Valuation-based market timing with PE10 can improve returns?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
novastepp
Posts: 340
Joined: Sun Apr 03, 2011 8:13 pm
Location: Indiana

Post by novastepp »

Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

novastepp wrote:Does the study take into account a strict rebalancing strategy that many Bogelheads use; say a 5% deviation or an annual effort?

I see that the study crosses its numbers with buy and hold, but I do not see mention as to the always-forgotten rebalancing strategy.
the study evaluates the following strategies:

- 100% equities buy-and-hold forever
- 50/50 balanced fund, rebalanced annually
- various market-timing strategies

The market-timing strategies produce much higher returns than the 50/50 balanced fund, with similar risk.

The market-timing strategies produce similar returns to the 100% equities buy-and-hold forever approach, but with much less risk.
User avatar
William Million
Posts: 1130
Joined: Wed May 05, 2010 4:41 am
Location: A Deep Mountain

Post by William Million »

Even though I don't think much of this study and consider PE10 pseudo-science, I admit there are some individuals who are consistently successful at timing the market. However, these individuals don't see to rely on any single valuation metric. They look at all of them, and then add some analysis on top.

There are too many events that affect the direction of stocks that have nothing to do with PE10. With perfect hindsight, we know that real estate was overvalued in 2007. Ultimately, this overvaluation - and the overextension of many banks - caused a deep recession that devastated the stock market. PE10 did not predict that chain of events. Peter Schiff did - because he looked and weighed many factors.

Having said that, most investors should stay out of market timing that goes beyond rebalancing.
User avatar
Raging Mage
Posts: 118
Joined: Fri Dec 31, 2010 12:37 pm

Post by Raging Mage »

yobria wrote:Japan had high P/Es in the 1950s and 60s, yet very strong stock performance for the next 30 years.
As a relative investing newbie, this is an observation that gives me pause when considering Pfau's research. For many investors, 30 years is the vast majority of their pre-retirement investing lifetimes. Even if, over a period of 139 years, the backtested market-timing strategy based on PE10 outperforms buy-and-hold, it's entirely possible that the results during your particular range of investment years will be different. What if you come face-to-face with a prolonged period of high P/E ratios and strong stock performance, and your portfolio light on stocks doesn't provide enough returns to fund your retirement?

Another concern I have is this: Normally, it's recommended that one increases his allocation to bonds (and decreases his stock holdings) as he advances in years, which of course makes sense, given the older investor's reduced capacity to take risk. Can this requirement coexist with a market-timing strategy that involves large swings between stocks and bonds based on PE10? What happens if an older investor following the market-timing strategy increases his allocation to stocks during a period of low P/E, only to watch the stock market tank and his nest egg get crushed while incapable of replacing the losses?

Again, I'm a newbie, and maybe my concerns don't have any merit. Don't get me wrong--I think it's fantastic that Pfau is putting so much work into this research. Maybe there's really something to it, and further research will back his documented market-timing approach. It's worth finding out. But until much more work is done on this front and I have reason to change my thinking, I'm certainly not going to flee from the time-tested, battle-approved buy/hold/rebalance approach, and I think others should be similarly cautious.
"Here are three easy ways to beat the market: Deception, irrelevance and bad math." - Alexander Green, Investment U
User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 7:51 pm

Post by bob90245 »

fredflinstone wrote:
555 wrote:
fredflinstone wrote:
555 wrote:Could someone please explain what the fuss is all about.
"Professor Pfau has identified a valuation-based asset allocation strategy which provides risk-adjusted returns that are greatly superior to the 50/50 balanced fund advocated by many Bogleheads."
I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
I don't belive taking more risk to earn higher returns is a refutation of Boglehead investing philosopy. As I already stated upthread, using "standard deviation" as measure of "risk" in Wade's paper is misleading. Here is a repost if this point isn't clear to you:

http://www.bogleheads.org/forum/viewtop ... 55#1066955
richard wrote:
bob90245 wrote:The P/E10 backtest tool will tell you to increase your stock allocation when business risk is high -- the very reason that causes P/E10 to fall. For those with short memories, I would imagine that the P/E10 tool would have told investors to load up on stocks during the financial meltdown of 2008-2009 when P/E10 fell below 15. However, I don't remember anyone telling investors that stocks were no longer as risky, do you?
This is a fundamental point. Low p/e means the market is telling you risk is high (and vice versa). You would expect to get higher returns because risk is higher. Of course, expected returns do not necessarily equal actual returns, or it would not be the case that risk was higher.

Getting higher returns by taking on more risk should not be surprising.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.
555
Posts: 4955
Joined: Thu Dec 24, 2009 6:21 am

Post by 555 »

fredflinstone wrote:
555 wrote:
fredflinstone wrote:
555 wrote:Could someone please explain what the fuss is all about.
"Professor Pfau has identified a valuation-based asset allocation strategy which provides risk-adjusted returns that are greatly superior to the 50/50 balanced fund advocated by many Bogleheads."
I think you are making a mountain out of a molehill.
"The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal."
Okay let me rephrase.
You are making a planet out of a pebble. :lol:
DUCY?
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

Raging Mage wrote:Normally, it's recommended that one increases his allocation to bonds (and decreases his stock holdings) as he advances in years, which of course makes sense, given the older investor's reduced capacity to take risk. Can this requirement coexist with a market-timing strategy that involves large swings between stocks and bonds based on PE10? What happens if an older investor following the market-timing strategy increases his allocation to stocks during a period of low P/E, only to watch the stock market tank and his nest egg get crushed while incapable of replacing the losses?
It would be straightforward to incorporate both age and PE10 into one's asset allocation decisions. For example, you could start with 100-AGE in equities then adjust upward or downward based on PE10.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

bob90245 wrote:
fredflinstone wrote:
555 wrote:
fredflinstone wrote:
555 wrote:Could someone please explain what the fuss is all about.
"Professor Pfau has identified a valuation-based asset allocation strategy which provides risk-adjusted returns that are greatly superior to the 50/50 balanced fund advocated by many Bogleheads."
I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
I don't belive taking more risk to earn higher returns is a refutation of Boglehead investing philosopy. As I already stated upthread, using "standard deviation" as measure of "risk" in Wade's paper is misleading. Here is a repost if this point isn't clear to you:

http://www.bogleheads.org/forum/viewtop ... 55#1066955
richard wrote:
bob90245 wrote:The P/E10 backtest tool will tell you to increase your stock allocation when business risk is high -- the very reason that causes P/E10 to fall. For those with short memories, I would imagine that the P/E10 tool would have told investors to load up on stocks during the financial meltdown of 2008-2009 when P/E10 fell below 15. However, I don't remember anyone telling investors that stocks were no longer as risky, do you?
This is a fundamental point. Low p/e means the market is telling you risk is high (and vice versa). You would expect to get higher returns because risk is higher. Of course, expected returns do not necessarily equal actual returns, or it would not be the case that risk was higher.

Getting higher returns by taking on more risk should not be surprising.
I don't understand this at all. Are you saying risk in the stock market was lowest when Nasdaq was at 5000 back in March 2000? Really?
User avatar
Mel Lindauer
Moderator
Posts: 35757
Joined: Mon Feb 19, 2007 7:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer »

letsgobobby wrote:Using a 401k and an IRA, most Bogleheads have substantial tax-advantaged accounts at their disposal. Thus many of the concerns Mel and Taylor have highlighted regarding fees and taxes are no longer dealbreakers.
It appears that you overlooked my message to Wade, reminding him that his study supposedly shows market timing to be better than buy-and-hold over a period of time when tax-deferred accounts weren't even available and sales loads and trading costs were extremely high compared to today. Thus, all trading would have had to be done in taxable accounts at these much higher costs.

And, higher tax rates were in effect for short-term capital gains, long-term capital gains and ordinary income. So if you want to use a time period to show something's better than another, then you need to use the costs that an investor actually incurred over that period using that strategy. And Wade didn't do that.

Ignoring the actual costs that the market-timing scheme incurred renders the study useless IMO.
Best Regards - Mel | | Semper Fi
User avatar
Raging Mage
Posts: 118
Joined: Fri Dec 31, 2010 12:37 pm

Post by Raging Mage »

fredflinstone wrote:It would be straightforward to incorporate both age and PE10 into one's asset allocation decisions. For example, you could start with 100-AGE in equities then adjust upward or downward based on PE10.
Forgive me, but I'm not convinced that addressing both factors in one's stock/bond ratio is really that simple. Mind you, there was already plenty of debate about appropriate stock/bond ratios for various age groups before adding the PE10 consideration into the mix. It's also not clear yet what impact adjustment for age would have on Pfau's strategy. We're talking about another level of complexity that, IMO, merits much closer examination and research.
"Here are three easy ways to beat the market: Deception, irrelevance and bad math." - Alexander Green, Investment U
User avatar
StoneReader
Posts: 394
Joined: Tue Feb 27, 2007 6:42 pm

Overvalued by Any Measure

Post by StoneReader »

A worrisome thing is that the stock market appears to be substantially overvalued, e.g. by ~70%, on any measure. Is this really the time to be indexing blindly? In any type of investment approach, one has to judge whether the asset has a reasonable chance of any positive return

q is the ratio between the value of companies according to the stock market and their net worth measured at replacement cost. CAPE is the cyclically adjusted PE ratio for the market. http://www.smithers.co.uk/page.php?id=34

Image

Shiller's 10-Year Average PE is now 23.52 at the end of last week.
Image
http://www.multpl.com/

From Pfau's paper:
Image


Note that according to Pfau's criterion, the portfolio is now in 100% cash.

A serious flaw with the Pfau's paper is that it looks in a rear-view mirror and uses average PE data from the past. If we knew what the average PE would be in the future, we could use that knowledge very profitably with much less risk.

The very long term comparison between Buy & Hold and trading with the knowledge of what the average PE would be in the 140 years ahead is not encouraging, especially when taxes are considered. This would say to stay with Buy&Hold Indexing if you were an institution with an infinite life.

Image

For an individual with a finite lifespan, on the other hand, 14 trades in 140 years is not excessive and the trading approach would have surely not seen many of the drastic drawdowns that could have impoverished an individual investor for the remaining years of his life
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

Mel Lindauer wrote:Ignoring the actual costs that the market-timing scheme incurred renders the study useless IMO.
I respectfully disagree. I don't care about taxes or the cost of indexing in 1890. To the extent that I have investments in taxable accounts, I care about taxes today and going forward. Costs for a modern day investor are negligible (except for thinly-traded microcaps) so I would prefer an analysis that ignores those. I guess we will agree to disagree on this.
User avatar
docneil88
Posts: 934
Joined: Mon Apr 30, 2007 3:39 pm
Location: Taxable

Re: Valuation-based market timing with PE10 can improve retu

Post by docneil88 »

fredflinstone wrote:Your figures are wrong. Look at Figure II on page 8 of the study. "Fixed 50/50" refers to a 50/50 balanced fund. "MT 0-100" refers to various all-or-nothing market timing strategies.
You're right. I confused SD with return on the 50/50 balanced fund. I edited my post above. Best, Neil
User avatar
Mel Lindauer
Moderator
Posts: 35757
Joined: Mon Feb 19, 2007 7:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer »

fredflinstone wrote:
Mel Lindauer wrote:Ignoring the actual costs that the market-timing scheme incurred renders the study useless IMO.
I respectfully disagree. I don't care about taxes or the cost of indexing in 1890. To the extent that I have investments in taxable accounts, I care about taxes today and going forward. Costs for a modern day investor are negligible (except for thinly-traded microcaps) so I would prefer an analysis that ignores those. I guess we will agree to disagree on this.
Well, if the costs that the market timing investor actually incurred during all those years weren't considered, then it's impossible to say that the market-timing scheme outperformed during those periods. That's basic 101 stuff.

You can't include periods in a study that show outperformance but ignore the fact that the study didn't take into account that those market-timing investos had to pay commissions when they bought and sold and taxes on those transaction. That's sheer folly.

Therefore, I stand by my statement that the study is flawed and useless. Either take out all those years or include the costs actually incurred.
Best Regards - Mel | | Semper Fi
User avatar
ObliviousInvestor
Posts: 4212
Joined: Tue Mar 17, 2009 9:32 am
Contact:

Post by ObliviousInvestor »

fredflinstone wrote:
Mel Lindauer wrote:Ignoring the actual costs that the market-timing scheme incurred renders the study useless IMO.
I respectfully disagree. I don't care about taxes or the cost of indexing in 1890. To the extent that I have investments in taxable accounts, I care about taxes today and going forward. Costs for a modern day investor are negligible (except for thinly-traded microcaps) so I would prefer an analysis that ignores those. I guess we will agree to disagree on this.
What if the historical predictive value of PE10 was just a market inefficiency that was allowed to exist because costs and taxes made it such that nobody could profit from it? Now that such costs and taxes are lower (or zero), should we expect this degree of market predictability to persist?
Mike Piper | Roth is a name, not an acronym. If you type ROTH, you're just yelling about retirement accounts.
yobria
Posts: 5978
Joined: Mon Feb 19, 2007 10:58 pm
Location: SF CA USA

Post by yobria »

Raging Mage wrote:
yobria wrote:Japan had high P/Es in the 1950s and 60s, yet very strong stock performance for the next 30 years.
As a relative investing newbie, this is an observation that gives me pause when considering Pfau's research. For many investors, 30 years is the vast majority of their pre-retirement investing lifetimes. Even if, over a period of 139 years, the backtested market-timing strategy based on PE10 outperforms buy-and-hold, it's entirely possible that the results during your particular range of investment years will be different.
Yes, it's also possible, as with the January Effect, that once discovered, it will simply stop working.

Let's use common sense: are the institutional investors that set the prices really going to consistently misvalue things, causing ineffiencies that only you can indentify and exploit at the time it happens?

Nick
User avatar
jeffyscott
Posts: 13438
Joined: Tue Feb 27, 2007 8:12 am

Post by jeffyscott »

William Million wrote:Even though I don't think much of this study and consider PE10 pseudo-science, I admit there are some individuals who are consistently successful at timing the market. However, these individuals don't see to rely on any single valuation metric. They look at all of them, and then add some analysis on top.
The study just shows (or confirms) that valuations are a factor that one might at least consider considering when deciding how to invest. I certainly would not go with a binary system, all stocks or none based on which side of a bright line (above or below average valuation) we are on. But for a study the system used sounds okay, that does not mean that it would be advisable to adopt something so simplistic for a real investment program.
User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 7:51 pm

Post by bob90245 »

fredflinstone wrote:
bob90245 wrote:
fredflinstone wrote:
555 wrote:
fredflinstone wrote: "Professor Pfau has identified a valuation-based asset allocation strategy which provides risk-adjusted returns that are greatly superior to the 50/50 balanced fund advocated by many Bogleheads."
I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
I don't belive taking more risk to earn higher returns is a refutation of Boglehead investing philosopy. As I already stated upthread, using "standard deviation" as measure of "risk" in Wade's paper is misleading. Here is a repost if this point isn't clear to you:

http://www.bogleheads.org/forum/viewtop ... 55#1066955
richard wrote:
bob90245 wrote:The P/E10 backtest tool will tell you to increase your stock allocation when business risk is high -- the very reason that causes P/E10 to fall. For those with short memories, I would imagine that the P/E10 tool would have told investors to load up on stocks during the financial meltdown of 2008-2009 when P/E10 fell below 15. However, I don't remember anyone telling investors that stocks were no longer as risky, do you?
This is a fundamental point. Low p/e means the market is telling you risk is high (and vice versa). You would expect to get higher returns because risk is higher. Of course, expected returns do not necessarily equal actual returns, or it would not be the case that risk was higher.

Getting higher returns by taking on more risk should not be surprising.
I don't understand this at all. Are you saying risk in the stock market was lowest when Nasdaq was at 5000 back in March 2000? Really?
Markets price for risk. During the time period of March 2000, the economy was growing briskly and unemployment was low. Thus, it was a low-risk business enviroment. With business risk perceived to be low, investors felt comfortable bidding up stocks to high levels.

But when stock are valuations high, investors have to expect low returns. And that is exactly what happened.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.
avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Post by avalpert »

fredflinstone wrote:
Mel Lindauer wrote:Ignoring the actual costs that the market-timing scheme incurred renders the study useless IMO.
I respectfully disagree. I don't care about taxes or the cost of indexing in 1890. To the extent that I have investments in taxable accounts, I care about taxes today and going forward. Costs for a modern day investor are negligible (except for thinly-traded microcaps) so I would prefer an analysis that ignores those. I guess we will agree to disagree on this.
But the people participating in the market then did care about taxes and costs - to ignore that impact erodes the comparability of prices across time.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Re: Valuation-based market timing with PE10 can improve retu

Post by fredflinstone »

docneil88 wrote:
fredflinstone wrote:Your figures are wrong. Look at Figure II on page 8 of the study. "Fixed 50/50" refers to a 50/50 balanced fund. "MT 0-100" refers to various all-or-nothing market timing strategies.
You're right. I confused SD with return on the 50/50 balanced fund. I edited my post above. Best, Neil
Thank you for that; much appreciated.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

Morgan wrote:Something that worries me, is that when you have to 'overbuy', your salary may be fixed or chopped to bits in a recession, or you could lose your job or business income. In other words, you may not have the spare capital to implement the strategy at excatly the wrong time.

We've rarely had recessions in which this became a significant issue though. In 1929 probably.

I don't think this is insurmountable, but it means you would ideally implement this strategy in regional overseas markets and work in a job that was service industry oriented somewhere else.
I don't understand this concern. The market-timing strategy can be implemented by people who have no income. When PE10 is low you shift from bonds to stocks and when PE10 is high you shift from stocks to bonds. There is no need to have income or spare capital.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

bob90245 wrote:
fredflinstone wrote:
bob90245 wrote:
fredflinstone wrote:
555 wrote: I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
I don't belive taking more risk to earn higher returns is a refutation of Boglehead investing philosopy. As I already stated upthread, using "standard deviation" as measure of "risk" in Wade's paper is misleading. Here is a repost if this point isn't clear to you:

http://www.bogleheads.org/forum/viewtop ... 55#1066955
richard wrote:
bob90245 wrote:The P/E10 backtest tool will tell you to increase your stock allocation when business risk is high -- the very reason that causes P/E10 to fall. For those with short memories, I would imagine that the P/E10 tool would have told investors to load up on stocks during the financial meltdown of 2008-2009 when P/E10 fell below 15. However, I don't remember anyone telling investors that stocks were no longer as risky, do you?
This is a fundamental point. Low p/e means the market is telling you risk is high (and vice versa). You would expect to get higher returns because risk is higher. Of course, expected returns do not necessarily equal actual returns, or it would not be the case that risk was higher.

Getting higher returns by taking on more risk should not be surprising.
I don't understand this at all. Are you saying risk in the stock market was lowest when Nasdaq was at 5000 back in March 2000? Really?
Markets price for risk. During the time period of March 2000, the economy was growing briskly and unemployment was low. Thus, it was a low-risk business enviroment. With business risk perceived to be low, investors felt comfortable bidding up stocks to high levels.

But when stock are valuations high, investors have to expect low returns. And that is exactly what happened.
If Mr. Market thought Nasdaq 5000 was low risk, it seems obvious in retrospect that Mr. Market was wrong. Very wrong. Nasdaq fell by more than 75 percent in just two and a half years after hitting 5000. Based on Pfau's research, it seems clear that a PE10-based market-timing approach is highly effective at exploiting such mis-judgments.
User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 7:51 pm

Post by bob90245 »

fredflinstone wrote:
bob90245 wrote:
fredflinstone wrote:
bob90245 wrote:
fredflinstone wrote: The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
I don't belive taking more risk to earn higher returns is a refutation of Boglehead investing philosopy. As I already stated upthread, using "standard deviation" as measure of "risk" in Wade's paper is misleading. Here is a repost if this point isn't clear to you:

http://www.bogleheads.org/forum/viewtop ... 55#1066955
richard wrote: This is a fundamental point. Low p/e means the market is telling you risk is high (and vice versa). You would expect to get higher returns because risk is higher. Of course, expected returns do not necessarily equal actual returns, or it would not be the case that risk was higher.

Getting higher returns by taking on more risk should not be surprising.
I don't understand this at all. Are you saying risk in the stock market was lowest when Nasdaq was at 5000 back in March 2000? Really?
Markets price for risk. During the time period of March 2000, the economy was growing briskly and unemployment was low. Thus, it was a low-risk business enviroment. With business risk perceived to be low, investors felt comfortable bidding up stocks to high levels.

But when stock are valuations high, investors have to expect low returns. And that is exactly what happened.
If Mr. Market thought Nasdaq 5000 was low risk, it seems obvious in retrospect that Mr. Market was wrong. Very wrong. Nasdaq fell by more than 75 percent in just two and a half years after hitting 5000. Based on Pfau's research, it seems clear that a PE10-based market-timing approach is highly effective at exploiting such mis-judgments.
It's not a misjudgement. I will repeat again.

When stock valuations are high, investors have to expect low returns. And that is exactly what happened.

Markets price for risk. During the time period of March 2000, the economy was growing briskly and unemployment was low. Thus, it was a low-risk business enviroment. With business risk perceived to be low, investors felt comfortable bidding up stocks to high levels.

I will agree that using the P/E10 tool to lighten up on stocks when valuations are high and loading up on stocks when valuations are low can produce higher returns. But it is misleading to think this is a free lunch.

In order to entice investors bear stock market risk when the business environment is poor, expected returns need to rise and this is accomplished by lower stock prices. It is here that you will see low P/E10 levels. Conversely, when the business environment is booming, it doesn't take much encouragement for investors to bid up stock prices. But this results in lower expected returns.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.
yobria
Posts: 5978
Joined: Mon Feb 19, 2007 10:58 pm
Location: SF CA USA

Post by yobria »

fredflinstone wrote:If Mr. Market thought Nasdaq 5000 was low risk, it seems obvious in retrospect that Mr. Market was wrong. Very wrong. Nasdaq fell by more than 75 percent in just two and a half years after hitting 5000. Based on Pfau's research, it seems clear that a PE10-based market-timing approach is highly effective at exploiting such mis-judgments.
Did you or Mr. Pfau actually make a fortune exploiting this at the time? In your lifetime, there are going to be many more "misjudgements" that were obvious with hindsight.

If you could identify even one of these, you could make millions in the market. But you're never going to.

Why not? Because when Mr. Market is wrong, you're likely to be very wrong, having only a portion of the information Mr. Market has.

Nick
letsgobobby
Posts: 12073
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby »

yobria wrote:
fredflinstone wrote:If Mr. Market thought Nasdaq 5000 was low risk, it seems obvious in retrospect that Mr. Market was wrong. Very wrong. Nasdaq fell by more than 75 percent in just two and a half years after hitting 5000. Based on Pfau's research, it seems clear that a PE10-based market-timing approach is highly effective at exploiting such mis-judgments.
Did you or Mr. Pfau actually make a fortune exploiting this at the time? In your lifetime, there are going to be many more "misjudgements" that were obvious with hindsight.

If you could identify even one of these, you could make millions in the market. But you're never going to.

Why not? Because when Mr. Market is wrong, you're likely to be very wrong, having only a portion of the information Mr. Market has.

Nick
well, 'fortune' is a relative term. i'm highly satisfied with the high-return, low-volatility outcome i've experienced.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

There are many good comments on this thread!

Some are about limitations of the original Fisher & Statman study: It only considers the historical period as a whole, and it only considers extreme strategies of all stocks or all bonds.

For individual investors, comparisons over shorter periods (such as 20-40 years, for instance) are more relevant, and also more realistic asset allocation strategies (such as a baseline of 50% stocks and then only moving to more are less stocks when PE10 moves far from its average) will be more meaningful. I am working on this topic and though I have preliminary results on my blog from March [the preliminary results I posted in January are okay, but more features are added in the March version], I don't have any finished papers.

JW Nearly Retired: I had an implicit assumption that choice of dates (January 1, etc.) and choice of valuations (PE6,...PE16, etc.) would not have any significant impacts on the results. As well, I was thinking in terms of something that would be easy to follow: only do a checkup at the end of each year rather than constant monitoring, and using PE10 which is readily available on the Internet. But you are right that I should check more of these possibilities as a form of sensitivity analysis. If I did accidentally cherry pick the only scenario that works, then of course it is a serious problem. I will check more about this. Thank you. -- The paper is not published yet, and the whole point of releasing the unpublished version is to get valuable feedback like this.

Richard, thanks for the support on some points. About your concern for the methodology: I'm doing the same thing as the Fisher and Statman study. Both are based only on annual data and so both check once a year about whether a change in asset allocation is needed. Fisher and Statman only have annual data, so they can't possibly be doing what you are suggesting they do. Monthly data is now available, so I can move in the direction that JW Nearly Retired indicates is important, but as of now, I'm doing the same thing as F & S. If you think I am missing something, please elaborate further.

rustymutt: Of course I would like to use more asset classes, but sufficient historical data is not available, especially with regard to valuation measures.

Raging Mage: Valuation-based allocation can be combined with age-based allocation. You don't see the full effect in this particular study because I am playing by Fisher and Statman's rules, but you could have stock allocations in the range of 70-100% when you are young, and the range would shift down to 10-50% by retirement, as one possible example.

bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.

As Mel pointed out, taxes are also important. I do need to consider modern day tax rates. For sure. I'm not so convinced about the importance of historical tax rates, because I don't think enough people would have ever followed this strategy for it to have important market impacts. It is too psychologically difficult as a number of posters mentioned, and many institutions can't afford to trail benchmarks for too long even if they are later vindicated. Only individuals with the right temperament and long-term horizon might pull it off. Anyone who panicked and sold stocks during 2008 should forget about it.

As well about historical tax rates and costs, before index funds, individual investors could not own the index. It would have been pretty much impossible. So there's not much to test. The whole research approach is DOA for anyone thinking this way. Though, Oblivious Investor does state the question well. Perhaps the days of this working are coming to an end. I guess to know about that, we would need data on actual investment flows and their motivations. Are valuation motives increasing now? I don't have any kind of data about that.

StoneReader: For my horse race figure, you are suggesting that institutions are better off with 100% stocks than with the valuation-based strategy. But most institutions cannot practically have a stock allocation of 100%. This is a limitation of only looking at extreme strategies, and why I am still working on the case of comparing more realistic asset allocation strategies.

This is actually rather important. Here is a table from my blog, using real data, instead of nominal data. But from this you can see, for instance, that 100% stocks accumulates about double the wealth of a fixed 80/20 asset allocation with annual rebalancing. This is very much why I am concerned that it is "unfair" that the original Fisher and Statman study only consider a 100% stocks strategy against a strategy with an average stock allocation in the neighborhood of 50/50. :

Image
User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 7:51 pm

Post by bob90245 »

wade wrote:bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.
Thanks for your acknowledgement. It is important to mention that "risk adjusted returns" is merely a mathematical abstraction and not something that investors experience while living in during robust or poor business environments that prompts P/E10 to rise or fall, respectively.

For your other comment, I agree that markets are cyclical. And barring a complete and permanent global economic collapse, markets should eventually recover from periods of decline.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

wade wrote:There are many good comments on this thread!

Some are about limitations of the original Fisher & Statman study: It only considers the historical period as a whole, and it only considers extreme strategies of all stocks or all bonds.

For individual investors, comparisons over shorter periods (such as 20-40 years, for instance) are more relevant, and also more realistic asset allocation strategies (such as a baseline of 50% stocks and then only moving to more are less stocks when PE10 moves far from its average) will be more meaningful. I am working on this topic and though I have preliminary results on my blog from March [the preliminary results I posted in January are okay, but more features are added in the March version], I don't have any finished papers.

JW Nearly Retired: I had an implicit assumption that choice of dates (January 1, etc.) and choice of valuations (PE6,...PE16, etc.) would not have any significant impacts on the results. As well, I was thinking in terms of something that would be easy to follow: only do a checkup at the end of each year rather than constant monitoring, and using PE10 which is readily available on the Internet. But you are right that I should check more of these possibilities as a form of sensitivity analysis. If I did accidentally cherry pick the only scenario that works, then of course it is a serious problem. I will check more about this. Thank you. -- The paper is not published yet, and the whole point of releasing the unpublished version is to get valuable feedback like this.

Richard, thanks for the support on some points. About your concern for the methodology: I'm doing the same thing as the Fisher and Statman study. Both are based only on annual data and so both check once a year about whether a change in asset allocation is needed. Fisher and Statman only have annual data, so they can't possibly be doing what you are suggesting they do. Monthly data is now available, so I can move in the direction that JW Nearly Retired indicates is important, but as of now, I'm doing the same thing as F & S. If you think I am missing something, please elaborate further.

rustymutt: Of course I would like to use more asset classes, but sufficient historical data is not available, especially with regard to valuation measures.

Raging Mage: Valuation-based allocation can be combined with age-based allocation. You don't see the full effect in this particular study because I am playing by Fisher and Statman's rules, but you could have stock allocations in the range of 70-100% when you are young, and the range would shift down to 10-50% by retirement, as one possible example.

bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.

As Mel pointed out, taxes are also important. I do need to consider modern day tax rates. For sure. I'm not so convinced about the importance of historical tax rates, because I don't think enough people would have ever followed this strategy for it to have important market impacts. It is too psychologically difficult as a number of posters mentioned, and many institutions can't afford to trail benchmarks for too long even if they are later vindicated. Only individuals with the right temperament and long-term horizon might pull it off. Anyone who panicked and sold stocks during 2008 should forget about it.

As well about historical tax rates and costs, before index funds, individual investors could not own the index. It would have been pretty much impossible. So there's not much to test. The whole research approach is DOA for anyone thinking this way. Though, Oblivious Investor does state the question well. Perhaps the days of this working are coming to an end. I guess to know about that, we would need data on actual investment flows and their motivations. Are valuation motives increasing now? I don't have any kind of data about that.

StoneReader: For my horse race figure, you are suggesting that institutions are better off with 100% stocks than with the valuation-based strategy. But most institutions cannot practically have a stock allocation of 100%. This is a limitation of only looking at extreme strategies, and why I am still working on the case of comparing more realistic asset allocation strategies.

This is actually rather important. Here is a table from my blog, using real data, instead of nominal data. But from this you can see, for instance, that 100% stocks accumulates about double the wealth of a fixed 80/20 asset allocation with annual rebalancing. This is very much why I am concerned that it is "unfair" that the original Fisher and Statman study only consider a 100% stocks strategy against a strategy with an average stock allocation in the neighborhood of 50/50. :

Image
I wish everyone responded to criticism in such a calm, clear, rational manner. I know I don't always do so. Professor Pfau, you are a class act.
User avatar
Raging Mage
Posts: 118
Joined: Fri Dec 31, 2010 12:37 pm

Post by Raging Mage »

wade wrote:Raging Mage: Valuation-based allocation can be combined with age-based allocation. You don't see the full effect in this particular study because I am playing by Fisher and Statman's rules, but you could have stock allocations in the range of 70-100% when you are young, and the range would shift down to 10-50% by retirement, as one possible example.
Wade, thanks for addressing my comments. I'm not sure how meaningful it would be to your research, but it would be interesting to find out exactly what kind of age-based adjustment shows the most promise in regards to the market-timing strategy outlined in your paper. For example, should a fixed range of percentage points (such as 30-40 in your comment above) always be available for market timing shifts? Should it become narrower as one approaches and enters retirement, or vice versa? I ask because it's a tricky question for which I don't have a good answer (made worse by the fact that each investor has a different risk tolerance and need for risk), and some investors who wish to apply your findings to their investment strategies may want a "rubber meets the road" dataset where the need to reduce risk with age is both taken into account and tested for optimization in combination with PE10 adjustments.

A couple other random thoughts I have on the subject are:

- What impact would using a "sliding scale" of yearly stock/bond ratio adjustments connected to PE10 have on a portfolio's risk/return profile? In other words, instead of just having two different ratios to switch between depending on PE10 (for example, 20-80 and 80-20), what if every ratio in between those two was also made available, so fine adjustments can be made year-by-year depending on how far above or below average the current PE10 is?

- It's widely known that more companies have shifted away from dividend payouts and towards stock buybacks compared to several decades ago, since this can provide greater benefits for both the investor (more payouts can take the form of long-term capital gains) and the company (managers' stock options can become more valuable, for instance). Can we expect this development to significantly alter the typical PE10 going forward? Will this have any impact on the validity of using past/current PE10 data to make asset allocation decisions? This is just food for thought, really.

Thanks for your continued work on this subject. Despite all the questioning and challenges presented to you, I'm sure the efforts are appreciated.
"Here are three easy ways to beat the market: Deception, irrelevance and bad math." - Alexander Green, Investment U
User avatar
Mel Lindauer
Moderator
Posts: 35757
Joined: Mon Feb 19, 2007 7:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer »

wade wrote:There are many good comments on this thread!

Some are about limitations of the original Fisher & Statman study: It only considers the historical period as a whole, and it only considers extreme strategies of all stocks or all bonds.

For individual investors, comparisons over shorter periods (such as 20-40 years, for instance) are more relevant, and also more realistic asset allocation strategies (such as a baseline of 50% stocks and then only moving to more are less stocks when PE10 moves far from its average) will be more meaningful. I am working on this topic and though I have preliminary results on my blog from March [the preliminary results I posted in January are okay, but more features are added in the March version], I don't have any finished papers.

JW Nearly Retired: I had an implicit assumption that choice of dates (January 1, etc.) and choice of valuations (PE6,...PE16, etc.) would not have any significant impacts on the results. As well, I was thinking in terms of something that would be easy to follow: only do a checkup at the end of each year rather than constant monitoring, and using PE10 which is readily available on the Internet. But you are right that I should check more of these possibilities as a form of sensitivity analysis. If I did accidentally cherry pick the only scenario that works, then of course it is a serious problem. I will check more about this. Thank you. -- The paper is not published yet, and the whole point of releasing the unpublished version is to get valuable feedback like this.

Richard, thanks for the support on some points. About your concern for the methodology: I'm doing the same thing as the Fisher and Statman study. Both are based only on annual data and so both check once a year about whether a change in asset allocation is needed. Fisher and Statman only have annual data, so they can't possibly be doing what you are suggesting they do. Monthly data is now available, so I can move in the direction that JW Nearly Retired indicates is important, but as of now, I'm doing the same thing as F & S. If you think I am missing something, please elaborate further.

rustymutt: Of course I would like to use more asset classes, but sufficient historical data is not available, especially with regard to valuation measures.

Raging Mage: Valuation-based allocation can be combined with age-based allocation. You don't see the full effect in this particular study because I am playing by Fisher and Statman's rules, but you could have stock allocations in the range of 70-100% when you are young, and the range would shift down to 10-50% by retirement, as one possible example.

bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.

As Mel pointed out, taxes are also important. I do need to consider modern day tax rates. For sure. I'm not so convinced about the importance of historical tax rates, because I don't think enough people would have ever followed this strategy for it to have important market impacts. It is too psychologically difficult as a number of posters mentioned, and many institutions can't afford to trail benchmarks for too long even if they are later vindicated. Only individuals with the right temperament and long-term horizon might pull it off. Anyone who panicked and sold stocks during 2008 should forget about it.

As well about historical tax rates and costs, before index funds, individual investors could not own the index. It would have been pretty much impossible. So there's not much to test. The whole research approach is DOA for anyone thinking this way. Though, Oblivious Investor does state the question well. Perhaps the days of this working are coming to an end. I guess to know about that, we would need data on actual investment flows and their motivations. Are valuation motives increasing now? I don't have any kind of data about that.

StoneReader: For my horse race figure, you are suggesting that institutions are better off with 100% stocks than with the valuation-based strategy. But most institutions cannot practically have a stock allocation of 100%. This is a limitation of only looking at extreme strategies, and why I am still working on the case of comparing more realistic asset allocation strategies.

This is actually rather important. Here is a table from my blog, using real data, instead of nominal data. But from this you can see, for instance, that 100% stocks accumulates about double the wealth of a fixed 80/20 asset allocation with annual rebalancing. This is very much why I am concerned that it is "unfair" that the original Fisher and Statman study only consider a 100% stocks strategy against a strategy with an average stock allocation in the neighborhood of 50/50. :

Image
Hi Again Wade:

You mentioned needing to include modern day taxes, but you're still choosing to ignore the real world costs of many of the years that you've included in your study to try to prove your market-timing scheme works.

If you don't want to include the costs (taxes and trading costs when both buying and selling) that would have actually been incurred by an investor using this market-timing scheme, then don't include those years in the study, since without these costs those years don't prove anything and it distorts the results.

I have no doubt that using the actual figures would present an entirely different picture. I'd hope that you're actually looking for the truth and not merely trying to take the easy way out by not using critically vital information simply because it may take some time and effort to gather and apply the data.

Regards,

Mel
Best Regards - Mel | | Semper Fi
letsgobobby
Posts: 12073
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby »

can he run the data assuming a tax-deferred only account? For many investors that holds a large enough portion of their assets to allow the tactical allocation to occur solely in tax-advantaged space.

Can he run the data starting only in the early 1980s, when the S&P500 index fund became widely available?
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Raging Mage: Thank you for the comments. I'm just getting started on combining age-based and valuation-based, but I will keep your comments in mind when I work on that. As for the sliding scale, my other (unfinished) research on this uses 3 categories instead of two: high, medium, and low, rather than just high and low. I think that is enough to get at the basic idea, but certainly you could make further refinements as a sliding scale.

Mel: I'm trying to acknowledge your point. Before index funds, it would have been impossible for individual investors to capture the market returns from either the fixed allocation strategy or the valuation-based strategy. It would have been too costly when there were no index funds. Where your point matters is if many people would have been following a valuation-based approach but were otherwise constrained from doing so due to these costs. I don't know if you are right or wrong about this. I don't know if people were really eager to do it but constrained by costs and taxes. Graham and Dodd made this basic point, and so certainly investors going back to the 1930s and 1940s could have known about it. But, psychologically, this strategy is difficult to do. I doubt whether such a historical analysis could ever be properly constructed. It's not a matter of not taking the time or effort, it just can't feasibly be done. So I leave this as a caveat. There is not much else I can do about it. People can then disagree about how important this point is as they make their own decisions and form their own views.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

letsgobobby wrote:can he run the data assuming a tax-deferred only account? For many investors that holds a large enough portion of their assets to allow the tactical allocation to occur solely in tax-advantaged space.

Can he run the data starting only in the early 1980s, when the S&P500 index fund became widely available?
I think Mel's point is that there were no tax-deferred accounts until recently, so the answer is no.

Yes, it could be run since only the early 1980s. And, recent years are when the strategy has not worked well due to the unprecedented bull market run in the 1990s with valuations reaching record levels. That's the kind of case where valuation-based strategies will fall behind their buy-and-hold counterparts. However, in this case, both strategies provide high wealth compared to most of history, so it is not necessarily a big problem. When wealth is high, the "marginal utility" of additional wealth is low.

Here are results for rolling 30-year periods comparing a 50/50 fixed allocation to an allocation that switches between 25, 50, or 75 percent stocks depending on PE10:

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

Post by Dick Purcell »

Wade –

I love your third paragraph beginning “For individual investors…”!!

To me, that means assess things in terms of the purpose, which for most of us (I hope) is best probabilistic prospects for results for our dollar plans and future needs and goals. Net of fees and taxes, as Mel emphasizes, and of course net of inflation.

But we have vastly different plans and goals. How much money we put in, when; and how much we hope to be able to take out, and when; and we have different priorities for high-side possibilities versus low-side perils. And we face different fees and taxes.

So I think the most valuable thing you could provide, from what you have been studying, is an expression or factor that each of us can incorporate in his/her own assessment(s) of result probabilities for his/her own dollar plan and goals. And it seems to me that Bob90245 just about defined the factor we need, in his last message just above:

What does history show to be the ratio between (a) the extent to which PE10 is above (below) historical average and (b) then extent to which the next-five-years’ average return rate arithmetic mean is below (above) historical average?

Frankly, for us individual investors, on the topic of this discussion I think that's all we need. Nothing more.

With that answer, each of us can simply adjust our return-rate-mean assumptions for the next five years based on current PE10, and then Monte Carlo away to assess and compare allocations for our particular dollar-value plans and goals. Test sensitivities of assumptions, of course, as Magellan advises . . .

My guess is that the effects of such adjustments are dwarfed by the vast long-term effects of uncertainty about assumptions for the basic return-rate means, which Rodc has illustrated in other discussions. At the same time, I think an adjustment to take advantage of what your research reveals can improve some folks' dollar-goal result probabilities.

Dick Purcell
lazyday
Posts: 3849
Joined: Wed Mar 14, 2007 10:27 pm

Post by lazyday »

fredflinstone wrote:
555 wrote:I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
No offence, and I think the paper may be a useful addition to knowledge even if I worry about people being led astray into painful experiences market timing, but there have been a few topics you've become quite interested in, to decide against later.
fredflinstone wrote:I look back at some of the dumb posts I wrote a year ago (right after I discovered this board), and I cringe. Much of what I wrote was so awful.
http://www.bogleheads.org/forum/viewtopic.php?t=72908

A year isn't long in an investing education.
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Re: Overvalued by Any Measure

Post by Bongleur »

StoneReader wrote:
q is the ratio between the value of companies according to the stock market and their net worth measured at replacement cost. CAPE is the cyclically adjusted PE ratio for the market. http://www.smithers.co.uk/page.php?id=34

Image
Note that the q used in the chart & defined above is not exactly the same as Tobin's q:

http://www.valuingwallstreet.com/VApp.pdf

EDIT: And an interesting technical factoid about mean reversion:

"unlike q, the stock price has no tendency to
mean-revert, so we cannot get any help at all from past average
values of the stock price. What does help us, however, is the fact
that q mean-reverts, since this implies that the stock price cannot
get too far away from net worth per share. This makes prediction
easier, but also more complicated in technical terms, since if
we want to predict stock prices, we need also to predict net worth."
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.
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

Dick Purcell wrote:Wade –
it seems to me that Bob90245 just about defined the factor we need, in his last message just above:

What does history show to be the ratio between (a) the extent to which PE10 is above (below) historical average and (b) then extent to which the next-five-years’ average return rate arithmetic mean is below (above) historical average?

Dick Purcell
I think Hussman has pontificated on that in one of his weekly articles. He tries to estimate for 7-10 years ahead. He also thinks that the expected returns for the upcoming period have pretty much already been "front loaded" during the last couple of years.
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
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

lazyday wrote:
fredflinstone wrote:
555 wrote:I think you are making a mountain out of a molehill.
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
No offence, and I think the paper may be a useful addition to knowledge even if I worry about people being led astray into painful experiences market timing, but there have been a few topics you've become quite interested in, to decide against later.
fredflinstone wrote:I look back at some of the dumb posts I wrote a year ago (right after I discovered this board), and I cringe. Much of what I wrote was so awful.
http://www.bogleheads.org/forum/viewtopic.php?t=72908

A year isn't long in an investing education.
so true! I am a flawed investor and a flawed human being. I will make no change to my ips for at least a month or two while all of this sinks in.
richard
Posts: 7961
Joined: Tue Feb 20, 2007 2:38 pm
Contact:

Post by richard »

Mel Lindauer wrote:You mentioned needing to include modern day taxes, but you're still choosing to ignore the real world costs of many of the years that you've included in your study to try to prove your market-timing scheme works.

If you don't want to include the costs (taxes and trading costs when both buying and selling) that would have actually been incurred by an investor using this market-timing scheme, then don't include those years in the study, since without these costs those years don't prove anything and it distorts the results.
I believe you've just shown that no study using long-term past data proves anything. While I largely believe this to be true, it would include much of the data presented by, among others, Jack Bogle and Vanguard. What percentage of studies that use data sets going back to the 1920s or 1870s try to adjust for taxes and costs?

What tax rates do you suggest Wade use? Considerations include that there were no income taxes prior to 1913, that rates were minimal during the initial years of income taxes, different people pay at different rates, trading is now dominated by institutions that don't take taxes into account.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

richard wrote:
Mel Lindauer wrote:You mentioned needing to include modern day taxes, but you're still choosing to ignore the real world costs of many of the years that you've included in your study to try to prove your market-timing scheme works.

If you don't want to include the costs (taxes and trading costs when both buying and selling) that would have actually been incurred by an investor using this market-timing scheme, then don't include those years in the study, since without these costs those years don't prove anything and it distorts the results.
I believe you've just shown that no study using long-term past data proves anything. While I largely believe this to be true, it would include much of the data presented by, among others, Jack Bogle and Vanguard. What percentage of studies that use data sets going back to the 1920s or 1870s try to adjust for taxes and costs?

What tax rates do you suggest Wade use? Considerations include that there were no income taxes prior to 1913, that rates were minimal during the initial years of income taxes, different people pay at different rates, trading is now dominated by institutions that don't take taxes into account.
+1
Roy
Posts: 970
Joined: Wed Sep 10, 2008 9:34 am

Post by Roy »

jeffyscott wrote: The study just shows (or confirms) that valuations are a factor that one might at least consider considering when deciding how to invest. I certainly would not go with a binary system, all stocks or none based on which side of a bright line (above or below average valuation) we are on. But for a study the system used sounds okay, that does not mean that it would be advisable to adopt something so simplistic for a real investment program.
I think Jeffy has a good way to view this study—or Shiller's peer-reviewed work that impelled it. Valuations either matter or they don't. They clearly do matter. What to do about them—if anything beyond a rebalance—is the issue.

The formulaic ways that an investor can implement strategy with this method (PE/10) are infinite (all-in/all-out, scaled-in with range-brackets, etc.), and that is a big problem for many who don't want to be bothered with any tactical management, even the occasional sort that a nod to PE/10 would require.

For professionals, note that even Shiller, around the March 2009 lows, was waiting for stocks to drop to PE/10 before going-in (I think they were around PE/12 or so in that interview and did not go lower). This year, Shiller said he had scaled-in some already at that time, so who knows what he really did? The point is that if even Shiller was unclear what to do, genuine clarity is unlikely for average folk—or even other pros. Remember, when fear is present, "buying low" feels suicidal, and even the act of rebalancing is abhorrent, never mind buying yet more equity as this method would require. That said, I think Shiller's work (and Wade's) has merit, and that valuations will continue to matter.

And if one prefers a Bogle-esque buy-and-hold strategy, I see nothing wrong there, provided those investors understand that valuations matter and that events like the last two Bear markets can crush equity-heavy portfolios. The flip side is that if someone screams "irrational exuberance", and you reduce equities, you might miss out on the next three stellar years, as happened in the late 90s. Personally, I fear downside more than regretting upside. But many things need be considered and little can be reduced to any formula.
User avatar
jeffyscott
Posts: 13438
Joined: Tue Feb 27, 2007 8:12 am

Post by jeffyscott »

wade wrote: bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.

I think it would also break down in the opposite situation, if the Dow 36,000 people had been correct, you'd be sitting in cash while the market rose to it's new permanent high plateau.

More realistically, it may be that due to various changes, the future average PE10 will be higher than the past average. I think if you fit a line to the PE10 data, the best fit actually does have a slight upward slope (not that that could continue forever). Of course, that can be addressed by not applying this as an on/off switch that immediately flips when PE10 exceeds it's average, as you have already suggested.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox »

richard wrote:
Mel Lindauer wrote:You mentioned needing to include modern day taxes, but you're still choosing to ignore the real world costs of many of the years that you've included in your study to try to prove your market-timing scheme works.

If you don't want to include the costs (taxes and trading costs when both buying and selling) that would have actually been incurred by an investor using this market-timing scheme, then don't include those years in the study, since without these costs those years don't prove anything and it distorts the results.
I believe you've just shown that no study using long-term past data proves anything. While I largely believe this to be true, it would include much of the data presented by, among others, Jack Bogle and Vanguard. What percentage of studies that use data sets going back to the 1920s or 1870s try to adjust for taxes and costs?

What tax rates do you suggest Wade use? Considerations include that there were no income taxes prior to 1913, that rates were minimal during the initial years of income taxes, different people pay at different rates, trading is now dominated by institutions that don't take taxes into account.
I was going to make a similar point. If not including taxes and transaction fees invalidates this study, it invalidates just about every study out there. Like the Trinity study, for instance.

Taking it further, before Bogle started the first index fund in 1975, it was impractical, if not impossible, to hold the S&P 500. So any study that assumes an investor holds the S&P 500 and that goes back before 1975 is invalid, by this criteria. In fact, the S&P 500 itself dates only from the late 1950s, before that was the S&P Composite. And a lot of the really early data was only recently reconstructed by graduate students from old newspaper archives, so it is somewhat dubious.

These studies are all just hypothetical. And I believe that the IRS has ruled that hypothetical investors don't have to pay taxes. Also, most brokerage firms waive their commission for hypothetical trades. Problem solved. Remember, no one actually experienced those returns shown in some hypothetical study. When Jeremy Siegel shows someone buys $1 worth of stocks in 1802 growing to $10.98 million by 2005, I ask who was it? I want names. Of course, nobody actually got, or could have gotten, that return. It's all just hypothetical.

Now I'm going to have to agree with Mel to some extent. There are limitations to what can be learned from historical data. That's why I don't put a whole lot of faith in stuff like a 4% Rule which are derived from backtesting. But you have to take them all with a grain of salt. You can't just cherry pick the one's that present results that you disagree with and say it's invalid because it ignores taxes and fees. Consistency.

Don't take any of this stuff too seriously.
letsgobobby
Posts: 12073
Joined: Fri Sep 18, 2009 1:10 am

Post by letsgobobby »

jeffyscott wrote:
wade wrote: bob90245's caveat is important. All of the risk measures I consider evaluation the risk after-the-fact, not the risk that people thought they were facing before the fact. But really, the only way for this to break down so that you are better off not following a quantitative rule would be for the markets to experience a big disaster and never again recovery. It could happen, but unless it's the end of the world, markets should bounce back even if it takes 10-15 years to happen.

I think it would also break down in the opposite situation, if the Dow 36,000 people had been correct, you'd be sitting in cash while the market rose to it's new permanent high plateau.

More realistically, it may be that due to various changes, the future average PE10 will be higher than the past average. I think if you fit a line to the PE10 data, the best fit actually does have a slight upward slope (not that that could continue forever). Of course, that can be addressed by not applying this as an on/off switch that immediately flips when PE10 exceeds it's average, as you have already suggested.
2 thoughts.

first, Dow 11,700 (in 2000) or Dow 14,300 (in 2007) or heck even Dow 36,000 means nothing to the value-based-allocator, because of his belief in reversion to the mean. Being largely out of the S&P500/total stocks from the late 90s for 10 years did not hurt this individual because eventually stocks collapsed to 1996 levels. furthermore, he got 1996 prices in 2009 without any of the intervening volatility. Isn't that the goal of portfolio theory - maximal efficiency meaning maximum returns with minimum risk?

second, I like your observation that PE may increase slowly over time. There may even be justification for that (the world is a safer place, etc). But it would still not explain PE10 at 2, 3, or 4 times their average values over long periods of time. I believe this model works best at the extremes - PE10 under 12 or especially 10, and above 24-25 or so. That's my visual take by looking at the 20 year returns posted here:

http://www.bogleheads.org/forum/viewtopic.php?p=746593

grayfox - yes Shiller kept waiting for the market to go lower, but he did 'scale in' as you say. My dad kept waiting til PE10 got to 5, so he didn't get much action. Sorry dad but that is the classic cognitive error of 'black or white thinking.' The data suggests that the cheaper (per PE10) stocks are, the better the long term value proposition. It works best at PE10 less than 12 and PE10 greater than 23, but has validity in the middle.

Using this model one should have a maximal equity exposure considering PE10 and all other factors, namely risk tolerance and age. So at my age, that might be 100% (which I would do if PE10 were 5). At Shiller's age, perhaps 50%. At PE10 10.4, which is what happened in early March 2009, one should certainly be 2/3 of the way there, maybe 3/4, in part because the journey from PE10 to PE5 has only happened about twice in the last 100 years.

Remember the model does not guarantee better results. It claims to offer a better expected return, and it does so with less volatility than buy and hold/rebalance.
User avatar
Mel Lindauer
Moderator
Posts: 35757
Joined: Mon Feb 19, 2007 7:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer »

grayfox wrote:
richard wrote:
Mel Lindauer wrote:You mentioned needing to include modern day taxes, but you're still choosing to ignore the real world costs of many of the years that you've included in your study to try to prove your market-timing scheme works.

If you don't want to include the costs (taxes and trading costs when both buying and selling) that would have actually been incurred by an investor using this market-timing scheme, then don't include those years in the study, since without these costs those years don't prove anything and it distorts the results.
I believe you've just shown that no study using long-term past data proves anything. While I largely believe this to be true, it would include much of the data presented by, among others, Jack Bogle and Vanguard. What percentage of studies that use data sets going back to the 1920s or 1870s try to adjust for taxes and costs?

What tax rates do you suggest Wade use? Considerations include that there were no income taxes prior to 1913, that rates were minimal during the initial years of income taxes, different people pay at different rates, trading is now dominated by institutions that don't take taxes into account.
I was going to make a similar point. If not including taxes and transaction fees invalidates this study, it invalidates just about every study out there. Like the Trinity study, for instance.

Taking it further, before Bogle started the first index fund in 1975, it was impractical, if not impossible, to hold the S&P 500. So any study that assumes an investor holds the S&P 500 and that goes back before 1975 is invalid, by this criteria. In fact, the S&P 500 itself dates only from the late 1950s, before that was the S&P Composite. And a lot of the really early data was only recently reconstructed by graduate students from old newspaper archives, so it is somewhat dubious.

These studies are all just hypothetical. And I believe that the IRS has ruled that hypothetical investors don't have to pay taxes. Also, most brokerage firms waive their commission for hypothetical trades. Problem solved. Remember, no one actually experienced those returns shown in some hypothetical study. When Jeremy Siegel shows someone buys $1 worth of stocks in 1802 growing to $10.98 million by 2005, I ask who was it? I want names. Of course, nobody actually got, or could have gotten, that return. It's all just hypothetical.

Now I'm going to have to agree with Mel to some extent. There are limitations to what can be learned from historical data. That's why I don't put a whole lot of faith in stuff like a 4% Rule which are derived from backtesting. But you have to take them all with a grain of salt. You can't just cherry pick the one's that present results that you disagree with and say it's invalid because it ignores taxes and fees. Consistency.

Don't take any of this stuff too seriously.
Hi Grayfox:

My point is this: When one group in the study would have paid much higher commissions, trading costs and taxes (the market timers) than the other group (the buy-and-holders), to ignore that in the study skews the results in favor of the market-timers. Folks like the OP and some others seem to be ignoring that simple fact.

And when you say "Don't take any of this stuff too seriously", I don't, but some folks (like the OP and a few others) obviously do when they make statements like this:
fredflinstone wrote:That's what Professor Wade Pfau says:
[M]arket timing [using Schiller's PE10 measure of valuation] provides comparable risks and the same average asset allocation as a 50/50 fixed allocation strategy, but with much higher returns.
http://mpra.ub.uni-muenchen.de/29448/1/ ... _29448.pdf

I have not yet read the paper closely. I am going to do so now. This seems to me to be a fundamental challenge to some of the most basic tenets of the Boglehead paradigm.

Note that Professor Pfau acknowledges this Forum on the first page of his paper.
fredflinstone wrote:
555 wrote:Could someone please explain what the fuss is all about.
Professor Pfau has identified a valuation-based asset allocation strategy which provides risk-adjusted returns that are greatly superior to the 50/50 balanced fund advocated by many Bogleheads.
and
Pfau's paper demonstrates convincingly that a valuation-based long-term market-timing approach can and does beat a 50/50 balanced fund--and does so by a large margin. I recommend that you read his paper.
and
The paper refutes a central tenet of the Boglehead investing philosophy. It's a big deal.
and
Based on Pfau's research, it seems clear that a PE10-based market-timing approach is highly effective at exploiting such mis-judgments.
and, as if it's a guarantee it to work in the future
True. But when an investment approach is strongly supported by both empirical evidence (139 years of U.S. data) and common sense (buy when cheap, sell when expensive) then it is probably going to continue to work well in the future.
and statements like this on the thread
One other issue about this approach: if in the heat of the moment one does NOT follow it (say, when PE10 falls to 10 one does not overincrease stocks), then the cost of NOT following the strategy is still zero, assuming one follows the basic Boglehead tenet of at least rebalancing to maintain AA. What other strategy gives you the opportunity to increase returns without sacrificing the returns straight indexing and rebalancing "should" give you in the first place?
It appears from this thread that at least a few folks think this is the Holy Grail of investing, and leaving this kind of stuff unchallenged is dangerous. Folks need to understand the limitations and flaws in this and other similar work.

Previous studies have shown lots of back-tested schemes that worked -- that is, until they stopped working.
Best Regards - Mel | | Semper Fi
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

just for fun, I came up with an equity allocation formula that takes into account three factors: the investor's age, the PE10 ratio, and the 5-year yield on TIPS.

Voila:

=MIN(95,(MAX(10,(152.5-(AGE*1.25)-(2.5*PE10)-(20*TIPS))))),

where AGE is the investor's age, PE10 is the Shiller PE10 ratio, and TIPS is the yield on 5-year TIPS bonds.

The maximum percentage to equities is 95 and the minimum is 10. Note that although I incorporate TIPS yields into the model, the investor is not restricted to buying TIPS in his or her fixed-income holdings.

Here are the resultant equity percentages under a few different scenarios:

Code: Select all

AGE	PE10	TIPS yield	% equities
20	   15	   0	          90
30	   15	   0	          77.5
40	   15	   0	          65
50	   15	   0	          52.5
60	   15	   0	          40
			
20	   25	   0             65
30	   25	   0	          52.5
40	   25	   0	          40
50	   25	   0	          27.5
60	   25	   0	          15
			
20	   35	   0	          40
30	   35	   0	          27.5
40	   35	   0	          15
50	   35	   0	          10
60	   35	   0	          10
These numbers look about right for me. Of course, others would probably want to modify it depending on their risk aversion, need for growth, and degree of tolerance for market-timing.
User avatar
Topic Author
fredflinstone
Posts: 2829
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Post by fredflinstone »

Mel Lindauer wrote:It appears from this thread that at least a few folks think this is the Holy Grail of investing, and leaving this kind of stuff unchallenged is dangerous.
If you ask me, a few folks think the standard Boglehead approach (fixed asset allocation; stay the course regardless of valuation) is the Holy Grail of Investing.
User avatar
Dick Purcell
Posts: 520
Joined: Tue Oct 26, 2010 1:58 am

Post by Dick Purcell »

Wade has brought attention to something that deserves a look – but what should be looked at is what it means for real people. Most of this discussion is lost in ivory tower abstractions. Dangit, let’s come down to Earth.

1. The purpose is best result probabilities for real people's dollar plans and goals. That is what should be the basis for actual decisions regarding what Wade’s research suggests.

2. Nobody invests for over 100 years. Maybe 20, 30, 40

3. For almost everyone, the plan has different amounts invested in different years, and dollar-weighted, the years that count most are quite fewer than the full life of the plan.

4. Because of 2 and 3, any strategy of jumping to sharply different allocations would be likely to make the dollar-years-weighted average allocation significantly different from what’s intended, with net adverse effect. Real people should not let this discussion’s topic lead to straying much, if any, from desired average allocations.

5. Different people have very different cash flow plans – how much in when, how much out when, etc. So what we need is something that can be incorporated in assessment of different plans, for assessment in future-dollar-result prospects for different goals.

6. For this purpose, the useful thing that Wade’s research suggests is that for higher PE10, return rates for the next few years are more likely to average lower, and vice versa. What we need is an assessment of whether and to what extent that has been true in history. Not as a binary hair-trigger jump, but as a slope.

7. Bob90245 has something close to what we need at his website, at http://www.bobsfinancialwebsite.com/Los ... sPE10.html. Look at his slope graph. But what’s there is stock-bond difference. What we need instead, relative to PE10 extent above/below average, is slope of next-few-years stocks return rate average being below/above long-term average. This we can incorporate in our estimates for future-return-rate means of next few years.

8. Then each of us can incorporate that in return-rate-probability estimates, in assessments of result probabilities for that investor’s real purpose of dollars, for that investor’s particular dollar plan and goals.

9. I’ll betcha that for most folks who view allocation targets in sensible round numbers, such as stocks 50% vs. 55% or 60% but not 54.309%, these adjustments will call for no changes in target allocations.

10. And I’ll betcha zillions that where these adjustments do point to different allocations, say 55% instead of 50%, the significance of these adjustments is just dwarfed by uncertainties about what our basic long-term-average estimates should be for return-rate means.

I think that for real people with sensible (dollar) goals, what Wade’s research suggests has validity – higher PE10, likely lower average returns next few years – but that its significance is dwarfed by more fundamental uncertainties about the future.

Dick Purcell
Last edited by Dick Purcell on Mon May 30, 2011 11:14 am, edited 1 time in total.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox »

Mel Lindauer wrote:
My point is this: When one group in the study would have paid much higher commissions, trading costs and taxes (the market timers) than the other group (the buy-and-holders), to ignore that in the study skews the results in favor of the market-timers. Folks like the OP and some others seem to be ignoring that simple fact.
I think that's a valid critique.

These studies exist in the realm of academia. Some researchers make a set of assumptions and then they make an assertion. They come up with a methodology to prove or disprove that assertion. Later, someone else comes along and questions the original assumptions or methodology.

Professor Pfau did this very thing. He questioned some of the assumptions and methodology in the original Fisher and Statman study. He came up with a different approach and got different results. Now you can question Pfau's set of assumptions and critique his methodology.

Fair enough.

As far as people taking it too seriously. Well this is thread filed under Investing Theory. I think it would be foolish to take some nascent theory discussed herein as actual advice and run with it.

I even read where Markowitz doesn't use Mean Variance Optimation. He just picks 50/50. Doesn't mean the MPT ideas aren't useful. There is the theoretical world of academia, where investors earned 10% return over 200+ years, and then there is the real world where we invest actual money in whatever investment products happen to be available at the time for 30 or 40 years, first making contributions, then taking withdrawals, paying taxes and fees along the way.

The real world is 1000x more complex than the hypothetical simulations and spreadsheets. The question is how much of that complexity needs to be captured in a model and how much can be ignored and still get useful results.
Last edited by grayfox on Mon May 30, 2011 11:21 am, edited 3 times in total.
User avatar
Mel Lindauer
Moderator
Posts: 35757
Joined: Mon Feb 19, 2007 7:49 pm
Location: Daytona Beach Shores, Florida
Contact:

Post by Mel Lindauer »

fredflinstone wrote:
Mel Lindauer wrote:It appears from this thread that at least a few folks think this is the Holy Grail of investing, and leaving this kind of stuff unchallenged is dangerous.
If you ask me, a few folks think the standard Boglehead approach (fixed asset allocation; stay the course regardless of valuation) is the Holy Grail of Investing.
I don't think I've seem folks pitching any "Boglehead approach" on this thread. Rather, we're merely pointing out the flaws in Wade's study which you (and some others) seem to feel is that Holy Grail I mentioned. I feel that's dangerous, and I'm simply pointing that out for present and future readers. Others posters are doing the same by pointing to the study's shortcomings and limitations.
Best Regards - Mel | | Semper Fi
Post Reply