Minimizing fat tail risk w/ lower equity allocation - myth?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Minimizing fat tail risk w/ lower equity allocation - myth?

Post by ddb » Wed Aug 24, 2011 1:02 pm

Larry and others have stated that they prefer higher risk equities with a lower overall allocation to equities, partly to reduce the fat tails of rate of return distributions (I'm paraphrasing, of course). 2008 is often cited as the prototypical environment for how such a strategy can be useful.

I've often wondered, once rebalancing is considered, does the above hold true? Remember, even a portfolio with a 1% allocation to equities will go to zero if equities drop by 100% over a prolonged period, because you are constantly rebalancing your bonds into equities.

I looked at a single example to see how this would have worked. I used the period of 12/31/2006 to 02/28/2009. I compared two portfolios:

Portfolio A: 60% DFA US Large Cap, 40% DFA 2-Year Global Fixed Income
Portfolio B: 40% DFA US Small Cap Value, 60% DFA 2-Year Global Fixed Income

I looked at month-end values, and rebalanced based on 5% bands, i.e. whenever the allocation to a given fund was outside of +/- 5% of the target allocation. Portfolio A required rebalancing on 08/31/2008 and 10/31/2008 (and also on the end date of 02/28/2009, which doesn't matter for this analysis). Portfolio B required rebalancing on 02/29/2008, 10/31/2008, and 01/31/2009.

For the entire time period, we have:

Code: Select all

                         Portfolio A      Portfolio B
Cumulative Return          -26.25%          -23.00%
StDev of Monthly Return      3.12%            2.73%
Admittedly, Portfolio B had a slightly better return with a slightly less volatility. Still, it doesn't seem to have stemmed off the fat tail very well.

So, a few questions for proponents of the Portfolio B-type strategy:

1. Is a 60/40 portfolio with TSM-style equities similar in risk to a 40/60 portfolio with all SCV?
2. Do you feel okay that in a terrible bear market, your "non-fat-tail" portfolio had approximately the same performance as a "fat tail" portfolio?

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Wed Aug 24, 2011 1:10 pm

Nisi --

What do you and others mean by "fat tail" in this kind of discussion?

Is it difference in nature of the probability distribution (EG, Cauchy v. Normal) ?

Or is it just bigger standard deviation?

Dick Purcell

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Post by ddb » Wed Aug 24, 2011 1:12 pm

Dick Purcell wrote:Nisi --

What do you and others mean by "fat tail" in this kind of discussion?

Is it difference in nature of the probability distribution (EG, Cauchy v. Normal) ?

Or is it just bigger standard deviation?

Dick Purcell
Not Nisi, but my interpreation has always been the former, i.e. a distribution with similar variance but more tightly centered around the mean.
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

Chuck
Posts: 2082
Joined: Thu May 21, 2009 12:19 pm

Post by Chuck » Wed Aug 24, 2011 1:13 pm

I've seen the suggestion to only rebalance out of the risky side, and never in. But not within the context of a tilted portfolio. This is a very interesting topic.

User avatar
Random Musings
Posts: 5551
Joined: Thu Feb 22, 2007 4:24 pm
Location: Pennsylvania

Post by Random Musings » Wed Aug 24, 2011 1:18 pm

I'm pretty sure the 73-74 bear would give a similar look, but I think the 2000-2002 bear would show portfolio B being the winner. Perhaps one has to look at a larger sample size of bear markets. At that point, it's the "best information available" for this specific look.

But I'm not sure if this can be looked at in a vacuum, I think the longer-term look is more important. Plus, one has to be able to handle the tracking error, as Larry S. has mentioned many a time.

RM

rwwoods
Posts: 1312
Joined: Thu Jun 07, 2007 7:54 pm
Location: The Villages, Florida

Post by rwwoods » Wed Aug 24, 2011 1:19 pm

I look for a smaller downside deviation since the idea is to minimize drawdowns. I am using 15% SCV, 15% EMSCV and 70% 3-5 yr T.
"I'm not so much concerned about the return on my money as the return of my money" - Will Rogers

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Wed Aug 24, 2011 1:23 pm

Ddb --

Thanks for your response.

But if so -- if "fat-tail-ness" is not reflected in standard deviation -- how can I detect "fat-tail-reduction" from the data Nisi provided?

Or more generally, how do I detect and compare portfolios' "fat-tail-ness"?

Dick Purcell

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Post by bob90245 » Wed Aug 24, 2011 2:39 pm

Random Musings wrote:I'm pretty sure the 73-74 bear would give a similar look, but I think the 2000-2002 bear would show portfolio B being the winner.
I think you are correct for the 2000-2002 bear. Small cap value stocks languished in the large cap stock bubble of the late 1990's. And when the bear finally arrived, it was the large caps that took the brunt of the beating.

I also have a chart of the fund components for the Coffeehouse Portfolio (2000-2004):

Image
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Post by ddb » Wed Aug 24, 2011 2:45 pm

Dick Purcell wrote:Ddb --

Thanks for your response.

But if so -- if "fat-tail-ness" is not reflected in standard deviation -- how can I detect "fat-tail-reduction" from the data Nisi provided?

Or more generally, how do I detect and compare portfolios' "fat-tail-ness"?

Dick Purcell
I have no idea. I leave this question up to those who claim that a lower equity allocation (with higher volatility equities) reduces fat tails.

(BTW, Nisi has not yet participated in this thread)

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Wed Aug 24, 2011 3:02 pm

Ddb --

Sorry about that. When I first saw this thread listed in the forum menu, I misread it as launched by Nisi. In my posts above, each time I said Nisi I should have addressed or named you.

As to what is meant and how it's detected when asset allocation is said to reduce fat tail, I remain puzzled.

Dick Purcell

Snowjob
Posts: 1609
Joined: Sun Jun 28, 2009 10:53 pm

Post by Snowjob » Wed Aug 24, 2011 3:07 pm

I would think sector has a bigger impact than size

staythecourse
Posts: 6993
Joined: Mon Jan 03, 2011 9:40 am

Post by staythecourse » Wed Aug 24, 2011 9:03 pm

Okay a couple of criticisms:

1. If your going to attack a portfolio rec. you have to get it right: 30% equity in SCV, ISV, ESV+ 70% 2 yr. treasuries. So your example of Mr. Swedroe's portfolio is not accurate.

2. A window as short as 2 years and a couple months, as your example has, is a small sample size to mean anything.

Good luck.
"The stock market [fluctuation], therefore, is noise. A giant distraction from the business of investing.” | -Jack Bogle

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

high tilt low beta

Post by larryswedroe » Wed Aug 24, 2011 9:38 pm

With rebalancing annually
I looked at period 73-10 and 100% S&P 500 returned 10% with SD of about 18% and 32% SV and 68% 1 year Treasury (think 5 year would have looked even better but travelling and doing this from memory and don't have access to data base) was something like 9.9% with less than half the SD. Worst year was well less than half the worst year for the S&P and best year was more than half (again from memory).

Now add global diversification to include portfolios like those in my books (Not as high tilt as the extreme high tilts I use) and something like 34% equity you had same return with about third the SD and worst loss of about 30% as bad as all equity and best year more than half

The data I have looked at shows you cut the worst losses more than you give up the best gains, but both "tails" are truncated.

Jut finished talk at Bradley University in Peoria

I hope this helps

User avatar
Noobvestor
Posts: 4767
Joined: Mon Aug 23, 2010 1:09 am
Contact:

Re: high tilt low beta

Post by Noobvestor » Thu Aug 25, 2011 1:48 am

pwm112 wrote:
larryswedroe wrote:With rebalancing annually
I looked at period 73-10 and 100% S&P 500 returned 10% with SD of about 18% and 32% SV and 68% 1 year Treasury (think 5 year would have looked even better but travelling and doing this from memory and don't have access to data base) was something like 9.9% with less than half the SD. Worst year was well less than half the worst year for the S&P and best year was more than half (again from memory).

Now add global diversification to include portfolios like those in my books (Not as high tilt as the extreme high tilts I use) and something like 34% equity you had same return with about third the SD and worst loss of about 30% as bad as all equity and best year more than half

The data I have looked at shows you cut the worst losses more than you give up the best gains, but both "tails" are truncated.
:shock: This is a perfect example of what happens when people misunderstand statistical terminology. All you've done is transformed relatively Gaussian tails into more Cauchy-like tails. You think the tails are truncated or thinner because the sample SD is lower, but in reality you've just "grafted the fat" further out on the tail where the SD doesn't measure it (which is the meaning of the term "fat tails"...the "fatness" refers to the infinitely tapered ends of the tails which are too fat to be ignored as one would normally do with infinitesimal things, so the approximation/measurement error becomes a huge issue). You've increased, not decreased, your fat tail risk because you've fattened the tails!

Even aside from the statistics of it, this should be clear from a pure engineering perspective. You've taken system components that would each somewhat function by themselves and replaced them with non-redundant components that depend on each other; So if any one of the components fails your entire strategy fails. Your design is not fail-safe, which means it's a bad design!
ddb wrote:I have no idea. I leave this question up to those who claim that a lower equity allocation (with higher volatility equities) reduces fat tails.
Clearly those making this claim (and selling it to their clients and book readers) don't know what the term "fat tails" even means! :shock:
I mean this in the best possible way but: can you repeat that in English please :shock:
"In the absence of clarity, diversification is the only logical strategy" -= Larry Swedroe

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Re: high tilt low beta

Post by ddb » Thu Aug 25, 2011 7:06 am

larryswedroe wrote:With rebalancing annually
I looked at period 73-10 and 100% S&P 500 returned 10% with SD of about 18% and 32% SV and 68% 1 year Treasury (think 5 year would have looked even better but travelling and doing this from memory and don't have access to data base) was something like 9.9% with less than half the SD. Worst year was well less than half the worst year for the S&P and best year was more than half (again from memory).
So, is a tilted portfolio with lower equity allocation a free lunch? Or was this merely a best-case scenario period, where investors were well-rewarded for the small and value premiums, plus treasury returns over the backtested period benefitted from interest rates which were in steady decline?

IMO, I do not think it is rational to expect that two passive low-cost portfolios can have the same expected return with vastly different expected variances going forward. I am frankly surprised to see backtesting of this nature as some sort of evidence for what to expect going forward. (I backtested in my original post simply to put some simple analysis to former claims that 2008 was a good year for tilted lower-equity portfolios)

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

Roy
Posts: 970
Joined: Wed Sep 10, 2008 9:34 am

Re: Minimizing fat tail risk w/ lower equity allocation - my

Post by Roy » Thu Aug 25, 2011 7:28 am

ddb wrote: Portfolio A: 60% DFA US Large Cap, 40% DFA 2-Year Global Fixed Income
Portfolio B: 40% DFA US Small Cap Value, 60% DFA 2-Year Global Fixed Income

1. Is a 60/40 portfolio with TSM-style equities similar in risk to a 40/60 portfolio with all SCV?
DDB, what is the ticker you used for the DFA US Large Cap, and is that equity profile similar to TSM? While the outcomes may be similar over this time period, perhaps just using TSM or (S&P 500, like Larry did) for the comparison would be better. I also think Larry uses fewer equities—like only 25-30%.
Last edited by Roy on Thu Aug 25, 2011 7:39 am, edited 1 time in total.

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 7:37 am

Dick Purcell wrote:What do you and others mean by "fat tail" in this kind of discussion?
Leptokurtosis.
Simplify the complicated side; don't complify the simplicated side.

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 7:38 am

Dick Purcell wrote:Or more generally, how do I detect and compare portfolios' "fat-tail-ness"?
Compare their kurtoses.
Simplify the complicated side; don't complify the simplicated side.

User avatar
woof755
Posts: 3160
Joined: Sun Aug 05, 2007 2:03 pm
Location: Honolulu

Re: high tilt low beta

Post by woof755 » Thu Aug 25, 2011 7:45 am

pwm112 wrote: :shock: This is a perfect example of what happens when people misunderstand statistical terminology. All you've done is transformed relatively Gaussian tails into more Cauchy-like tails. You think the tails are truncated or thinner because the sample SD is lower, but in reality you've just "grafted the fat" further out on the tail where the SD doesn't measure it (which is the meaning of the term "fat tails"...the "fatness" refers to the infinitely tapered ends of the tails which are too fat to be ignored as one would normally do with infinitesimal things, so the approximation/measurement error becomes a huge issue). You've increased, not decreased, your fat tail risk because you've fattened the tails!

Even aside from the statistics of it, this should be clear from a pure engineering perspective. You've taken system components that would each somewhat function by themselves and replaced them with non-redundant components that depend on each other; So if any one of the components fails your entire strategy fails. Your design is not fail-safe, which means it's a bad design!
ddb wrote:I have no idea. I leave this question up to those who claim that a lower equity allocation (with higher volatility equities) reduces fat tails.
Clearly those making this claim (and selling it to their clients and book readers) don't know what the term "fat tails" even means! :shock:
I am interested in this thread, and agree that I'd like to hear you further explain this, but I can't say your tone helps very much. You have been a member for 3 weeks and you are insulting someone who is a vital resource on this board.
"By singing in harmony from the same page of the same investing hymnal, the Diehards drown out market noise." | | --Jason Zweig, quoted in The Bogleheads' Guide to Investing

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

few thoughts

Post by larryswedroe » Thu Aug 25, 2011 7:47 am

First, the low beta high tilt portfolio is meant to cut the risk of what Taleb called the Black Swan type events, the potential for large losses--the term fat tail is just used to indicate that type event(yes technically it is not correct)

Second, the logic is very simple. If you have a high tilt low beta portfolio and invest the remainder in safe fixed income investments it is simply impossible to have the same size losses in bad times. Just look at a 100% equity portfolio vs the 1/3 equity with high tilt. If stocks drop 50% you lose 50% with the 100% equity. With the 30% equity I don't care how negative the size and value premiums are you cannot lose more than 30%. This of course is extreme example, but makes the point that the low beta and high tilt cuts the potential for the big loss while at same time cutting opportunity for the large gain, without changing the expected return (of course we can only estimate expected returns)

Now historically the high tilt low beta portfolio has enabled the investor to avoid much more of the left side risk while giving up much less of the right side risk. Will that be true in future, I don't know but I do believe that by diversifying the sources of risk away from beta to other factors you are likely to see that happen (you get a diversification benefit just like you do when you diversify from the S&P to include international and add other asset classes).

Now if you also believe that value is a behavioral story, or at least partly one, then YES it is a FREE LUNCH. And my own view is that the value premium is to a good degree a risk story but it has been too high to be totally explained by risk, so part is behavioral--a free stop at dessert tray if you will

User avatar
woof755
Posts: 3160
Joined: Sun Aug 05, 2007 2:03 pm
Location: Honolulu

Re: few thoughts

Post by woof755 » Thu Aug 25, 2011 8:12 am

larryswedroe wrote:With the 30% equity I don't care how negative the size and value premiums are you cannot lose more than 30%.
...But if you continually rebalance to maintain your 30 / 70 allocation, you could conceivably lose more than 30%. I think this is ddb's question--does rebalancing eliminate the benefit of such a portfolio? It seems hard to conceive of a rebalancing interval that makes sense--unless you do as mentioned earlier in the thread--rebalance out of the high-risk class, but not back into it.
"By singing in harmony from the same page of the same investing hymnal, the Diehards drown out market noise." | | --Jason Zweig, quoted in The Bogleheads' Guide to Investing

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Re: few thoughts

Post by ddb » Thu Aug 25, 2011 8:13 am

larryswedroe wrote:Second, the logic is very simple. If you have a high tilt low beta portfolio and invest the remainder in safe fixed income investments it is simply impossible to have the same size losses in bad times. Just look at a 100% equity portfolio vs the 1/3 equity with high tilt. If stocks drop 50% you lose 50% with the 100% equity. With the 30% equity I don't care how negative the size and value premiums are you cannot lose more than 30%. This of course is extreme example, but makes the point that the low beta and high tilt cuts the potential for the big loss while at same time cutting opportunity for the large gain, without changing the expected return (of course we can only estimate expected returns).
Larry, with all due respect, the above paragraph implies that you don't understand the impact of rebalancing (which I know isn't actually the case). A portfolio with even 1% equities can hypothetically drop 100% if equities drop 100%. In a steadily-declining stock environment, rebalancing can have a very negative impact on returns.

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell » Thu Aug 25, 2011 9:23 am

Magician --

Your Kurtosis answers illustrate my original question.

In discussions of asset allocation to "reduce fat tails," I see reports of mean and SD but not Kurtosis, which makes me wonder if what is meant is reduction of SD rather than changing nature of distribution as measured by Kurtosis.

And I wonder about sufficiency of data for meaningful comparisons of Kurtoses.

Not that reducing SD is bad -- it is certainly good. I'm just seeking my own better understanding of the benefit being sought.

? ? ?

Dick Purcell

User avatar
Aptenodytes
Posts: 3762
Joined: Tue Feb 08, 2011 8:39 pm

Re: few thoughts

Post by Aptenodytes » Thu Aug 25, 2011 9:31 am

ddb wrote: A portfolio with even 1% equities can hypothetically drop 100% if equities drop 100%. In a steadily-declining stock environment, rebalancing can have a very negative impact on returns.
- DDB
This doesn't smell right to me. A quick test with a spreadsheet shows that if equities drop 50% per year for 50 straight years, and bonds are flat, then after 50 years a 1/99 portfolio will drop in value by about 22%, if you rebalance once a year. You could argue that if you rebalance more often the losses might be bigger. True, but you wouldn't do that. The smaller your equity percentage, the less often you have to rebalance. With anything like conventional bands a 1/99 portfolio would seldom need to rebalance.

Even if equities drop 100% per year (which isn't even possible) then a 1/99 portfolio loses about 38% of its value.

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 9:43 am

Dick Purcell wrote:Your Kurtosis answers illustrate my original question.

In discussions of asset allocation to "reduce fat tails," I see reports of mean and SD but not Kurtosis, which makes me wonder if what is meant is reduction of SD rather than changing nature of distribution as measured by Kurtosis.
I think - as you apparently do, too - that when many finance-types talk about "fat tails" they really don't understand what they're discussing. Most, I suspect, wouldn't recognize kurtosis if they stepped on it.

(I'm not talking about the folks here: we have some pretty savvy finance-types. I'm talking about the vast sea of people giving financial advice who haven't a clue.)
Dick Purcell wrote:And I wonder about sufficiency of data for meaningful comparisons of Kurtoses.
Does one need more data to be able to compare higher moments (e.g., skewness, kurtosis) than to be able to compare lower moments (e.g., mean, standard deviation)? I, for one, don't know.
Dick Purcell wrote: ? ? ?
Yup.
Last edited by magician on Tue Dec 06, 2011 8:20 am, edited 1 time in total.
Simplify the complicated side; don't complify the simplicated side.

User avatar
Aptenodytes
Posts: 3762
Joined: Tue Feb 08, 2011 8:39 pm

Post by Aptenodytes » Thu Aug 25, 2011 9:49 am

magician wrote:
Dick Purcell wrote:What do you and others mean by "fat tail" in this kind of discussion?
Leptokurtosis.
Taking the overall conversation into account, it seems that there are three uses at play:

1) Leptokurtosis. The tail is "fat" compared to a normal distribution.

2) High standard deviation -- this is what Swedroe says he means by it. As you travel away from the mean, the height of the curve makes it look "fat" compared to a distribution with lower SD.

3) Low probability events regardless of the governing distribution, or even regardless of any known distribution whatsoever (e.g. I saw a column this morning referring to the possibility that the U.S. will convene a new constitutional convention as a black swan event. We don't know the probability distribution of such a thing, but we can say it has low probability.

Under the circumstances it seems we should avoid the term altogether and instead d

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 9:55 am

Aptenodytes wrote:. . . e.g. I saw a column this morning referring to the possibility that the U.S. will convene a new constitutional convention as a black swan event. We don't know the probability distribution of such a thing, but we can say it has low probability.
As a single event that will either occur or else it won't, there isn't a probability distribution: just a single (unknown) probability that it will happen.
Aptenodytes wrote:Under the circumstances it seems we should avoid the term altogether and instead d
Are you tormenting us intentionally?

This is like the end-of-season cliffhanger; tune in next season to see the outcome.

;)
Simplify the complicated side; don't complify the simplicated side.

User avatar
Aptenodytes
Posts: 3762
Joined: Tue Feb 08, 2011 8:39 pm

Post by Aptenodytes » Thu Aug 25, 2011 10:03 am

magician wrote:
Aptenodytes wrote:. . . e.g. I saw a column this morning referring to the possibility that the U.S. will convene a new constitutional convention as a black swan event. We don't know the probability distribution of such a thing, but we can say it has low probability.
As a single event that will either occur or else it won't, there isn't a probability distribution: just a single (unknown) probability that it will happen.
Aptenodytes wrote:Under the circumstances it seems we should avoid the term altogether and instead d
Are you tormenting us intentionally?

This is like the end-of-season cliffhanger; tune in next season to see the outcome.

;)
Sorry -- real work intruded and I hit send prematurely. I meant we should not use "fat tail" and instead use unambiguous terms we all agree on (rare event, high standard deviation, skewed distribution, etc.).

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 10:23 am

pwm112 wrote:Well it could happen twice or one could consider the distribution of the different ways it could happen (ie happening badly with twice the impact vs happening mildly with half the impact).
That's changing the original statement. Aptenodytes was taking about a single event - convening a constitutional convention - occurring or not. Either a constitutional convention is convened, or else it isn't. If it happens twice, then it happens: outcome = TRUE.
pwm112 wrote:That's precisely what makes fat tails tricky: even if all the fat tail events are each almost impossible, the consolidated impact may be non-negligible (or even guaranteed). That's what makes statistical analysis so nasty: unlike other types of science/engineering, when it comes to statistics, infinitely small may not be small enough.
Yes, there's a difference between the probability of an event occurring (a single number) and the probability distribution of the possible consequences should that event occur.

Yesterday I interviewed a brace of engineers on the possibility of a particular risk event occurring on their construction project. The probability of the event occurring is a single number. The range of possible impacts has a probability distribution.
Simplify the complicated side; don't complify the simplicated side.

etm
Posts: 138
Joined: Sun May 23, 2010 1:31 pm

Post by etm » Thu Aug 25, 2011 11:11 am

Larry's portfolio reminds me a little bit of Vanguard's Wellsley fund--68% corporate bonds and 1/3 valuish large cap stocks. Obviously Larry's portfolio is different--treasuries plus the higher expected return of SCV, EM, etc. They both are like a portfolio on lithium--your highes aren't too high but you avoid very destructive lows.

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

ddb

Post by larryswedroe » Thu Aug 25, 2011 1:04 pm

with all due respect I certainly due understand the issues related to rebalancing.
First the lower beta high tilt has less of that risk than a higher beta lower tilt portfolio has. It is clearly relatively less risky, cut the likelihood of big losses.
Sure you can continue to rebalance your self into bankruptcy, but plans are living documents and one should alter them to prevent such an occurrence. For example, many clients cut expenses during the bear market so they would not run as great a risk of running out of assets.

Second, one can adopt a policy like I do which has rebalancing only when exceed equity target, not when below it. Again, cuts risk of the left side of the distribution while cutting opportunity for the right side. Again, depends on marginal utility of wealth and need to take risk

Best wishes
Larry

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

woof

Post by larryswedroe » Thu Aug 25, 2011 1:06 pm

As I explained to DDB, the same risk you have with a higher beta lower tilt portfolio is there that you describe, only more so. So however you want to look at the risk of large losses is less. The lower beta higher tilt portfolio cuts off the "risks" of the tails of the lower tilt higher beta portfolios. You have less chance of getting the good right tail and less chance of getting the bad left tail

Also as I noted above, one does not necessarily have to rebalance and one can alter plan, adopting Plan Bs if the worst case does show up.

I hope that is helpful

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

pwm

Post by larryswedroe » Thu Aug 25, 2011 1:09 pm

by definition if you will there is no fat tail of a riskless asset. Either it is riskless or not. And btw, if that risk where to show up (US defaults on tbills) the risk on the equity side would almost certainly be FAR FAR worse. Thus the answer IMO remains the same, the low beta high tilt portfolio has less risk of the big loss (and less opportunity for the big gain--which risk averse investors care far less about in general)

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

one last comment

Post by larryswedroe » Thu Aug 25, 2011 1:11 pm

My high tilt low beta portfolio is very much like Taleb's suggested portfolio---most of portfolio in very low risk assets and small amount in highest risk assets.

Best wishes
Larry

Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 10:36 am

Post by Bongleur » Thu Aug 25, 2011 1:19 pm

magician wrote: Yesterday I interviewed a brace of engineers on the possibility of a particular risk event occurring on their construction project. The probability of the event occurring is a single number. The range of possible impacts has a probability distribution.
Risk vs Hazard ; probability vs bad consequences
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Bongleur

Post by larryswedroe » Thu Aug 25, 2011 1:29 pm

The lesson taught by Pascal's Wager is that the consequences of one's decisions should dominate the probabilities of outcomes

Best wishes
Larry

Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 10:36 am

Re: Minimizing fat tail risk w/ lower equity allocation - my

Post by Bongleur » Thu Aug 25, 2011 1:31 pm

ddb wrote: I looked at a single example to see how this would have worked. I used the period of 12/31/2006 to 02/28/2009. I compared two portfolios:
- DDB
What are the year by year results if you rebalance out of equity gains but not losses?

Wondering how a withdrawal stage would look, so you could also pick some reasonable percent as being spent as well.

I suppose the best strategy is to spend from bond yields first, then take from equity, and then the remainder of equity gains is "eligible" to be moved to bonds if it still exceeds the band rule.

Hmmm... When equities are doing poorly... spending from equity prevents that amount from being lost during the following period, which you assume will also be bad (follow momentum since you don't know when it will change).

You might need to devise a rule to re-fund the equity once momentum has changed...

But given that you are using a heavy SV tilt, you have lowered the expectation that there will be a lot of sequential poor returns for the equity side, so you don't expect to re-fund the equity very many times in your 30 to 40 year timespan.
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Re: ddb

Post by ddb » Thu Aug 25, 2011 1:36 pm

larryswedroe wrote:with all due respect I certainly due understand the issues related to rebalancing.
First the lower beta high tilt has less of that risk than a higher beta lower tilt portfolio has. It is clearly relatively less risky, cut the likelihood of big losses.
Sure you can continue to rebalance your self into bankruptcy, but plans are living documents and one should alter them to prevent such an occurrence. For example, many clients cut expenses during the bear market so they would not run as great a risk of running out of assets.
Right. And of course, the flexibility also applies to a higher beta portfolio. Can still cut off the left tail by implementing some sort of "plan b". What you suggest is not unique to the low beta strategy.
Second, one can adopt a policy like I do which has rebalancing only when exceed equity target, not when below it. Again, cuts risk of the left side of the distribution while cutting opportunity for the right side. Again, depends on marginal utility of wealth and need to take risk
This is the first I've heard of an "upside only" rebalancing strategy. Have you written about this before, and I've just missed it? Such a strategy would seem to eliminate some of the benefits of a more typical rebalancing strategy. Basically, you can never "buy low" unless adding new money.

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

tms
Posts: 400
Joined: Mon Dec 15, 2008 11:44 am

Re: ddb

Post by tms » Thu Aug 25, 2011 1:49 pm

ddb wrote: This is the first I've heard of an "upside only" rebalancing strategy. Have you written about this before, and I've just missed it? Such a strategy would seem to eliminate some of the benefits of a more typical rebalancing strategy. Basically, you can never "buy low" unless adding new money.

- DDB
Interesting. With this strategy, I wonder if interest and dividend income is reinvested in the same fund, or allowed to go to cash and then redeployed based on the target allocation.

In that case, there would be some buying low, but you would never use fixed income principal to buy stocks.

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 1:50 pm

Bongleur wrote:
magician wrote: Yesterday I interviewed a brace of engineers on the possibility of a particular risk event occurring on their construction project. The probability of the event occurring is a single number. The range of possible impacts has a probability distribution.
Risk vs Hazard ; probability vs bad consequences
?
Simplify the complicated side; don't complify the simplicated side.

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

ddb

Post by larryswedroe » Thu Aug 25, 2011 2:02 pm

I don't know if I have ever mentioned that specifically here though I know I have discussed it with many people.

And yes adopting Plan B is critical part or should be of all IPS's, (though my guess is absent from the vast majority)

BTW- one last point, as prices go lower you are buying at times of higher and higher expected returns (which of course doesn't mean you will get them)
Best
Larry

Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 10:36 am

Post by Bongleur » Thu Aug 25, 2011 2:23 pm

Risk is the probability of an occurrence, which can usually be quantified more accurately than the cost of its consequences (Hazard)
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
Topic Author
ddb
Posts: 5509
Joined: Mon Feb 26, 2007 12:37 pm
Location: American Gardens Building, West 81st St.

Re: ddb

Post by ddb » Thu Aug 25, 2011 2:29 pm

larryswedroe wrote:I don't know if I have ever mentioned that specifically here though I know I have discussed it with many people.

And yes adopting Plan B is critical part or should be of all IPS's, (though my guess is absent from the vast majority)

BTW- one last point, as prices go lower you are buying at times of higher and higher expected returns (which of course doesn't mean you will get them)
Agreed, somewhat. Too bad your "upside only" rebalancing strategy can't take advantage of the higher expected returns!

I must admit, I'm thoroughly taken aback by this rebalancing idea of yours. Definitely never heard of such a thing.

- DDB
"We have to encourage a return to traditional moral values. Most importantly, we have to promote general social concern, and less materialism in young people." - PB

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

ddb

Post by larryswedroe » Thu Aug 25, 2011 2:36 pm

You are right about not being able to take advantage of higher expected returns--though I do admit at some level I could be again tempted to buy--maybe P/Es like 5 or 6.

But would not suggest this to anyone else unless they had same need to take risk as I did and virtually zero marginal utility of wealth. I won the game already and so I have no need to play. Basically in capital preservation mode.

Best wishes
Larry

User avatar
magician
Posts: 1571
Joined: Mon May 02, 2011 1:08 am
Location: Yorba Linda, CA
Contact:

Post by magician » Thu Aug 25, 2011 2:39 pm

Bongleur wrote:Risk is the probability of an occurrence, which can usually be quantified more accurately than the cost of its consequences (Hazard)
I'm not sure that the probability of occurrence can usually be quantified more accurately than the cost of consequences if the risk event occurs, but I am sure that risk isn't (merely) the probability of an occurrence.

For example, if the risk is that we may have to tie the foundation into the bedrock with steel dowels, it's pretty easy to quantify the cost of the consequence: the cost of the dowels, the labor to drill the bedrock and install the dowels, and so on. Whether the nature of the rock sixty feet down will require such dowels (or, whether a given inspector will insist on it or not) is, I believe, much more difficult to quantify.

But we digress.
Simplify the complicated side; don't complify the simplicated side.

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: one last comment

Post by richard » Thu Aug 25, 2011 2:50 pm

larryswedroe wrote:My high tilt low beta portfolio is very much like Taleb's suggested portfolio---most of portfolio in very low risk assets and small amount in highest risk assets.
Also see Bodie's suggested portfolio - lots of TIPS plus long-term equity call options.

larryswedroe
Posts: 16001
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Richard

Post by larryswedroe » Thu Aug 25, 2011 3:24 pm

I am aware of Bodie's similar idea but don't like idea of buying options so did not mention that
Best
Larry

User avatar
Joe S.
Posts: 479
Joined: Sat May 05, 2007 12:11 pm

Post by Joe S. » Thu Aug 25, 2011 3:58 pm

Dick Purcell wrote:Nisi --

What do you and others mean by "fat tail" in this kind of discussion?

Is it difference in nature of the probability distribution (EG, Cauchy v. Normal) ?

Or is it just bigger standard deviation?

Dick Purcell
All these curves have the same mean and standard deviations. The blue curve is the standard Gaussian curve with 0 kurtosis. The green curves (with increased kurtosis) show more extreme events, or fat tails.

Image

Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 10:36 am

Post by Bongleur » Thu Aug 25, 2011 4:00 pm

magician wrote:
Bongleur wrote:Risk is the probability of an occurrence, which can usually be quantified more accurately than the cost of its consequences (Hazard)
I'm not sure that the probability of occurrence can usually be quantified more accurately than the cost of consequences if the risk event occurs, but I am sure that risk isn't (merely) the probability of an occurrence.

For example, if the risk is that we may have to tie the foundation into the bedrock with steel dowels, it's pretty easy to quantify the cost of the consequence: the cost of the dowels, the labor to drill the bedrock and install the dowels, and so on. Whether the nature of the rock sixty feet down will require such dowels (or, whether a given inspector will insist on it or not) is, I believe, much more difficult to quantify.

But we digress.
The risk is the probability of a collapse without the dowels.
The hazard is the cost of the consequences of the collapse.
Having to use them or not is merely a requirement or an option
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

richard
Posts: 7961
Joined: Tue Feb 20, 2007 3:38 pm
Contact:

Re: Richard

Post by richard » Thu Aug 25, 2011 4:03 pm

larryswedroe wrote:I am aware of Bodie's similar idea but don't like idea of buying options so did not mention that
What don't you like about long-term options? For that matter, is there a liquid market in LEAPS?

Post Reply