Join Early Retirement Today
Reply
 
Thread Tools Display Modes
An oddball idea?
Old 05-31-2006, 08:42 PM   #1
Recycles dryer sheets
 
Join Date: Apr 2006
Posts: 190
An oddball idea?

dory,

Don't know if you're familiar with Fast Fourier Transforms. They are a relatively small programming function that convert time domain data streams into frequency domain.

Meaning, if you digitized a sinusoid of 10 Hz and then passed the resulting data stream through an FFT, what would emerge would be a single spike value at the 10 Hz point of the chart with the magnitude of all other frequencies zero.

If you passed your actual market data through an FFT, you would get several/many spikes of varying magnitude showing the relative presence of various frequencies of cycles within the data stream.

Where I'm going with this is your Monte Carlo work. There have been a number of folks who suggest the market is not well modeled by pure random numbers and they have tried to adjust them to look like the market. Well, if you took the FFT output of the historical market and used it as a filter on the pseudo random noise emerging from your random number generator, I think semi-theoretically you would have a MC model that would increase the available market-like sample data points for analysis up from 130 years to millions and . . . it would be cool, too.

Don't want to create work for you that is of no interest and this is semi sophisticated stuff, but it has a high nifty coefficient should you have some rainy days.

Just a thought . . . .
modlair is offline   Reply With Quote
Join the #1 Early Retirement and Financial Independence Forum Today - It's Totally Free!

Are you planning to be financially independent as early as possible so you can live life on your own terms? Discuss successful investing strategies, asset allocation models, tax strategies and other related topics in our online forum community. Our members range from young folks just starting their journey to financial independence, military retirees and even multimillionaires. No matter where you fit in you'll find that Early-Retirement.org is a great community to join. Best of all it's totally FREE!

You are currently viewing our boards as a guest so you have limited access to our community. Please take the time to register and you will gain a lot of great new features including; the ability to participate in discussions, network with our members, see fewer ads, upload photographs, create a retirement blog, send private messages and so much, much more!

Re: An oddball idea?
Old 06-03-2006, 10:08 AM   #2
Early-Retirement.org Founder
Developer of FIRECalc
dory36's Avatar
 
Join Date: Jun 2002
Posts: 1,840
Re: An oddball idea?

I guess I am not sure what question this would answer. The SWR tools look to identify the bottom edge of a wide range of possible outcomes, so when the overall outcome stretches from -300K to +4.3M, "precision" is not a likely goal. All (to my knowledge) research and simulations that consider market volatility converge on the same general results, and only get significant differences when they change the rules for how withdrawals are calculated.

That is not to say that such a line of inquiry might not be valuable for purposes other than SWR investigation. There are many useful and predictive frequency (and other) patterns in the data, and there are mutual funds and software products that attempt to detect and exploit these, using every imaginable technique.

Long ago I did a good bit of similar research regarding oil prices. My employer (in the Middle East) at the time was sufficiently intrigued by the idea to spend a lot of money on the effort. (Including costs for the staff I had assigned to the project, they probably spent in the high six figures for the two year project.) Yet after spending all that money, they abandoned the project when they decided that regional and world events (such as military invasions, SCUD missiles hitting at our doorstep, and more recently terrorist bombings* ) so overwhelmingly impacted the results that, in my boss's words, our efforts were akin to carefully measuring the spacing for the deck chairs on the Titanic.

FWIW, no single component, and not even a double-handful, were sufficient to provide interesting results, whether we were using "common sense" understandings of data relationships by the handful of experts (phd plus decades of relevant experience)* or neural networks that sought relationships too subtle for humans to detect them.

SWR is much easier!
__________________
Often uninformed, seldom undecided.

Twenty years from now you will be more disappointed by the things you didn't do than by the ones you did do. So throw off the bowlines. Sail away from the safe harbor. Catch the trade winds in your sails. Explore. Dream. Discover. Mark Twain
dory36 is offline   Reply With Quote
Re: An oddball idea?
Old 06-03-2006, 04:46 PM   #3
Recycles dryer sheets
 
Join Date: Apr 2006
Posts: 190
Re: An oddball idea?

Understood. And thanks for reading the item.

I don't think I laid a foundation for motivation before indulging in the bark on the trees of the forest.

The 130some years of data (I assume you do all your work with yearly and not daily or monthly results) is not, by traditional statistical standards, much of a sample size. This is why Monte Carlo techniques exist, of course, but in almost all areas in which MC techniques are applied there is immedate doubt that the waveform produced is valid.

It's a somewhat perverse analytical reality that when a process is modeled and a desire for enough samples to draw valid statistical conclusions exists, as soon as the MC runs are complete there is an inevitable challenge to the validity of the result because . . . the process modeled is presumed to be non random and purely random numbers are potentially not valid.

A very good example is blackjack simulation. If you're familiar with the game you know that Basic Strategy emerged from computer runs numbering in the millions and the house advantage was precisely calculated. "Card counting" schemes were invented and tested against the models and it was discovered that the house could be beaten and people wrote books and got famous (after they realized they could not remember all their card counting requirements).

But was the house advantage really well calculated? It is not clear at all that a Gaussian distributed pseudo random number generator is a correct emulation of a shuffled deck of cards. If it is not, then what?

MC runs for SWR do not yield the same numbers as the real 130some year market samples. This seems to be the case in repeated runs and I would venture to suggest that this is not surprising. What is surprising is the results are consistently lower. That reality is disquieting.

It means that, clearly, the 130some samples of the real history market is not well modeled by a Gaussian waveform. But it also means the 130some samples are not necessarily representative of all market 130sample sequences. That particular sequence of years could have been an aberration. People often speculate that the next 130 yrs will be better, or be worse.

So what I was suggesting above is a postulate that human nature does not change and there should be detectable in a 130 year waveform some measurable and characterizable core aspects of human behavior in markets. If one characterizes those aspects in the frequency domain (postulating that human behavior manifests itself in cycles), then the random noise (numbers) of thousands of 130some point sequences can be shaped by a digital filter whose design derives from the actual history sequence. You really don't have to understand digital signal processing to understand the concept. It's not at all unusual in many applications to filter frequencies in noise and shape noise to a certain form in the frequency domain.

Anyhow, that's what this would be. One runs the real historical data through an FFT and gets a plot of (Nyquist 1/2 frequency) points that represent magnitudes of
the possible frequencies embedded in that waveform.

Normalize these results.

Then generate 130some random numbers, FFT them, apply the normalized results above to this FFT'ed noise, run an inverse FFT and you have for the first time . . . perhaps ever . . . a second (in addition to the original historical data) sequence of 130some real "market-like" data on which to compute an SWR.

Obviously one then does this a thousand times, averages the obtained SWRs and you may then for the first time have not just a better estimate of the mean value of true SWR given inital conditions specified, but the most important thing that you then have is a valid, defensible measure of SWR standard deviation.

To perhaps some extent what we're saying here is that card shuffling very likely is well (not perfectly, just well) modeled by pure random numbers. But given that MC runs do not come out the same as real data runs, we have some reason to believe that the market is not a deck of cards and for MC techniques to be valid, the noise needs to be shaped.
modlair is offline   Reply With Quote
Re: An oddball idea?
Old 06-03-2006, 06:55 PM   #4
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,856
Re: An oddball idea?

And people wonder how Elliott Wave theory has survived after all these years.

As for the difference between MC & historical-return results, it's possible that investment returns are more correlated than MC allows for. I think that markets are a lot like the weather model, where persistence is more likely than change. MC tends to throw in a randomly-generated number without regard for persistence, which raises volatility while tending to lower returns. But that reasoning's not as complex as a FFT function...
__________________
*

Co-author (with my daughter) of “Raising Your Money-Savvy Family For Next Generation Financial Independence.”
Author of the book written on E-R.org: "The Military Guide to Financial Independence and Retirement."

I don't spend much time here— please send a PM.
Nords is offline   Reply With Quote
Re: An oddball idea?
Old 06-03-2006, 11:53 PM   #5
Thinks s/he gets paid by the post
 
Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: An oddball idea?

rodmail,

As Nords points out, correlations are what is lost using a Monte Carlo simulation. While the correlations may be too complex to quantify mathematically, stock returns are correlated to bond returns are correlated to inflation. While it is far too complex to quantify with a simple mathematical expression, returns and inflation are causal. Something causes them to rise and fall. That means that this year's numbers are correlated to last year's, and the year before, and . . .

By definition, historical simulations account for these correlations. Monte carlo cannot. That means that monte carlo simulations overestimate the extremes possible in the economy and therefore underestimates SWR.

The main strength of monte carlo SWR simulations is that you can consider asset classes with less than 130 years of data.

One interesting approach to monte carlo SWR simulations is to use the 130 years of data, but randomly select the order of the years. This captures some of the correlations, but not the year-to-year correlations. The technique also still suffers the limitation that you need 130 years of data to include an asset class.
sgeeeee is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 08:46 AM   #6
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
Ed_The_Gypsy's Avatar
 
Join Date: Dec 2004
Location: the City of Subdued Excitement
Posts: 5,588
Re: An oddball idea?

rodmail,

As I recall, FFTs have been applied to market data in the past without success. The system is open-ended. (I wish I could remember a reference. Sorry.)
__________________
I have outlived most of the people I don't like and I am working on the rest.
Ed_The_Gypsy is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 09:28 AM   #7
Recycles dryer sheets
 
Join Date: Apr 2006
Posts: 190
Re: An oddball idea?

Quote:
That means that monte carlo simulations overestimate the extremes possible in the economy and therefore underestimates SWR.
1) I am familiar with the adjusted MC approach on raddr's site. I think I understand the concept of why MC yields, in pure form, an absence of continuity and trends. (I am not confident I understand why large jumps in year to year market performance must necessarily lead to lower SWRs. But no matter. We do see MCs yielding lower SWRs than real world market results -- which means in some way random numbers do not well model our single 130 yr sample series.)

2) The attempt to undo that absence of continuity and trends is by selective compression of possible data points emerging from the Gaussian data stream -- to prevent "the next point" from diverging too much from the previous point.

3) What may not be clear unless you think of this in terms of signal processing is that such a compression is, in the frequency domain, a low pass filter. It passes low frequencies and stops high frequencies. It's a crude form of what I'm offering as an idea. Compressing data like that is equivalent to not allowing a high frequency to exist -- because a high frequency will have a very narrow time duration (high freq. = short wavelength) between peak and trough of the sinusoid. What I'm proposing is far more refined than that.

3) I am completely aware that this could all be worthless. I don't know if there are cycles to human behavior over 130 yrs of time. If there are not, though, an FFT should discover that by showing no frequencies with any profound presence vs others. But also vice versa. Regardless of that, the result may yield mean values of SWRs not particularly different from what is obtained by MC runs now (though I doubt it). It may also yield a value for sigma that is very large and teaches us nothing. If we faced a task of inventing code for an FFT, I wouldn't even mention this because the risk (in terms of dory's work magnitude) / reward (no confident increase in knowledge) ratio would be very poor. But FFT code is readily available and should be easily ported -- and as noted above someone has already made a "1st order" attempt to extract frequencies and shape noise. But that attempt was made "in the dark". Good reasoning was applied to why the random numbers are not market-like, but the technique used is not defensible against our own suspicions that it is contrived. An FFT-derived result can allay those suspicions.

4) We have one and only one sample of how a market behaves over 130+ yrs. That looks very poor viewed as a single series. We have 130+ samples of a market's annual performance. This is better, but still not good. We have < 100 samples of 40 year periods within that sequence. That's pretty poor as a sample size of a universe of human behavior. Modifying the noise of MC is a possible way to get more samples that have some rationale behind being defensibly "market-like".

Quote:
As I recall, FFTs have been applied to market data in the past without success. The system is open-ended. (I wish I could remember a reference. Sorry.)
I believe what has been done in the past is application of FFT techniques to daily data. The intent was to apply technical analysis moving averages and Bollinger bands and the usual TA toolbox of witchcraft to data in the frequency domain rather than time domain. I don't know of any attempts to create additional series of 130+ yr sequences of frequency-shaped random noise in order to create additional samples of market-like data.

But . . . that doesn't mean someone hasn't done it. They probably have. And like I say above, maybe it yields nothing interesting. Seems a new enough application that not much work may have been done. Dunno.



modlair is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 09:48 AM   #8
Recycles dryer sheets
 
Join Date: Jun 2002
Posts: 376
Re: An oddball idea?

Quote:
Originally Posted by rodmail
1) I am familiar with the adjusted MC approach on raddr's site.* I think I understand the concept of why MC yields, in pure form, an absence of continuity and trends.* (I am not confident I understand why large jumps in year to year market performance must necessarily lead to lower SWRs.* *But no matter.* We do see MCs yielding lower SWRs than real world market results -- which means in some way random numbers do not well model our single 130 yr sample series.)

2) The attempt to undo that absence of continuity and trends is by selective compression of possible data points emerging from the Gaussian data stream --* to prevent "the next point" from diverging too much from the previous point.*

3) What may not be clear unless you think of this in terms of signal processing is that such a compression is, in the frequency domain, a low pass filter.* It passes low frequencies and stops high frequencies.* It's a crude form of what I'm offering as an idea.* Compressing data like that is equivalent to not allowing a high frequency to exist -- because a high frequency will have a very narrow time duration (high freq. = short wavelength) between peak and trough of the sinusoid.* What I'm proposing is far more refined than that.*

3) I am completely aware that this could all be worthless.* I don't know if there are cycles to human behavior over 130 yrs of time.* If there are not, though, an FFT should discover that by showing no frequencies with any profound presence vs others.* But also vice versa.* Regardless of that, the result may yield mean values of SWRs not particularly different from what is obtained by MC runs now (though I doubt it).* It may also yield a value for sigma that is very large and teaches us nothing.* If we faced a task of inventing code for an FFT, I wouldn't even mention this because the risk (in terms of dory's work magnitude) / reward (no confident increase in knowledge) ratio would be very poor.* But FFT code is readily available and should be easily ported -- and as noted above someone has already made a "1st order" attempt to extract frequencies and shape noise.* But that attempt was made "in the dark".* Good reasoning was applied to why the random numbers are not market-like, but the technique used is not defensible against our own suspicions that it is contrived.* An FFT-derived result can allay those suspicions.*

4) We have one and only one sample of how a market behaves over 130+ yrs.* That looks very poor viewed as a single series.* We have 130+ samples of a market's annual performance.* This is better, but still not good.* We have < 100 samples of 40 year periods within that sequence.* That's pretty poor as a sample size of a universe of human behavior.* Modifying the noise of MC is a possible way to get more samples that have some rationale behind being defensibly "market-like".

I believe what has been done in the past is application of FFT techniques to daily data.* The intent was to apply technical analysis moving averages and Bollinger bands and the usual TA toolbox of witchcraft to data in the frequency domain rather than time domain.* I don't know of any attempts to create additional series of 130+ yr sequences of frequency-shaped random noise in order to create additional samples of market-like data.

But . . . that doesn't mean someone hasn't done it.* They probably have.* And like I say above, maybe it yields nothing interesting.* Seems a new enough application that not much work may have been done.* Dunno.
So Rodmail, if I understand you correctly, you're suggesting like 4%, give or take?

Cb
Cb is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 10:06 AM   #9
Thinks s/he gets paid by the post
 
Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: An oddball idea?

Quote:
Originally Posted by rodmail
1) I am familiar with the adjusted MC approach on raddr's site. *I think I understand the concept of why MC yields, in pure form, an absence of continuity and trends. *(I am not confident I understand why large jumps in year to year market performance must necessarily lead to lower SWRs. * . . .
There is nothing in most monte carlo simulations to keep the sequence from producing extremely high inflation with very low bond returns for extended periods of time, for example. *This is a highly unlikely situation in real life, but since monte carlo treats each performance metric as a random event -- independent of the other performance metrics, it has no problem producing such sequences. *Notice that monte carlo is equally likely to produce overly optimistic sequences, but they have no contribution to SWR since SWR is determined only by worst case analysis.

Quote:
2) The attempt to undo that absence of continuity and trends is by selective compression of possible data points emerging from the Gaussian data stream -- *to prevent "the next point" from diverging too much from the previous point.
* Yes. *It is absolutely theortically possible to force appropriate correlations. *Rather than choose annual stock return, bond return, and inflation using three unique random numbers for each year in the sequence, you could (and probably should) choose a single random number to describe the movement of each of these metrics in relationship to the previous year's values. *This could be done (theoretically) using an approach like you describe -- or several others. *It is a much harder problem and I'm not sure that we really have enough real world data to capture all the relationships, but some of us would find the capability fun to play with. *While I am one those people who would love to play around with such a simulator, I also admit that it is a little bit like measuring with a micrometer and cutting with a chain saw.

Quote:
3) What may not be clear unless you think of this in terms of signal processing is that such a compression is, in the frequency domain, a low pass filter. . .
*
I get it. *I've suggested alternative approaches to capture more of the correlations to Raddr and gummy. *I even talked with dory about one possible approach to quantifying the correlation impact with a version of monte carlo when he was developing it. *It's easy for me to offer advice, but I'm mostly retired now -- too lazy to continue to write monte carlo code. *Beggars can't be choosers.

Quote:
3) I am completely aware that this could all be worthless. *I don't know if there are cycles to human behavior over 130 yrs of time. *If there are not, though, an FFT should discover that by showing no frequencies with any profound presence vs others. . .
I don't think it is likely to lead to changes in the way people choose retirement spending habits. *It's not likely to change the fact that a 4% SWR is a pretty good first SWAG and that events outside of the realm of simulators keeps us from getting more refinement. *But I think such a tool in the hands of the right person could lead to a greater understanding of some of the issues and risks in retirement.

Quote:
4) We have one and only one sample of how a market behaves over 130+ yrs. *That looks very poor viewed as a single series. *We have 130+ samples of a market's annual performance. *This is better, but still not good. *We have < 100 samples of 40 year periods within that sequence. *That's pretty poor as a sample size of a universe of human behavior. *Modifying the noise of MC is a possible way to get more samples that have some rationale behind being defensibly "market-like".
Yeah. *But remember that everything about the monte carlo simulation -- performance distributions, inflation distributions, correlations, FFT filter variable determination -- is all calibrated using only that historical data. *If you determine the mc variables empirically using the historical data, there is no guarantee that it describes anything causally. *If you work hard at it, you can guarantee that your monte carlo simulation produces nearly identical probability results as the historical record, but you can't guarantee that it has any more predictive capability than the historical record it is based on.

sgeeeee is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 10:44 AM   #10
Early-Retirement.org Founder
Developer of FIRECalc
dory36's Avatar
 
Join Date: Jun 2002
Posts: 1,840
Re: An oddball idea?

Quote:
Originally Posted by sgeeeee
Quote:
Originally Posted by rodmail
1) I am familiar with the adjusted MC approach on raddr's site. *I think I understand the concept of why MC yields, in pure form, an absence of continuity and trends. *(I am not confident I understand why large jumps in year to year market performance must necessarily lead to lower SWRs. * . . .
There is nothing in most monte carlo simulations to keep the sequence from producing extremely high inflation with very low bond returns for extended periods of time, for example. *This is a highly unlikely situation in real life, but since monte carlo treats each performance metric as a random event -- independent of the other performance metrics, it has no problem producing such sequences. *Notice that monte carlo is equally likely to produce overly optimistic sequences, but they have no contribution to SWR since SWR is determined only by worst case analysis.

Quote:
2) The attempt to undo that absence of continuity and trends is by selective compression of possible data points emerging from the Gaussian data stream -- *to prevent "the next point" from diverging too much from the previous point.
* Yes. *It is absolutely theortically possible to force appropriate correlations. *Rather than choose annual stock return, bond return, and inflation using three unique random numbers for each year in the sequence, you could (and probably should) choose a single random number to describe the movement of each of these metrics in relationship to the previous year's values. *This could be done (theoretically) using an approach like you describe -- or several others. *It is a much harder problem and I'm not sure that we really have enough real world data to capture all the relationships, but some of us would find the capability fun to play with. *While I am one those people who would love to play around with such a simulator, I also admit that it is a little bit like measuring with a micrometer and cutting with a chain saw.

Quote:
3) What may not be clear unless you think of this in terms of signal processing is that such a compression is, in the frequency domain, a low pass filter. . .
*
I get it. *Check out my background: *http://www.golio.net/* I've suggested alternative approaches to capture more of the correlations to Raddr and gummy. *I even talked with dory about one possible approach to quantifying the correlation impact with a version of monte carlo when he was developing it. *It's easy for me to offer advice, but I'm mostly retired now -- too lazy to continue to write monte carlo code. *Beggars can't be choosers.

Quote:
3) I am completely aware that this could all be worthless. *I don't know if there are cycles to human behavior over 130 yrs of time. *If there are not, though, an FFT should discover that by showing no frequencies with any profound presence vs others. . .
I don't think it is likely to lead to changes in the way people choose retirement spending habits. *It's not likely to change the fact that a 4% SWR is a pretty good first SWAG and that events outside of the realm of simulators keeps us from getting more refinement. *But I think such a tool in the hands of the right person could lead to a greater understanding of some of the issues and risks in retirement.

Quote:
4) We have one and only one sample of how a market behaves over 130+ yrs. *That looks very poor viewed as a single series. *We have 130+ samples of a market's annual performance. *This is better, but still not good. *We have < 100 samples of 40 year periods within that sequence. *That's pretty poor as a sample size of a universe of human behavior. *Modifying the noise of MC is a possible way to get more samples that have some rationale behind being defensibly "market-like".
Yeah. *But remember that everything about the monte carlo simulation -- performance distributions, inflation distributions, correlations, FFT filter variable determination -- is all calibrated using only that historical data. *If you determine the mc variables empirically using the historical data, there is no guarantee that it describes anything causally. *If you work hard at it, you can guarantee that your monte carlo simulation produces nearly identical probability results as the historical record, but you can't guarantee that it has any more predictive capability than the historical record it is based on.
What he said.
__________________
Often uninformed, seldom undecided.

Twenty years from now you will be more disappointed by the things you didn't do than by the ones you did do. So throw off the bowlines. Sail away from the safe harbor. Catch the trade winds in your sails. Explore. Dream. Discover. Mark Twain
dory36 is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 11:26 AM   #11
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,708
Re: An oddball idea?

Quote:
Originally Posted by sgeeeee
Check out my background: http://www.golio.net/
Oh. No. This CANT be you!

http://mtt.org/awards/images/golio_web.jpg

__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 11:32 AM   #12
Thinks s/he gets paid by the post
 
Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: An oddball idea?

. . .
sgeeeee is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 11:59 AM   #13
Recycles dryer sheets
 
Join Date: Jun 2002
Posts: 376
Re: An oddball idea?

Hmmm. It would appear that SG's "body of work" is a bit larger than the half dozen FIREcalc runs with various expense ratios I mentioned yesterday.

Fascinating.

Cb
Cb is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 01:21 PM   #14
Recycles dryer sheets
 
Join Date: Apr 2006
Posts: 190
Re: An oddball idea?

Quote:
Yes. It is absolutely theortically possible to force appropriate correlations. Rather than choose annual stock return, bond return, and inflation using three unique random numbers for each year in the sequence, you could (and probably should) choose a single random number to describe the movement of each of these metrics in relationship to the previous year's values. This could be done (theoretically) using an approach like you describe -- or several others. It is a much harder problem and I'm not sure that we really have enough real world data to capture all the relationships, but some of us would find the capability fun to play with.
This does presume that a large sample size existed from which samples of relationships between these variables were taken, a correlation coefficient computed and its statistical significance tested. That hasn't been done with a solid conclusion, as best I know, and given that the Fed has been increasing rates for X months now with the market rising to a near all time high . . . how would you persuade anyone that a experimental model should presume higher rates reduce stock prices in 2006 and onward? They certainly seem to on days of announcements, but the last 18 mos are powerful counter evidence.

Now I do think what you're discussing is interesting, but I also think that sort of thing is all about market prediction and pursuit of higher returns, and surely people have already done statistically rigorous correlations of stocks to other variables to seek a predictive edge. OTOH, anyone who found such a thing would, of course, be crazy to publish it. Hmmm.

Note that I have not talked about any variable other than the raw equity market's performance. This does not mean that I should not, but I haven't. I've talked only about increasing our sample size of intermediate term human behavior trading stocks.

My suggestion is clearly more simplistic than multivariate correlation in the frequency (or time) domain. The 130+ yr sample we have could have been an aberration with many end of year numbers at sinusoidal troughs. Days to come could be far rosier. Or the reverse. The SWR from that single sample might be substantially deceptive in either direction. I just want more samples of "market-like" data and I'm defining "market like" as "having an equivalent Nyquist-limited frequency profile". Probably some value in noting that raddr's "1st order" model does yield a sigma around a mean SWR that suggests a different value from the historical data.

No question at all that the hypothesized definition of "market-like" is unproven. No one has shown that a waveform of equivalent frequency "content" is a good definition of "market-like". I know this is only an experiment, and again, I would not even suggest it if the work magnitude looked huge.

Quote:
I don't think it is likely to lead to changes in the way people choose retirement spending habits. It's not likely to change the fact that a 4% SWR is a pretty good first SWAG and that events outside of the realm of simulators keeps us from getting more refinement. But I think such a tool in the hands of the right person could lead to a greater understanding of some of the issues and risks in retirement.
Don't know how good 4% is from one sample. I'd like to find the time to study raddr's methodology more carefully to determine how he obtained different numbers. But hey, you're right, no question news events could change human behavior . . . well, that's badly phrased. I guess any correlation we look for presumes human behavior does not change, but an asteroid strike is likely going to trump the variance on the waveform produced by human behavior.

Quote:
Yeah. But remember that everything about the monte carlo simulation -- performance distributions, inflation distributions, correlations, FFT filter variable determination -- is all calibrated using only that historical data. If you determine the mc variables empirically using the historical data, there is no guarantee that it describes anything causally. If you work hard at it, you can guarantee that your monte carlo simulation produces nearly identical probability results as the historical record, but you can't guarantee that it has any more predictive capability than the historical record it is based on.
Dead on correct. We are using a historical record that we suspect of aberration as the source of a frequency profile to which we are molding random noise. And news events can undo everything.

But.

We already have done the work using the historical record on the presumption that it has some value. And we have no control of news. If there is a composite waveform in that data, of multiple frequencies, we could find them and model it and have more samples of market-like data.

If we were sitting here reading this in the year . . . what, 5000 AD or something and we had 100's of 40 year duration samples and discovered that the SWR for those in total was 6%, would we still worship the 4% number? No, of course not. The equivalent work that has produced the 100% certain 4% number would have yielded a different number. That first 130+ yr period would be exposed as an aberration.

Oh well. It does seem curious to me that random numbers yield consistently lower SWRs and rationally, non-contrived, low-pass filtered random numbers yield different SWRs than the historical record. If the frequency content of the historical record was uniformly distributed, I don't think these results would differ. Dunno.

modlair is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 02:48 PM   #15
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,708
Re: An oddball idea?

Quote:
Originally Posted by sgeeeee
You're still one ugly mutha%^@#$...
__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 03:02 PM   #16
Thinks s/he gets paid by the post
 
Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: An oddball idea?

Quote:
Originally Posted by rodmail
. . . If we were sitting here reading this in the year . . . what, 5000 AD or something and we had 100's of 40 year duration samples and discovered that the SWR for those in total was 6%, would we still worship the 4% number?*
You realize that this is not possible. *The SWR from historical simulation can never be larger than it is today. *It can only go down from additional history. *Remember that the SWR is a worst case search of all possible historical sequences. *Since we already have sequences that produce SWRs as low as 4%, it can never go up.

Quote:
Oh well. *It does seem curious to me that random numbers yield consistently lower SWRs and rationally, non-contrived, low-pass filtered random numbers yield different SWRs than the historical record. *If the frequency content of the historical record was uniformly distributed, I don't think these results would differ.
*
I don't understand your issue here. *Stock performance, bond performance and inflation rate are not random events. *They are causal. *The causes are complex and not readily described with mathematical models, but that does not mean they are random. *Analysis that works perfectly well for cards and dice, does not apply as well to finance. *Conversion in Fourier space and application of filter techniques is not a causal description of financial issues. *It may or may not provide a better fit to actual performance, but it does not provide a causal description that can legitimately be used as a prescription for the future. *
sgeeeee is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 03:03 PM   #17
Thinks s/he gets paid by the post
 
Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: An oddball idea?

And don't you forget it.* :P
sgeeeee is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 03:13 PM   #18
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,708
Re: An oddball idea?

Quote:
Originally Posted by rodmail
would we still worship the 4% number
I still dont see any "worship" of the 4% number, or "beating people up" for proposing other numbers.

Over 4% fails for the data we already have now. Adding 40,000,000 years more data wont remove that failed sequence.

Although you might get 99.999% at 6% from the 40,000,000 years worth instead of the 80% you get now.

Presuming our economy/stock markets dont turn downwards going forward from the past. A lot of our prior 'returns' come from when our economy was practically an 'emerging market'. As a maturing market, we can plausibly count on lowered returns going forward as the risk premium shrinks.

Which might produce some return sequences that say 2% fails. Or that 4% is pretty aggressive.

The great news is that everyone can use any method of calculating and withdrawal, and any percentage they like!

In fact, I'll bet half of my next years withdrawal that we could talk about this for 30,000 more posts, and absolutely nobody would change their mind about what they think!
__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 03:33 PM   #19
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
REWahoo's Avatar
 
Join Date: Jun 2002
Location: Texas: No Country for Old Men
Posts: 50,004
Re: An oddball idea?

Quote:
Originally Posted by sgeeeee
Nooooo.

That was me. This is me now:

http://www.rfglobalnet.com/content/n...F236F09}#golio

OMG!!



Willy Wonka




+ Young Dr Frankenstein




= Dr Golio sgeeeeeee!


__________________
Numbers is hard
REWahoo is offline   Reply With Quote
Re: An oddball idea?
Old 06-04-2006, 03:37 PM   #20
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,708
Re: An oddball idea?

__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Reply


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Using your Roth as part of your emergency fund--good idea? CompoundInterestFan FIRE and Money 14 05-21-2007 12:08 AM
my dog's idea of a nice gift bright eyed Other topics 3 05-04-2007 03:48 PM
$100,000 idea contest Nicholas Other topics 5 10-28-2005 10:38 PM
Investment idea: Deluxe Corp. brewer12345 FIRE and Money 9 03-30-2005 03:08 PM
Another dumb investment idea Ted Other topics 18 09-12-2003 08:58 AM

» Quick Links

 
All times are GMT -6. The time now is 01:56 AM.
 
Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2024, vBulletin Solutions, Inc.