate warnings and a Java warning about the product requesting 'unrestricted access' which the warning said would also include the webcam and microphone.
Not sure why that would be, but it caused me to decide not to run it.
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

Diogenes wrote: Tue Feb 27, 2018 11:41 am Just tried to use it (downloaded version) on a Mac. It prompted some unusual security certificate warnings and a Java warning about the product requesting 'unrestricted access' which the warning said would also include the webcam and microphone.
Not sure why that would be, but it caused me to decide not to run it.
I have never gotten those warnings (running High Sierra) but MacOS in general does not like Java. It's perfectly safe I've used it for years.

Post your issue here and the owner will respond, he's very active:

http://www.flexibleretirementplanner.com/phpBB3/
User avatar
Monster99
Posts: 628
Joined: Tue Aug 29, 2017 10:28 am

Re: The Flexible Retirement Planner

Post by Monster99 »

WoodSpinner wrote: Wed Jan 17, 2018 2:43 pm
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
OP,

I have been using it for about a year and have built some very detailed models for my retirement planning. Personally, I think it works really well and gives you a great amount of control.

Of course, it’s up to you to figure out the input, but the tool has worked well for me.

Did you have any specific questions?

WoodSpinner 8-)
Will the tool model RMD withdrawals from the tax defered assets? Or do you need to imput each withdrawal?
User avatar
corn18
Posts: 2867
Joined: Fri May 22, 2015 6:24 am

Re: The Flexible Retirement Planner

Post by corn18 »

Monster99 wrote: Tue Feb 27, 2018 9:28 pm
Will the tool model RMD withdrawals from the tax defered assets? Or do you need to imput each withdrawal?
FRP models RMD's for you.
Consistently sets low goals and fails to achieve them.
User avatar
WoodSpinner
Posts: 3504
Joined: Mon Feb 27, 2017 12:15 pm

Re: The Flexible Retirement Planner

Post by WoodSpinner »

calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
I used extensively for modeling my plans and understanding risk. I ended up putting detailed entries to build the model.

Do you have specific questions? Or just seeing if anyone is using it?

I think using a deterministic model (Excel) and a Probabalistic model (FRP) are a good approach for gaining an understanding of retirement .

WoodSpinner
WoodSpinner
User avatar
AtlasShrugged?
Posts: 699
Joined: Wed Jul 15, 2015 6:08 pm

Re: The Flexible Retirement Planner

Post by AtlasShrugged? »

I have used it for about three years and like it very much.
I used extensively for modeling my plans and understanding risk. I ended up putting detailed entries to build the model.
Woodspinner, Admiral....Is it worth getting the license key? Or is the freeware version with three scenarios good enough? Does the license key unlock further capabilities not present in the freeware version?
“If you don't know, the thing to do is not to get scared, but to learn.”
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

AtlasShrugged? wrote: Wed Feb 28, 2018 12:26 pm
I have used it for about three years and like it very much.
I used extensively for modeling my plans and understanding risk. I ended up putting detailed entries to build the model.
Woodspinner, Admiral....Is it worth getting the license key? Or is the freeware version with three scenarios good enough? Does the license key unlock further capabilities not present in the freeware version?
I use the freeware version and have found no limitations. I think it's free for non-commercial use. I have had more than three scenarios modeled at once (I think) so I'm not even sure about that limitation...
User avatar
AtlasShrugged?
Posts: 699
Joined: Wed Jul 15, 2015 6:08 pm

Re: The Flexible Retirement Planner

Post by AtlasShrugged? »

I use the freeware version and have found no limitations. I think it's free for non-commercial use.
Admiral...I have the freeware version as well. I know it is a hell of a tool, just having a hard time grasping all the different aspects, and fully use the functionality of FRP.
“If you don't know, the thing to do is not to get scared, but to learn.”
User avatar
WoodSpinner
Posts: 3504
Joined: Mon Feb 27, 2017 12:15 pm

Re: The Flexible Retirement Planner

Post by WoodSpinner »

AtlasShrugged? wrote: Wed Feb 28, 2018 12:26 pm
I have used it for about three years and like it very much.
I used extensively for modeling my plans and understanding risk. I ended up putting detailed entries to build the model.
Woodspinner, Admiral....Is it worth getting the license key? Or is the freeware version with three scenarios good enough? Does the license key unlock further capabilities not present in the freeware version?
I got the license key ($20??)—it was a great investment and a good way to pay some back for all of the hard work it took to write and support the tool. I had 11 models going at my max — variations looking at early deaths, stock market crashes, really high inflation etc.

Ended up keeping 3 models up to date....

It’s pretty powerful, but you have to know what to enter and how to interpret. Their support forum is excellent.

Joel
WoodSpinner
jmk
Posts: 643
Joined: Tue Nov 01, 2011 7:48 pm

Re: The Flexible Retirement Planner

Post by jmk »

I like it a lot. Best feature is the sensitivity analysis. I like how easy it is to check scenarios (as simple as checking or unchecking a box.) Biggest problems are (1) only normal distribution for monte carlo, (2) limited number of withdrawal strategies, (3) cruder tax modeling than other calculators.
Last edited by jmk on Fri May 24, 2019 10:40 am, edited 2 times in total.
esteen
Posts: 439
Joined: Thu May 23, 2019 12:31 am

Re: The Flexible Retirement Planner

Post by esteen »

I'm thinking of downloading the FRP for personal use but how does it compare to Portfolio Analyzer's Monte Carlo -> Financial Goals (multistage) feature? Is it far more robust, or not much difference? Would love to hear from people who have tried both.

Sorry if this compare has already been discussed but my quick forum search did not turn up anything.

Thanks!
This post is for entertainment or information only, and should not be construed as professional financial advice. | | "Invest your money passively and your time actively" -Michael LeBoeuf
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: The Flexible Retirement Planner

Post by Beliavsky »

willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise.
You could fit an autoregressive model to past stock returns. If the AR coefficients are negative, for example an AR(3) model

r(t) = e(t) + c1*r(t-1) + c2*r(t-2) + c3*r(t-3)

where c1 = c2 = c3 = -0.05,

you can simulate it to model a mean-reverting stock market. But if you believe mean reversion in the stock market can be quantified well enough to simulate it, you are saying that returns are somewhat predictable, which raises the question of why you should have a static asset allocation.
deskpilot65
Posts: 10
Joined: Fri Mar 08, 2019 6:28 pm

Re: The Flexible Retirement Planner

Post by deskpilot65 »

I use FRP extensively for all of my planning. I find the MC analysis to be the least important feature. For me, the strength of the tool is in the ability to model so many cash flow scenarios at different life stages - i.e. pre/post retirement, pre/post medicare age, pre/post SS, inheritance, RMDs... and then with a few clicks you can add/remove different what-if scenarios, like obamacare repeal, SS haircut, major house repair, etc...
The sensitivity analysis feature lets you graphically see the effect of things like portfolio return or retirement age on success rates and spending levels.
klrjaa
Posts: 65
Joined: Sat Mar 23, 2019 10:55 am

Re: The Flexible Retirement Planner

Post by klrjaa »

deskpilot65 wrote: Thu May 23, 2019 6:16 pm I use FRP extensively for all of my planning. I find the MC analysis to be the least important feature. For me, the strength of the tool is in the ability to model so many cash flow scenarios at different life stages - i.e. pre/post retirement, pre/post medicare age, pre/post SS, inheritance, RMDs... and then with a few clicks you can add/remove different what-if scenarios, like obamacare repeal, SS haircut, major house repair, etc...
The sensitivity analysis feature lets you graphically see the effect of things like portfolio return or retirement age on success rates and spending levels.
+ 1. Tremendous capabilities. I took my .frp file and sent to a financial advisor who was intimately familiar with the tool. I wouldn't use a CFP that was not totally familiar with it
targetconfusion
Posts: 90
Joined: Sun Aug 12, 2018 1:02 pm

Re: The Flexible Retirement Planner

Post by targetconfusion »

Beliavsky wrote: Thu May 23, 2019 6:01 am You could fit an autoregressive model to past stock returns. If the AR coefficients are negative, for example an AR(3) model

r(t) = e(t) + c1*r(t-1) + c2*r(t-2) + c3*r(t-3)

where c1 = c2 = c3 = -0.05,

you can simulate it to model a mean-reverting stock market. But if you believe mean reversion in the stock market can be quantified well enough to simulate it, you are saying that returns are somewhat predictable, which raises the question of why you should have a static asset allocation.
+1
Without disputing @willthrill81's conclusion that this (and maybe other) sim tools are overly conservative, it does seem hard to know a priori the right amount of mean reversion to assume. It's been pointed out by smarter people than me on this forum that every "worst case scenario" of the past was worse than preceding worst case scenario. To me, that suggests it's reasonable to expect things in the future could break worse than they ever have in the past and, thus, not to automatically reject a model allowing it as a possibility.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

targetconfusion wrote: Fri May 24, 2019 1:40 pm
Beliavsky wrote: Thu May 23, 2019 6:01 am You could fit an autoregressive model to past stock returns. If the AR coefficients are negative, for example an AR(3) model

r(t) = e(t) + c1*r(t-1) + c2*r(t-2) + c3*r(t-3)

where c1 = c2 = c3 = -0.05,

you can simulate it to model a mean-reverting stock market. But if you believe mean reversion in the stock market can be quantified well enough to simulate it, you are saying that returns are somewhat predictable, which raises the question of why you should have a static asset allocation.
+1
Without disputing @willthrill81's conclusion that this (and maybe other) sim tools are overly conservative, it does seem hard to know a priori the right amount of mean reversion to assume. It's been pointed out by smarter people than me on this forum that every "worst case scenario" of the past was worse than preceding worst case scenario. To me, that suggests it's reasonable to expect things in the future could break worse than they ever have in the past and, thus, not to automatically reject a model allowing it as a possibility.
It's quite possible that the worst declines in stocks are in front of us, though it would be difficult to have a bigger drawdown than the -89% of the Great Depression. But where we have evidence that simulations without some kind of mean reversion produce tails, both positive and negative, that are too 'fat' is when we compare their results to that of history. If we assume historical returns to be a representative sample of all possible returns, then simulations produce extreme sequences of returns at a rate that is statistically greater than what history says that they should. Statistics cannot prove anything, but they can provide evidence of effects.

If one isn't able or willing to use a simulation that incorporates mean reversion, an alternative possibility is to simply 'throw out' something like the bottom and top 5% of return sequences it produces.
The Sensible Steward
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: The Flexible Retirement Planner

Post by Beliavsky »

willthrill81 wrote: Fri May 24, 2019 2:26 pm
targetconfusion wrote: Fri May 24, 2019 1:40 pm
Beliavsky wrote: Thu May 23, 2019 6:01 am You could fit an autoregressive model to past stock returns. If the AR coefficients are negative, for example an AR(3) model

r(t) = e(t) + c1*r(t-1) + c2*r(t-2) + c3*r(t-3)

where c1 = c2 = c3 = -0.05,

you can simulate it to model a mean-reverting stock market. But if you believe mean reversion in the stock market can be quantified well enough to simulate it, you are saying that returns are somewhat predictable, which raises the question of why you should have a static asset allocation.
+1
Without disputing @willthrill81's conclusion that this (and maybe other) sim tools are overly conservative, it does seem hard to know a priori the right amount of mean reversion to assume. It's been pointed out by smarter people than me on this forum that every "worst case scenario" of the past was worse than preceding worst case scenario. To me, that suggests it's reasonable to expect things in the future could break worse than they ever have in the past and, thus, not to automatically reject a model allowing it as a possibility.
It's quite possible that the worst declines in stocks are in front of us, though it would be difficult to have a bigger drawdown than the -89% of the Great Depression. But where we have evidence that simulations without some kind of mean reversion produce tails, both positive and negative, that are too 'fat' is when we compare their results to that of history. If we assume historical returns to be a representative sample of all possible returns, then simulations produce extreme sequences of returns at a rate that is statistically greater than what history says that they should. Statistics cannot prove anything, but they can provide evidence of effects.
You have advocated trend-following strategies. Do you think mean reversion operates on a longer time horizon and momentum on a shorter one?
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

Beliavsky wrote: Fri May 24, 2019 2:33 pm
willthrill81 wrote: Fri May 24, 2019 2:26 pm
targetconfusion wrote: Fri May 24, 2019 1:40 pm
Beliavsky wrote: Thu May 23, 2019 6:01 am You could fit an autoregressive model to past stock returns. If the AR coefficients are negative, for example an AR(3) model

r(t) = e(t) + c1*r(t-1) + c2*r(t-2) + c3*r(t-3)

where c1 = c2 = c3 = -0.05,

you can simulate it to model a mean-reverting stock market. But if you believe mean reversion in the stock market can be quantified well enough to simulate it, you are saying that returns are somewhat predictable, which raises the question of why you should have a static asset allocation.
+1
Without disputing @willthrill81's conclusion that this (and maybe other) sim tools are overly conservative, it does seem hard to know a priori the right amount of mean reversion to assume. It's been pointed out by smarter people than me on this forum that every "worst case scenario" of the past was worse than preceding worst case scenario. To me, that suggests it's reasonable to expect things in the future could break worse than they ever have in the past and, thus, not to automatically reject a model allowing it as a possibility.
It's quite possible that the worst declines in stocks are in front of us, though it would be difficult to have a bigger drawdown than the -89% of the Great Depression. But where we have evidence that simulations without some kind of mean reversion produce tails, both positive and negative, that are too 'fat' is when we compare their results to that of history. If we assume historical returns to be a representative sample of all possible returns, then simulations produce extreme sequences of returns at a rate that is statistically greater than what history says that they should. Statistics cannot prove anything, but they can provide evidence of effects.
You have advocated trend-following strategies. Do you think mean reversion operates on a longer time horizon and momentum on a shorter one?
I have never advocated trend-following; I've merely discussed its pros and cons along with that of buy-and-hold along with my own experiences with trend following.

The historic record for U.S. stocks is very clear that holding stocks over longer periods has reduced the variation in total returns. For instance, the standard deviation of 5 year real returns was 7.92% as opposed to 5.13% for 10 year returns, 2.96% for 20 year returns, and 1.67% for 30 year returns. Similarly, the range of returns has shrunk as the holding period increased.
The Sensible Steward
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: The Flexible Retirement Planner

Post by Beliavsky »

willthrill81 wrote: Fri May 24, 2019 2:44 pm The historic record for U.S. stocks is very clear that holding stocks over longer periods has reduced the variation in total returns. For instance, the standard deviation of 5 year real returns was 7.92% as opposed to 5.13% for 10 year returns, 2.96% for 20 year returns, and 1.67% for 30 year returns. Similarly, the range of returns has shrunk as the holding period increased.
The range of annualized returns may shrink, but the range of cumulative (non-annualized) returns does not. There are not too many non-overlapping 30-year periods of stock market data, so the standard errors of the standard deviation of 30-year annualized real returns is high. I'm not saying that mean reversion does not exist, just that there is not that much data available to estimate its magnitude.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

Beliavsky wrote: Fri May 24, 2019 2:55 pm
willthrill81 wrote: Fri May 24, 2019 2:44 pm The historic record for U.S. stocks is very clear that holding stocks over longer periods has reduced the variation in total returns. For instance, the standard deviation of 5 year real returns was 7.92% as opposed to 5.13% for 10 year returns, 2.96% for 20 year returns, and 1.67% for 30 year returns. Similarly, the range of returns has shrunk as the holding period increased.
The range of annualized returns may shrink, but the range of cumulative (non-annualized) returns does not. There are not too many non-overlapping 30-year periods of stock market data, so the standard errors of the standard deviation of 30-year annualized real returns is high. I'm not saying that mean reversion does not exist, just that there is not that much data available to estimate its magnitude.
True, we don't have as much data we would like, but we must work with what we have. We can't wait 800 more years until we have 30 non-overlapping 30 year periods to determine what to do.
The Sensible Steward
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

It is quite simple to ignore high or low values with FRP. For each future year, it reports the median, top 10%, and bottom 10% portfolio value. You can use, or ignore, whatever values you wish. All results are in current dollars.
YRT70
Posts: 1289
Joined: Sat Apr 27, 2019 8:51 am

Re: The Flexible Retirement Planner

Post by YRT70 »

willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
I like your thoughts about MC testing. Does the Vanguard tool have the same problems?
https://retirementplans.vanguard.com/VG ... ggCalc.jsf

I'm also trying to understand why the tool gives higher succes rates to higher bond allocations that historic simulations do. Any thoughts on that?
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

YRT70 wrote: Sat May 25, 2019 11:24 am
willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.

This is a significant drawback because it is now understood that MC analysis overstates the 'tails' of return distributions (i.e. it predicts more very good sequences and very poor sequences than the historic record implies should exist). Consequently, when compared to the historic record, the results of MC analysis will likely lead you to a more conservative conclusion than otherwise. Some view this as a real problem, others argue that an extra measure of conservatism is warranted. It's up to you to determine what to use and how to use it.
I like your thoughts about MC testing. Does the Vanguard tool have the same problems?
https://retirementplans.vanguard.com/VG ... ggCalc.jsf

I'm also trying to understand why the tool gives higher succes rates to higher bond allocations that historic simulations do. Any thoughts on that?
Yes, the Vanguard retirement planning tool suffers from the same issue because, while they are using real historic returns, they are only using one year at a time, which almost entirely negates any mean reversion. It could pick 1929 for one year, 1973 for the next, then 2008, etc.
For each year of each simulation, we randomly select one year of stock, bond, and cash returns from our database. Using those values, we calculate what would happen to your savings - subtracting your spending, adjusting for inflation, and adding your investment return. We repeat this process, one year at a time, until the end of your retirement or until your savings are exhausted.
The Sensible Steward
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: The Flexible Retirement Planner

Post by Beliavsky »

calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
Does the Flexible Retirement Planner allow you to specify the cost basis of assets in your taxable account and charge you capital gains taxes at the correct level when you sell appreciated assets? FRP does look interesting.
User avatar
WoodSpinner
Posts: 3504
Joined: Mon Feb 27, 2017 12:15 pm

Re: The Flexible Retirement Planner

Post by WoodSpinner »

Beliavsky wrote: Sun May 26, 2019 10:06 am
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
Does the Flexible Retirement Planner allow you to specify the cost basis of assets in your taxable account and charge you capital gains taxes at the correct level when you sell appreciated assets? FRP does look interesting.
No it does not. You can model your own tax situation though but you have to provide the necessary input. This is one of the reasons I think a Deterministic model which incorporates your investments, cash flows, and taxes really helps in building the Probabilistic model in FRP.

For instance, I can model a Roth Conversion Strategy in my deterministic model and develop an expected tax rate (Fed + State). This can then be entered into the FRP model as a series of Credits to the IRA accounts, Debits to the Roth accounts, and expenses related to taxes. The probabilistic model can help gain a better understanding of your retirement plan — note, it’s still just a model, not a guaranteed reality.

WoodSpinner
WoodSpinner
jmk
Posts: 643
Joined: Tue Nov 01, 2011 7:48 pm

Re: The Flexible Retirement Planner

Post by jmk »

willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.
One that does is portfolio visualizer. You can have the bootstrap pick a sample of n year units (rather than each year being its own independent sample), thereby modeling rtm to varying degrees.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: The Flexible Retirement Planner

Post by willthrill81 »

jmk wrote: Sun May 26, 2019 4:01 pm
willthrill81 wrote: Wed Jan 17, 2018 10:40 am It seems that the Monte Carlo analysis used in the software does not incorporate mean reversion; few MC tools do.
One that does is portfolio visualizer. You can have the bootstrap pick a sample of n year units (rather than each year being its own independent sample), thereby modeling rtm to varying degrees.
Yes, thanks for pointing that out. PV really is a fantastic tool, and it's pretty amazing that it's free.
The Sensible Steward
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

WoodSpinner wrote: Sun May 26, 2019 11:22 am
Beliavsky wrote: Sun May 26, 2019 10:06 am
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
Does the Flexible Retirement Planner allow you to specify the cost basis of assets in your taxable account and charge you capital gains taxes at the correct level when you sell appreciated assets? FRP does look interesting.
No it does not. You can model your own tax situation though but you have to provide the necessary input. This is one of the reasons I think a Deterministic model which incorporates your investments, cash flows, and taxes really helps in building the Probabilistic model in FRP.

For instance, I can model a Roth Conversion Strategy in my deterministic model and develop an expected tax rate (Fed + State). This can then be entered into the FRP model as a series of Credits to the IRA accounts, Debits to the Roth accounts, and expenses related to taxes. The probabilistic model can help gain a better understanding of your retirement plan — note, it’s still just a model, not a guaranteed reality.

WoodSpinner
RPM is a useful tool for these types of granular/data-driven calculations. I find that RPM and FRP tend to produce similar results (though one is current and the other inflated dollars).
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: The Flexible Retirement Planner

Post by Beliavsky »

Admiral wrote: Tue May 28, 2019 9:06 am
WoodSpinner wrote: Sun May 26, 2019 11:22 am
Beliavsky wrote: Sun May 26, 2019 10:06 am
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
Does the Flexible Retirement Planner allow you to specify the cost basis of assets in your taxable account and charge you capital gains taxes at the correct level when you sell appreciated assets? FRP does look interesting.
No it does not. You can model your own tax situation though but you have to provide the necessary input. This is one of the reasons I think a Deterministic model which incorporates your investments, cash flows, and taxes really helps in building the Probabilistic model in FRP.

For instance, I can model a Roth Conversion Strategy in my deterministic model and develop an expected tax rate (Fed + State). This can then be entered into the FRP model as a series of Credits to the IRA accounts, Debits to the Roth accounts, and expenses related to taxes. The probabilistic model can help gain a better understanding of your retirement plan — note, it’s still just a model, not a guaranteed reality.

WoodSpinner
RPM is a useful tool for these types of granular/data-driven calculations. I find that RPM and FRP tend to produce similar results (though one is current and the other inflated dollars).
What does RPM stand for?
Admiral
Posts: 5039
Joined: Mon Oct 27, 2014 12:35 pm

Re: The Flexible Retirement Planner

Post by Admiral »

Beliavsky wrote: Tue May 28, 2019 10:11 am
Admiral wrote: Tue May 28, 2019 9:06 am
WoodSpinner wrote: Sun May 26, 2019 11:22 am
Beliavsky wrote: Sun May 26, 2019 10:06 am
calqueuelater wrote: Wed Jan 17, 2018 8:17 am Good Morning - Longtime lurker and infrequent poster here. I am familiar with many of the free financial tools often referenced on the site (i.e. FIRECalc), but was wondering whether any posters are familiar with The Flexible Retirement Planner? I have been using it to model out different financial scenarios and find it fairly user friendly; don't know if there are any cautions or drawbacks I should be aware of. Any insights would be appreciated.
Does the Flexible Retirement Planner allow you to specify the cost basis of assets in your taxable account and charge you capital gains taxes at the correct level when you sell appreciated assets? FRP does look interesting.
No it does not. You can model your own tax situation though but you have to provide the necessary input. This is one of the reasons I think a Deterministic model which incorporates your investments, cash flows, and taxes really helps in building the Probabilistic model in FRP.

For instance, I can model a Roth Conversion Strategy in my deterministic model and develop an expected tax rate (Fed + State). This can then be entered into the FRP model as a series of Credits to the IRA accounts, Debits to the Roth accounts, and expenses related to taxes. The probabilistic model can help gain a better understanding of your retirement plan — note, it’s still just a model, not a guaranteed reality.

WoodSpinner
RPM is a useful tool for these types of granular/data-driven calculations. I find that RPM and FRP tend to produce similar results (though one is current and the other inflated dollars).
What does RPM stand for?
See here:

https://www.bogleheads.org/wiki/Retiree_Portfolio_Model
HoosierJim
Posts: 987
Joined: Wed Mar 24, 2010 7:11 pm

Re: The Flexible Retirement Planner

Post by HoosierJim »

Reminder.... tonight AT 7 PM central time is the author/developer product demo and Q & A session on the FLEXIBLE RETIREMENT PLANNER

5 pm pacific
6 pm mountain
7 pm central
8 pm eastern

ALL BOGLEHEAD members and lurkers are invited.

FLEXIBLE RETIREMENT PLANNER 3/24/21 at 7pm central time.
You can download an individual version here


Please visit the websites above and try out the products to have the best discussion and questions about your situation.


If you would like to attend the Zoom meeting please contact me via email at chicagovirtualbogleheads@gmail.com.



If you would like to attend the Zoom meeting please contact me via email at chicagovirtualbogleheads@gmail.com or Private Message me with your EMAIL ( Email is easier for me than the PM system - thank you.) or join/and ask your local chapter coordinator.

You can find the local chapters here.

You can also find a calendar of upcoming events here.
Post Reply