FIRECalc et al. - user experience

Have a question about your personal investments? No matter how simple or complex, you can ask it here.
Post Reply
Topic Author
mass_biker
Posts: 208
Joined: Thu Feb 17, 2011 2:53 pm

FIRECalc et al. - user experience

Post by mass_biker » Sun Jan 28, 2018 12:22 pm

I use these types of monte carlo simulators to run various iterations of:

- Early retirement
- Increasing after tax savings
- Portfolio mixes
- Spend down
- Other sources (rental income) and uses (college tuition)

In most cases, and even when stressing the scenarios pretty heavily, I end up with a 100% success rate, and in some cases of the upper end of the bounds/percentiles, some goofily large #s.

Question for the All-Knowing Forum: how have the estimators/calculators worked out for those who have planned ER or portfolio actions based on the simulation? Wondering how much faith to place on these as I think about scenario planning.

Many thanks
m_b

dbr
Posts: 32811
Joined: Sun Mar 04, 2007 9:50 am

Re: FIRECalc et al. - user experience

Post by dbr » Sun Jan 28, 2018 12:48 pm

Some generalizations are:

1. The whole thing is a rough guideline to broad effects. If you are always getting 100% success you probably haven't stressed the system with a high enough rate of spending. It isn't hard to drive FireCalc to 0% success. Asset allocation during retirement has a weak effect on success rates but a strong effect on residual wealth. If you want to drive the result to more failures putting in severely not enough in stocks will do that.

2. The very nature of approaching retirement withdrawal from a safe withdrawal rate point of view is that eliminating the worst cases results in the rest of the outcomes amounting to dying with fabulous wealth. It is an economics fact that a portfolio of volatile assets is a terrible match to a problem of providing a certain income stream. Certain algorithms such as variable percentage withdrawal attempt to ameliorate that problem to a certain degree. I am not sure there is a way to ameliorate the problem of variable returns during accumulation. Luck of history is a severe taskmaster.

skierrex
Posts: 57
Joined: Tue Oct 03, 2017 10:03 pm

Re: FIRECalc et al. - user experience

Post by skierrex » Sun Jan 28, 2018 1:01 pm

As I understand it, firecalc is not a monte carlo analysis tool. Those use probabilities to perform the analysis.
Firecalc uses historical data instead.
I turn up one of the variables (like withdrawal rate) until I get something like 90% success, then vary something else (like AA) to see the effect.
If you believe that the future will not be any worse than the past, it is very informative.

livesoft
Posts: 71894
Joined: Thu Mar 01, 2007 8:00 pm

Re: FIRECalc et al. - user experience

Post by livesoft » Sun Jan 28, 2018 1:06 pm

I think FIRECalc and cfiresim are very conservative.

If you are worried about things, please read the series of 20+ articles starting at:
https://earlyretirementnow.com/2016/12/ ... t-1-intro/
Wiki This signature message sponsored by sscritic: Learn to fish.

Ron Scott
Posts: 1090
Joined: Tue Apr 05, 2016 5:38 am

Re: FIRECalc et al. - user experience

Post by Ron Scott » Sun Jan 28, 2018 1:13 pm

The basic Firecalc engine--which uses historical returns on various investment instruments--asks you to bet the house on the belief that the past hundred years or so can be used to predict the future. This is a horrible proposition in my opinion that simply ends up assuring those who have net worth at retirement between $500k and $4M that they'll be OK.

I operate under the assumption that I have no idea what the future of investment returns will be like and believe the best approach retirement planning is assuming real returns will be much lower than in the past. If this means working longer and spending less, so be it.

I have no doubt Firecalc produces accurate results given the inputs they use. I have no faith in their answers.
Retirement is a game best played by those prepared for more volatility in the future than has been seen in the past. The solution is not to predict investment losses but to prepare for them.

BBQ Nut
Posts: 168
Joined: Sun Jun 22, 2014 9:19 am

Re: FIRECalc et al. - user experience

Post by BBQ Nut » Sun Jan 28, 2018 1:18 pm

skierrex wrote:
Sun Jan 28, 2018 1:01 pm
As I understand it, firecalc is not a monte carlo analysis tool. Those use probabilities to perform the analysis.
Firecalc uses historical data instead.
I turn up one of the variables (like withdrawal rate) until I get something like 90% success, then vary something else (like AA) to see the effect.
If you believe that the future will not be any worse than the past, it is very informative.
Firecalc can do Monte Carlo simulations.

Under the "Your Portfolio" tab, you can opt to run analyses using portfolio of consistent growth or portfolio of random growth - both of these options result in Monte Carlo simulations.

dbr
Posts: 32811
Joined: Sun Mar 04, 2007 9:50 am

Re: FIRECalc et al. - user experience

Post by dbr » Sun Jan 28, 2018 1:32 pm

Ron Scott wrote:
Sun Jan 28, 2018 1:13 pm
I operate under the assumption that I have no idea what the future of investment returns will be like and believe the best approach retirement planning is assuming real returns will be much lower than in the past. If this means working longer and spending less, so be it.
You don't see a contradiction between " I have no idea what the future of investment returns will be like." and "the best approach retirement planning is assuming real returns will be much lower than in the past."? Either you think you know something or you don't. Note that if a person really does think they know how the future will be different from the past then they can modify the results from the past accordingly.

Stonebr
Posts: 1472
Joined: Wed Jan 21, 2009 11:19 am
Location: Maine

Re: FIRECalc et al. - user experience

Post by Stonebr » Sun Jan 28, 2018 1:33 pm

We retired 6.5 years ago and did cash flow modeling on both FireCalc and Otar. I found Jim Otar's spreadsheet more flexible for my needs, and more detailed. Results, however, were fairly similar, giving me confidence that neither was leading me far astray. The free calculators at the various mutual fund sites were fairly useless by comparison. I was a math major in college, and worked in financial services for 30 years, so I actually found this modeling to be fun. Otar's retirement planning book, essentially a manual for using his spreadsheet, was also excellent.

We made our best guess at living expenses in retirement and then added some additional splurges and started testing. Otar calculator was great, showing survivability of various plans and scenarios under best, worst, and median conditions. My wife and I have a healthy lifestyle and decent DNA, so we projected out to the max -- age 100+ for both -- and found an asset allocation and a spending level that worked, at least historically, through thick and thin. Note that Otar's worst case uses a sequence of returns from the Great Depression of the 1930s. Clearly, the last 6.5 years have been nothing like the 1930s, at least in the stock market, so we've dodged a big bullet. Because we are deferring Social Security until 70, these are some of our biggest cash withdrawal years, and sequence of returns has been favorable. Our net worth has risen every year in spite of the drawdown.

You certainly don't need to go through all this, however. A simple rule like 3.5% withdrawal rate will work fine. My testing of 4% showed that it failed too much for my own comfort. It's easy to say that you can adjust spending, but sometimes life gets in the way. So I'd rather not take the chance. In any case, the Otar (and Firecalc) models indicated we could actually spend significantly more in retirement than during our working life. Neither of us is much into consumerism, so our biggest budget item last year was charity.
"have more than thou showest, | speak less than thou knowest" -- The Fool in King Lear

User avatar
WoodSpinner
Posts: 1472
Joined: Mon Feb 27, 2017 1:15 pm

Re: FIRECalc et al. - user experience

Post by WoodSpinner » Sun Jan 28, 2018 1:41 pm

OP,

For Monte Carlo style simulations, I highly recommend, Flexible Retirement Planner and building out you own customized model. It’s not magic, you have to have a solid understanding of the inputs and learn to see the trends in the outputs.

For my planning I started with a deterministic model (Excel, a customized model) and then used that insight as a starting point for the Probabalistic (FRP).

For the immediate term it gave me great confidence in my plan and a set of tools I can use to adapt to unanticipated events (somehow we are really seriously contemplating a Class B RV purchase).

Modeling, decision making, and financial risk analysis were part of my job at the MegaCorp I retired from in Dec2017. I figure if the approach worked for business decisions in the 100’s of billions, it should work for my modest retirement. :greedy

I ended up cross checking my models against i-ORP, cFireSim, FireCalc just to see if there were any major anomalies. Fortunately nothing really stood out that made me rethink the use of one tool over another.

Now we get to see if how well the plan works—still learning and adapting. But now it’s REAL! :beer

WoodSpinner :D

Ron Scott
Posts: 1090
Joined: Tue Apr 05, 2016 5:38 am

Re: FIRECalc et al. - user experience

Post by Ron Scott » Sun Jan 28, 2018 4:43 pm

dbr wrote:
Sun Jan 28, 2018 1:32 pm
Ron Scott wrote:
Sun Jan 28, 2018 1:13 pm
I operate under the assumption that I have no idea what the future of investment returns will be like and believe the best approach retirement planning is assuming real returns will be much lower than in the past. If this means working longer and spending less, so be it.
You don't see a contradiction between " I have no idea what the future of investment returns will be like." and "the best approach retirement planning is assuming real returns will be much lower than in the past."? Either you think you know something or you don't. Note that if a person really does think they know how the future will be different from the past then they can modify the results from the past accordingly.
No. The first statement simply acknowledges the obvious while the second asserts a conservative assumption.

In ny event, do you actually have a point that relates to the Firecalc topic or are you just letting off steam for some reason?
Retirement is a game best played by those prepared for more volatility in the future than has been seen in the past. The solution is not to predict investment losses but to prepare for them.

dbr
Posts: 32811
Joined: Sun Mar 04, 2007 9:50 am

Re: FIRECalc et al. - user experience

Post by dbr » Sun Jan 28, 2018 5:24 pm

Ron Scott wrote:
Sun Jan 28, 2018 4:43 pm
dbr wrote:
Sun Jan 28, 2018 1:32 pm
Ron Scott wrote:
Sun Jan 28, 2018 1:13 pm
I operate under the assumption that I have no idea what the future of investment returns will be like and believe the best approach retirement planning is assuming real returns will be much lower than in the past. If this means working longer and spending less, so be it.
You don't see a contradiction between " I have no idea what the future of investment returns will be like." and "the best approach retirement planning is assuming real returns will be much lower than in the past."? Either you think you know something or you don't. Note that if a person really does think they know how the future will be different from the past then they can modify the results from the past accordingly.
No. The first statement simply acknowledges the obvious while the second asserts a conservative assumption.

In ny event, do you actually have a point that relates to the Firecalc topic or are you just letting off steam for some reason?
It's relevant to the question of how much credence to put in future planning from a program constructed around investment data of the last hundred years. I think that is a genuine dilemma, but the problem is what to do about it. The zero order assumption is that the future will be like past. The refinement is to start with the past and propose an adjustment. Real returns will be lower than in the past is a plausible assumption someone might propose. That assumption is far from saying we have no idea what future investment returns will be. In fact it is a definite statement that we do suspect what future investment returns will be. This causes a problem for Firecalc using historical returns. The program won't work and we have to try other models that allow that adjustment to be made. If we really think we have NO idea what will happen in the future, then our only choice is to walk away with no model at all, which is a completely different thing. Possibly just short of that we can run a model that allows returns to be adjusted and look at a whole set of outcomes for a whole range of returns and think about what it would be like to get one of those rather than another one.

2015
Posts: 2906
Joined: Mon Feb 10, 2014 2:32 pm

Re: FIRECalc et al. - user experience

Post by 2015 » Sun Jan 28, 2018 9:37 pm

Pre-retirement I ran about 876,000 calculators at least monthly. I only retired after factoring in a very large cushion and considering worst case scenarios. In retirement I'm lazy (at least as far as tinkering with investing and my PF goes) and only run Fidelity's retirement calculator, and mostly for amusement.

I've learned--and I sort of knew this pre-retirement--I was overly conservative pre-retirement. This has not changed post-retirement and I don't intend for it to change because the future is always unknown.

Post Reply