fire clac calculator

[socfp]

first time caller, I just went to use the FIRECALC calculator and saw wild numbers generated. In the detail printout I saw the years began in 1871 ?? Has anyone had similar problems

Best
SOCfp

 

[REWahoo]

Read up on how Firecalc works here:

http://www.fireseeker.com/explain.htm

If that doesn't answer your questions, post again and
I'm sure someone will help.

REW

 

[obryanjf]

I was intrigued by a post a week or so ago that discussed a variable withdrawl rate in addition to the normal fixed rate set at retirement. I added that feature to my spreadsheet, which got me to look more at FIRECALC. My issue is the partial periods at the end of every simulation. For example, using a 40 year retirement, I would have 91 full 40 periods and 40 partial periods. The 40 partials are counted as pass/fail depending on the outcome. My sense is the 40 partials should not be counted either way as they are not full periods. Especially when the partials are small, they are generally a pass even though from 0 to 10 years would have been used to get that years result and for that time period, with any diversification it is nearly impossible to run out of money using near standard WR's.

I believe the results presented then are generally optimistic.

Does anybody that knows FIRECALC know the reasoning behind including these partials in the final pass/fail percentage.

job

 

[Cut-Throat]

I believe the results presented then are generally optimistic.

Does anybody that knows FIRECALC know the reasoning behind including these partials in the final pass/fail percentage.

I think you have a good point! I also think that the Percent of pass fail should be given for whole periods only.

Maybe there could be an explanation added and another percent. Such as Partial Periods for years 1966 and Beyond are 99% safe so far.........

But the author of FireCalc is too busy because he's retired and no longer screwing off at work to make any modifications to the program.

Dory?

 

[dory36]

You can always view the detailed results, count the red lines, and divide by 91 if you want a different figure.

When I was developing Firecalc, the other tool used for these calculations was Intercst's spreadsheet, which ignored the partial periods (at least in that version). But there were some awful times in the 1960-2000 timeframe which were being ignored because they weren't full 40 year periods.

So the partial periods were added to avoid being overly optimistic by ignoring yesterday's calamities.

 

[obryanjf]

I would be happy to change it(program it) to something if you/? wanted to.  My current model rolls after it runs out of years, to the first year i do have data for.  I'm not sure that is good either.  Another option is to randomly pick years to use when partial years come in.   I think both of those are better than partials; however i am not a statistician, just an engineer that likes to see how money grows and maybe grow faster, better, etc.

job

 

[dory36]

I guess I am not sure the problem you describe is real, even though it sounds logical.

The best test is an actual run.

If I run a quick 40 year test with all the default settings, but change the the years to 40 and the withdrawal to 30290 (to get close to 90%), then 14 failing years result out of 91 full and 40 partial sequences.

If we looked only at the full sequences, then the success rate would be 93.4%. Only 6 out of 91 sequences fail.

But in the remaining 40 partial sequences, there are 8 failures, even though the shorter durations would lead you to believe these are safer. The 80% rate of these 40 partial periods pulls the overall success rate down to 89.4%.

So clearly, in a 40 year run, including the partial sequences prevents an overly optimistic result.

So... I'm not inclined to drop these out of the calculations.

Dory36

 

[obryanjf]

I agree with the simulation results. I had a run the other day where i had some failures on the fulls and none on the partials, but I cannot re-create it. I forgot the inputs. I do not have any better ideas or alternatives. Sorry to trouble you.

job

 

[Cut-Throat]

But in the remaining 40 partial sequences, there are 8 failures, even though the shorter durations would lead you to believe these are safer. The 80% rate of these 40 partial periods pulls the overall success rate down to 89.4%.

So clearly, in a 40 year run, including the partial sequences prevents an overly optimistic result.

So... I'm not inclined to drop these out of the calculations.

Dory36

Yup, I've changed my mind again! - It can go both ways, as you point out. Probably working fine as is.

 

[Olav23]

I am wondering if anyone has given thought to adding international historical numbers to FireCalc. I wonder what effects it would have on the outcome if results from developing/emerging countries were added. It might be harder to come across reliable data for other countries, but who knows if it might be more realistic as it has real data to act upon no matter what state the Country is in in the future. (I am assuming firecalc is basically a USA-outcome based simulation)

 

[Nords]

It might be harder to come across reliable data for other countries, but who knows if it might be more realistic as it has real data to act upon no matter what state the Country is in in the future.

I think the technical term for this is "Garbage in, garbage out." Heck, we can't even decide what numbers FIRECalc should use for inflation.

No one has anything nice to say about the SEC or GAAP, but they're shiny paragons of virtue among the rest of the world's reporting & accounting standards.  Let alone anything involving the words "developing" or "emerging"...

 

[sgeeeee]

By including partial periods in the probability of success calculation, you catch periods that would fail before the partial period is over (ie. a failure at year 5 is still a failure if the whole period is simulated), but might optimistically assume success when only 5 years of a 30 year period are actually simulated. Depending on the studies you are doing, either approach to computing probability of success or SWR can give you strange results. The real solution is to be aware of the calculation and look at the detailed printouts.

 

[Nords]

Re: FIRECalc calculator

The real solution is to be aware of the calculation and look at the detailed printouts.   

Phew, what a relief, I was afraid that someone was going to discover that ***** was right after all and that I'd need to start drafting a resume'...

 

[sgeeeee]

Re: FIRECalc calculator

Phew, what a relief, I was afraid that someone was going to discover that ***** was right after all and that I'd need to start drafting a resume'...

Glad I could help, Nords. The thought of you tossing and turning at night worried about what to put on your resume, worried about whether or not to cut your hair, etc. . . well, that would make us all feel miserable.

 

[bosco]

I am wondering if anyone has given thought to adding international historical numbers to FireCalc. I wonder what effects it would have on the outcome if results from developing/emerging countries were added. It might be harder to come across reliable data for other countries, but who knows if it might be more realistic as it has real data to act upon no matter what state the Country is in in the future. (I am assuming firecalc is basically a USA-outcome based simulation)

Dimson, Marsh, and Staunton have published some reputedly good data on historical international returns. Their numbers have been used by Ibbotson & Associates and others, and if I'm not mistaken, a lot of their data went into the historical numbers used by Firecalc etc.

The problem is, I have been unable to find the actual data anywhere on the internet. Their book, "The Triumph of Optimists" is expensive, and although it has graphs, apparantly does not actually publish the data.

If anyone knows where to get these numbers, I would love a set. I'd like to plug them into a couple of retirement calculators including gummy-stuffs sensible withdrawal calculator.