Page 1 of 1

probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:35 am
by atrchi
Hello,

I understand the PP has not had a losing year since 1982 or so.

This reminds me that San Francisco has not had a devastating earthquake since 1908.

That got me wondering.

Using purely statistical methods, like those used to predict earthquakes, what is the probability the PP will lose money next year?

e.g. When you google "probability of earthquake next year" you'll find stuff like this: :o

http://www.guardian.co.uk/world/2012/ja ... four-years

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:42 am
by craigr
There are no guarantees with investing. Best you can do is widely diversify because nobody knows what may or may not happen. Past returns can disprove theories, but can't prove them going forward. Something could happen that would impact the portfolio assets all at once and cause a large loss, but then again most other portfolios are probably in the same boat.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 8:23 am
by jimbojones
Using the PP return data on the website, the arithmetic mean return is 9.945% with a standard deviation of 7.99%.  A return of 0% is 1.244 standard deviations below the mean.  That roughly translates to the 10.7th percentile.  So from a purely statistical perspective, I'd say there's a 10.7% chance of a negative nominal return.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:01 pm
by Xan
Are PP returns normally distributed, though?

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:20 pm
by jimbojones
Xan wrote: Are PP returns normally distributed, though?
It isn't perfect, but the returns are bell-shaped.  I made a histogram and I'd post the picture here, but I don't know how.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:27 pm
by Pointedstick
If you don't already have an image file of your histogram, take a screenshot of your screen. Crop the resulting file with software of your choice (GIMP, Preview, etc.). Upload the file to http://imgur.com/. Copy the URL to your image, and head back here to make a new post. Click on the little picture icon on the button bar, and paste the URL between the IMG and /IMG tags that appear. Voila!

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:39 pm
by Gumby
While it's not a guarantee, when the permanent portfolio had a bad year, it was always preceded and followed by years with greater than 10% nominal returns.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:41 pm
by jimbojones
Image

Thanks for the tip, Pointedstick!

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:43 pm
by TripleB
The problem is choosing the end points of the year.

Your question is based on the common fallacy of caring about what happens from Jan 1 through Dec 31.

If you look at 2008/2009, the PP did something like +0.2% one of those years (I forgot which was the really bad one).

However, Long Term Treasuries bonds gained around 40% in the last 2 months of that bad year. Thus, if you calculated a year return as from Oct 1 through Sep 31st of that time period, the PP might have been down 15% (I don't know, I'm speculating/estimating). It was carried by LTTs in the last couple of months for that calendar annual year.

Since you're interested (or should be interested) in holding the PP for a long period of time, then it shouldn't matter.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 12:51 pm
by Ad Orientem
If you flip a coin once a day for forty years, and every day it come up heads, what are your statistical odds of it coming up tails tomorrow? 50/50. The future is unpredictable. If you're using the PP for all or most of your money, yes you will have an occasional down year. History suggests that those down years will be rare and shallow. So relax and enjoy the ride.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 4:06 pm
by Mark Leavy
Ad Orientem wrote: If you flip a coin once a day for forty years, and every day it come up heads, what are your statistical odds of it coming up tails tomorrow? 50/50.
Or as Fat Tony would say, "That coin’s gotta be loaded. . .”?  and bet appropriately.

http://en.wikipedia.org/wiki/Ludic_fallacy

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 4:22 pm
by craigr
Mark Leavy wrote:
Ad Orientem wrote: If you flip a coin once a day for forty years, and every day it come up heads, what are your statistical odds of it coming up tails tomorrow? 50/50.
Or as Fat Tony would say, "That coin’s gotta be loaded. . .”?  and bet appropriately.

http://en.wikipedia.org/wiki/Ludic_fallacy
Amen! In investing the odds are not that of a casino. Your primary defense is wide diversification, and even then it can't guarantee against all negative outcomes. It can just put the odds (!) in your favor that something very bad happening in this world won't wipe you out entirely. There are no guarantees in investing!

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 4:51 pm
by jimbojones
TripleB wrote: The problem is choosing the end points of the year.

Your question is based on the common fallacy of caring about what happens from Jan 1 through Dec 31.

If you look at 2008/2009, the PP did something like +0.2% one of those years (I forgot which was the really bad one).

However, Long Term Treasuries bonds gained around 40% in the last 2 months of that bad year. Thus, if you calculated a year return as from Oct 1 through Sep 31st of that time period, the PP might have been down 15% (I don't know, I'm speculating/estimating). It was carried by LTTs in the last couple of months for that calendar annual year.

Since you're interested (or should be interested) in holding the PP for a long period of time, then it shouldn't matter.
I used a sample size of 40, which should be enough.  Adding hundreds of additional samples will improve the data but the results shouldn't be materially different.  I thought I saw that someone ran the histogram over rolling 12-month periods, but maybe that wasn't on this site.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 5:52 pm
by MediumTex
jimbojones wrote:
TripleB wrote: The problem is choosing the end points of the year.

Your question is based on the common fallacy of caring about what happens from Jan 1 through Dec 31.

If you look at 2008/2009, the PP did something like +0.2% one of those years (I forgot which was the really bad one).

However, Long Term Treasuries bonds gained around 40% in the last 2 months of that bad year. Thus, if you calculated a year return as from Oct 1 through Sep 31st of that time period, the PP might have been down 15% (I don't know, I'm speculating/estimating). It was carried by LTTs in the last couple of months for that calendar annual year.

Since you're interested (or should be interested) in holding the PP for a long period of time, then it shouldn't matter.
I used a sample size of 40, which should be enough.  Adding hundreds of additional samples will improve the data but the results shouldn't be materially different.  I thought I saw that someone ran the histogram over rolling 12-month periods, but maybe that wasn't on this site.
Is there another site looking at the PP at the molecular level?

I thought it was just us.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 6:22 pm
by Xan
I think that rolling 12-months thing was on this forum; I remember seeing that, but I don't know where exactly.

Re: probability of PP losses next year.

Posted: Tue Oct 30, 2012 8:54 pm
by jimbojones
Turns out the rolling 12-month graph was from the "Sorry guys, I broke PP" thread:

http://gyroscopicinvesting.com/forum/in ... ic=3384.60

Slotine's graph only went back to 2000, but it would be interesting to see 40 years of rolling 12-month data.