Author Topic: Monte Carlo measure  (Read 2336 times)

momcpa

  • Stubble
  • **
  • Posts: 102
Monte Carlo measure
« on: October 16, 2016, 06:21:24 PM »
I see and read on this forum different methods of checking the probability of investments making it through your retirement.   A person helping us look at our balances says they used the Monte Carlo method/standards (whatever the correct term is).   

Guess I'm looking for opinions from those more savvy than I am.............   Does this method provide reliable information on the percentage of your chances to have enough money throughout retirement?  I know that nothing can predict the future with 100% accuracy.  But should I feel some confidence in the range/percentage this is giving me?   Thanks again for help from those wiser than me  !!!


Indexer

  • Handlebar Stache
  • *****
  • Posts: 1463
Re: Monte Carlo measure
« Reply #1 on: October 16, 2016, 06:44:39 PM »
Monte carlo simulation just means you ran many different possible scenarios(hundreds to potentially millions) so you can look at a range and see where the results tend to congregate(a mean). They use it in many different fields. In financial planning it is normally used for running projections of what the future could look like.

Whether it is reliable depends on the inputs.

Some monte carlo simulators just use historic data. The upside is that it captures various relationships(stocks Vs bonds Vs inflation, etc.).  The downside is a very limited amount of projections. If you run 30-40 years of possibilities you only have so many backtesting scenarios to use.

Pure monte carlo:  This normally looks at the mean and standard deviation of different investments. You can run this type of simulation infinite times. The downside is that it normally fails to capture the various interrelationships and you can end up with some crazy potential results. If you run 10,000 scenarios, 1 could show your money growing 100%/yr forever.

What you want is something that takes the second, and based on a lot of research tunes it to be more reliable/realistic. The problem is this will probably require a lot of analysts/mathematicians so it can be expensive. The big investment firms have this sort of thing. So if they ran the projections for you I would put more weight on it than if you just used a free pure simulator or if you just used historic data.

ender

  • Walrus Stache
  • *******
  • Posts: 7402
Re: Monte Carlo measure
« Reply #2 on: October 16, 2016, 08:43:27 PM »
Variability in the current "price" of stocks matters a lot more than whatever Monte Carlo simulation you run.

For example, if 30 years from now market returns suggest stocks are 30% overpriced, this will have a much more significant impact on any analysis that sequence of returns (though sequence of returns will be impacted by such analysis).

When it comes to modeling your retirement, by far the most sensitive input value is this - the current "market valuation" of your inputs. The 4% rule is an attempt to guesstimate based on this input, but the results are only as accurate as the input.


momcpa

  • Stubble
  • **
  • Posts: 102
Re: Monte Carlo measure
« Reply #3 on: October 17, 2016, 11:25:39 AM »
Thanks for the replies.   The simulation was 'run' by a Wells Fargo investment person.  So as one poster mentioned, it was done by one of the larger institutions.

Our current score indicates that we are at a "91".   Meaning we have a really good chance of having plenty of money in our retirement.  Just wanted to check with those on this board who are a lot more knowledgeable than me as to how much I could reply on that number.

Thanks again.

ender

  • Walrus Stache
  • *******
  • Posts: 7402
Re: Monte Carlo measure
« Reply #4 on: October 17, 2016, 11:45:28 AM »
Thanks for the replies.   The simulation was 'run' by a Wells Fargo investment person.  So as one poster mentioned, it was done by one of the larger institutions.

Our current score indicates that we are at a "91".   Meaning we have a really good chance of having plenty of money in our retirement.  Just wanted to check with those on this board who are a lot more knowledgeable than me as to how much I could reply on that number.

Thanks again.

If this is your interest, post a case study - it's by far the best way to answer that sort of question.

AlanStache

  • Magnum Stache
  • ******
  • Posts: 3175
  • Age: 44
  • Location: South East Virginia
Re: Monte Carlo measure
« Reply #5 on: October 17, 2016, 12:25:20 PM »
Without knowing what specifically they ran it is impossible to say if it is at all good, garbage in-garbage out...  But if run by a "professional" at WF it is probably worth giving some weight to.  I would look at MC-simulation as one part of an over all picture that by itself does not give an answer one way or another.  You knowing your specific case need to understand your position and not blindly follow some algorithm.  And as is always said keep an open mind and be flexible and adapt your spending to changes in the world. 

my 0.02$

FWIW my business cards have the words "senior", "simulation" and "engineer" on them.

 

Wow, a phone plan for fifteen bucks!