The cult of PE10 - so what?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Yipee-Ki-O
Posts: 167
Joined: Sun May 24, 2009 12:42 pm

Re: The cult of PE10 - so what?

Post by Yipee-Ki-O » Sun Oct 27, 2013 2:30 pm

Prokofiev wrote:
FFFollower wrote:I have been reading quite a bit here recently.
I notice a rather odd paradox, and I would like to solicit thoughts.

Why then the interest in debating PE10?
What is actionable as a result of this calculation, whilst also considering the prove points of Market Timing?

- Jason
You cannot accuse the BogleHeads of being a part of the "Cult of PE10", because Bogleheads.org was originally founded in an effort to ban a poster (Rob Bennett) from discussing PE10 and "valuation-informed investing" (his term). If anything, we are the anti-cult.
Just a point of clarification, Mr. Bennett wasn't banned because he wanted to "discuss" his investing strategy. He was banned because he demanded that his views and beliefs be accepted without question and that he be afforded the appropriate degree of adulation commensurate with his having found the "Holy Grail" of investing; high returns at no risk. And for his stupendous insight that safe withdrawal rates determined for past periods of time might not necessarily apply to future periods of time. And when, oddly enough, sufficient deference and praise were not directed his way he then proceeded to hijack each and every thread so that he and his collection of beliefs would assume their proper roles as the center-of-attention.

User avatar
docneil88
Posts: 927
Joined: Mon Apr 30, 2007 3:39 pm
Location: Taxable

Re: The cult of PE10 - so what?

Post by docneil88 » Sun Oct 27, 2013 3:47 pm

MnD wrote:PE10 is barely more predictive than PE1.
I wonder if anyone has run numbers to see whether combining PE1 and PE10 is more "predictive" in long-term backtests than either individually, and if so by how much. My hunch is that the combo is more predictive. Best, Neil

User avatar
Kevin M
Posts: 10197
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: The cult of PE10 - so what?

Post by Kevin M » Sun Oct 27, 2013 4:36 pm

I think the question in the OP is a good one. The most valuable thing for me in this thread was the link to the Vanguard paper on forecasting stock returns; thanks for that!

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

JOtar
Posts: 21
Joined: Sun Aug 23, 2009 5:35 pm

Re: The cult of PE10 - so what?

Post by JOtar » Mon Oct 28, 2013 12:36 am

While PE10 might be important concept, it mostly applies to accumulation portfolios. Thousands are retiring every day, and its impact is a lot more severe to distribution portfolios, as you are withdrawing from assets.

Several years ago (June 2007) I wrote an article that looks at the correlation between the (distribution) portfolio life and the prevailing PE at the time of retirement. Here is the link to that article: http://www.retirementoptimizer.com/arti ... CLE103.pdf

This (PE versus portfolio longevity) is covered under Warning Signal #1. You might find it interesting.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 1:53 pm

William Million wrote:We have PE10 because we have a decimal (base-10) number system. We came to use base-10 because our ancestors had 10 fingers.

There is no logical reason that PE10 is a more accurate predictor than PE7 or PE9 or PE11. While it is due to cultural evolution that we have base-10 and PE10, we should not expect equities to conform to these cultural biases.

Some researchers back-test to prove PE10 is predictive. However, the San Diego Padres posting a winning record also proved highly predictive of bull markets. (Well, there are 30 teams, so it had to work with at least one of them.)
So the data shouldn't lead our thought because someone made a spurious correlaton somewhere?

I haven't tested PE7 or PE9 or PE11. But I have tested PE5 and it isn't useful as a predictor. PE10 is. There is a lot of academic research to back this up so let's go with PE10.

PE10 does have some value as a long-term predictor. I believe that it is a promising line of research that is beginning to bear fruit in the form of models with weak explanatory power. That may sound like a bad thing, but we have to seriously question any variable with strong predictive power as being spurious. If that weren't true, well I'd be a lot richer than I am.

These weak variables are promising IMO because they may hold the key to adding an additional point or two of long-term annual return to a portfolio. Personally, I think PE10 is limited. But in conjunction with other variables (such as 10 year treasury rates), the evidence suggests that we have a useful model.

Case in point. I use a formula which predicts the 10 year equity risk premium based upon treasuries and PE10. This indicator has recently dropped into its lowest decile. Historically, this has resulted in a 10-year CAGR of -2.9% to +6.3% with a mean of 1.8%. In contrast, in 2009 we were looking at 10 year returns ranging from 1.5% to 10.6% with a mean of 6.5%. Not bad at all.

What you do with that information is up to you. The Boglehead philosophy is a good one. If you don't want to stress out about a couple of extra points of return, then ignore it and you'll probably do just fine. Alternatively, you might decide to reduce your equity allocation somewhat based on this information. Also probably a good decision (which comes with a little more work). For me, I'll take +1-2% annual over 20 years, thank you very much :-)

One thing is for certain, these discussions about whether or not to pull out of the market based on PE10 show a lack of basic understanding of what this indicator is. It can't be used for that purpose. If the potential for market timing exists (which I doubt), PE10 is not the answer.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 1:54 pm

docneil88 wrote:
MnD wrote:PE10 is barely more predictive than PE1.
I wonder if anyone has run numbers to see whether combining PE1 and PE10 is more "predictive" in long-term backtests than either individually, and if so by how much. My hunch is that the combo is more predictive. Best, Neil
Yes, some time ago. I found no value in PE1 either alone or in conjunction with other variables I tested. For what that's worth.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 2:07 pm

Rodc wrote: Also, it makes no sense to look at valuations in isolation. If bond returns are very low that also effects how the future may play out. I am very surprised we never seem to see a P/E10 AND bond yield market prediction. That would make more sense but I still doubt it would have much predictive power.
Actually, your intuition is right on. Those two variables combined predict 10 year average returns with an R2 of .67. Again, what you can do with the information is pretty limited but the relationship is most definitely real.

User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: The cult of PE10 - so what?

Post by Clearly_Irrational » Tue Oct 29, 2013 2:11 pm

It's not that 10 years is particularly special, it's just that longer time periods tend to give you a better idea of what "normal" is. The longer the period the more smoothing effect it will have but the more data you'll lose at the beginning of the graph. Going too long might also introduce problems by including info from other "eras" where things were significantly different, I'm not sure PE50 for example would be that good an measure. I haven't run the numbers, but my guess is the sweet spot is probably somewhere more than 5 but less than 25. Here's a link to the S&P 500 Earnings by year if someone wants to do a fancy analysis:

http://www.multpl.com/s-p-500-earnings/table

User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: The cult of PE10 - so what?

Post by Clearly_Irrational » Tue Oct 29, 2013 2:12 pm

ot1138 wrote:
Rodc wrote: Also, it makes no sense to look at valuations in isolation. If bond returns are very low that also effects how the future may play out. I am very surprised we never seem to see a P/E10 AND bond yield market prediction. That would make more sense but I still doubt it would have much predictive power.
Actually, your intuition is right on. Those two variables combined predict 10 year average returns with an R2 of .67. Again, what you can do with the information is pretty limited but the relationship is most definitely real.

Hmm, tie it into safe withdrawal rates maybe?

User avatar
Kevin M
Posts: 10197
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: The cult of PE10 - so what?

Post by Kevin M » Tue Oct 29, 2013 2:14 pm

ot1138 wrote: Yes, some time ago. I found no value in PE1 either alone or in conjunction with other variables I tested. For what that's worth.
Did you read the Vanguard research paper linked above? They found PE10 only slightly more predictive than PE1 for 10-year returns--very different than your conclusion of no value in PE1. Since neither one explains more than about 40% of variation in returns, I wouldn't have a lot of confidence in a strategy relying on either one generating excess return.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 2:20 pm

Clearly_Irrational wrote:
ot1138 wrote:
Rodc wrote: Also, it makes no sense to look at valuations in isolation. If bond returns are very low that also effects how the future may play out. I am very surprised we never seem to see a P/E10 AND bond yield market prediction. That would make more sense but I still doubt it would have much predictive power.
Actually, your intuition is right on. Those two variables combined predict 10 year average returns with an R2 of .67. Again, what you can do with the information is pretty limited but the relationship is most definitely real.

Hmm, tie it into safe withdrawal rates maybe?
I saw that comment earlier (was it you?). I think that's pretty genius. Definitely going to look into it some more. If you can point me anywhere for hints, I'd be truly grateful.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 2:26 pm

Kevin M wrote:
ot1138 wrote: Yes, some time ago. I found no value in PE1 either alone or in conjunction with other variables I tested. For what that's worth.
Did you read the Vanguard research paper linked above? They found PE10 only slightly more predictive than PE1 for 10-year returns--very different than your conclusion of no value in PE1. Since neither one explains more than about 40% of variation in returns, I wouldn't have a lot of confidence in a strategy relying on either one generating excess return.

Kevin
No, I haven't. But I've tested it in the past and found that PE1 wasn't terribly useful. PE10 is quite useful but you have to go beyond a simple regression to tease out the data (decile analysis is incredibly useful for this). In any case, my time frame isn't one year and I'm not interested in market timing, so the 10 year version is more suitable to me.

User avatar
Kevin M
Posts: 10197
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: The cult of PE10 - so what?

Post by Kevin M » Tue Oct 29, 2013 3:04 pm

^The Vanguard paper looked at 10-year returns, not 1-year returns. Why don't you read it and tell us how your research is superior to theirs?

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: The cult of PE10 - so what?

Post by Rodc » Tue Oct 29, 2013 3:09 pm

ot1138 wrote:
Clearly_Irrational wrote:
ot1138 wrote:
Rodc wrote: Also, it makes no sense to look at valuations in isolation. If bond returns are very low that also effects how the future may play out. I am very surprised we never seem to see a P/E10 AND bond yield market prediction. That would make more sense but I still doubt it would have much predictive power.
Actually, your intuition is right on. Those two variables combined predict 10 year average returns with an R2 of .67. Again, what you can do with the information is pretty limited but the relationship is most definitely real.

Hmm, tie it into safe withdrawal rates maybe?
I saw that comment earlier (was it you?). I think that's pretty genius. Definitely going to look into it some more. If you can point me anywhere for hints, I'd be truly grateful.
Poster cjking has done some work along these lines using P/E10 only. See first page of this thread for a mention. A search may turn up more.

The monthly Shiller data would be a good dataset as it has P, E, and 10-year bond rates. Might not be the best bonds rates to use, but it at least lines up in time with the other data.

My personal take is that moderating withdrawal rates is one of the more likely sensible uses for valuations, as opposed to tactical allocation moves (or whatever market timing description one wishes to use).
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 3:39 pm

Kevin M wrote:^The Vanguard paper looked at 10-year returns, not 1-year returns. Why don't you read it and tell us how your research is superior to theirs?

Kevin
I'm not interested in comparing sizes with Vanguard but I will tell you that I have a research background so I'm well equipped to do the analysis. I can tell you that on a monthly basis using data using data from April 1953 to Sept 2002, I get a R2=.51 when regressing PE10 against 10 year S&P 500. It's an ok relationship, not great. Sounds like I have a slightly better result than them but the results are consistent (perhaps they used a different timeframe... why don't you tell me?).

In any case, I think we're getting way off the point. I'm not saying that PE10 is particularly reliable when used in isolation. It is reliable when used in conjunction with treasury rates. The R2 of my predicted 10-year CAGR vs actual is .67. Using the most recent 5 year (out-of-sample) data, my R2 increases to .71. Hardly surprising, my results are pretty consistent with the academic literature.

It's pretty well proven that there are weak long-term predictors of stock market performance. The real question is what can you do with this information?

User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: The cult of PE10 - so what?

Post by Clearly_Irrational » Tue Oct 29, 2013 4:28 pm

ot1138 wrote:In any case, I think we're getting way off the point. I'm not saying that PE10 is particularly reliable when used in isolation. It is reliable when used in conjunction with treasury rates. The R2 of my predicted 10-year CAGR vs actual is .67. Using the most recent 5 year (out-of-sample) data, my R2 increases to .71. Hardly surprising, my results are pretty consistent with the academic literature.

It's pretty well proven that there are weak long-term predictors of stock market performance. The real question is what can you do with this information?
Out of curiousity which data are you using for treasury rates and how are you combining the two?

User avatar
Kevin M
Posts: 10197
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

Re: The cult of PE10 - so what?

Post by Kevin M » Tue Oct 29, 2013 4:51 pm

ot1138 wrote: ... but I will tell you that I have a research background so I'm well equipped to do the analysis.
That would make your comments on the paper more valuable.
ot1138 wrote:I can tell you that on a monthly basis using data using data from April 1953 to Sept 2002, I get a R2=.51 when regressing PE10 against 10 year S&P 500. It's an ok relationship, not great. Sounds like I have a slightly better result than them but the results are consistent (perhaps they used a different timeframe... why don't you tell me?).
They use 10-year annualized total real returns for "the broad US stock market" from 1926-2011. R^2 for PE10 is 0.43 and for PE1 it's 0.38. The paper discusses different smoothed PEs quite a bit, but finds no one that is particularly more reliable than the others. I think it's an interesting discussion, and it contradicts your dismissal of PE 1, which is what I was interested in learning more about.
U.S. stock market returns. S&P 90 from January 1926 through March 3, 1957; S&P 500 Index from March 4, 1957, through December 1974; Dow Jones Wilshire 5000 Index from January 1957 through April 22, 2005; MSCI US Broad Market Index
thereafter. Returns are geometrically annualized and converted to real terms using the CPI data described above.
ot1138 wrote:In any case, I think we're getting way off the point. I'm not saying that PE10 is particularly reliable when used in isolation. It is reliable when used in conjunction with treasury rates. The R2 of my predicted 10-year CAGR vs actual is .67. Using the most recent 5 year (out-of-sample) data, my R2 increases to .71. Hardly surprising, my results are pretty consistent with the academic literature.
The Vanguard paper evaluates a number of forecasting metrics. One of them is the Fed model (for which R^2 was only 0.16):
Fed Model: the spread between U.S. stock earnings yield and the long-term government bond yield (the spread between the inverse of P/E1 and the level of the 10-year Treasury yield).
I didn't see a multi-variable model combining PE 10 with treasury rates, which I understand is what you are doing. Vanguard does mention that they use a proprietary forecasting model:
the Vanguard Capital Markets Model incorporates various valuation metrics (including those discussed here) blended in a
proprietary manner to produce a distribution of expected equity risk premiums.
ot1138 wrote:It's pretty well proven that there are weak long-term predictors of stock market performance. The real question is what can you do with this information?
Yes. For me the answer right now is nothing. The predictive power is too weak, and the distributions too large to convince me that any changes to my AA are warranted.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)

alexfrey
Posts: 33
Joined: Tue Oct 29, 2013 6:02 pm
Location: San Francisco
Contact:

Re: The cult of PE10 - so what?

Post by alexfrey » Tue Oct 29, 2013 6:36 pm

I'm new here, but thought this was an interesting discussion, so much so that I had to jump in. A few points:
  • A correlation of .4 (or anything over .2 really) is not something to be dismissed when dealing with noisy stock market data. A "coin flip" would be a correlation of 0, not .5 as someone suggested.
  • Regardless of the empirical data, there is strong theoretical reason to predict that PE10 should (weakly) predict future returns, because the earnings yield (inverse of PE) is in some sense a measure of what you are getting for your $ when you buy stocks
  • I don't think having a 7-10 year holding period is "asking too much." If you don't have that long of a horizon you shouldn't be investing in the stock market, period.
Is PE10 a "silver bullet?" Of course not. But I don't see why that means it can't be a useful that will be combined with other things when deciding how to allocate your assets or when and how to rebalance.

Some ways it can be used:
  • Shifting allocations to US, Europe, Japan on the margins based on their relative attractiveness as measured by PE10
  • Shifting allocations to stocks / bonds on the margins based on difference between inverse PE10 ("cyclically adjusted earnings yield") and 10 year TIPs yield
  • Using to estimate a long-run return for retirement planning purposes

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 8:43 pm

I couldn't agree more with this. The misunderstanding of the utility of PE10 seems to stem from confusing the ability predict a specific outcome vs. a probability of outcomes. PE10 is useful in the latter regard only. That suggests that it *should* be useful as a tactical allocation tool.

How? I don't know. But let's not rule it out. If such a thing is possible, it's not at all in conflict with Boglehead philosophy... it would simply be a nuance that we just don't fully understand or appreciate (yet).

Investment theory has come a long way in 50 years. I can't imagine there's nothing new yet to learn.

ot1138
Posts: 146
Joined: Sun Mar 10, 2013 12:46 pm

Re: The cult of PE10 - so what?

Post by ot1138 » Tue Oct 29, 2013 9:15 pm

Kevin M wrote:
ot1138 wrote: ... but I will tell you that I have a research background so I'm well equipped to do the analysis.
That would make your comments on the paper more valuable.
Ok, I'll bite.
Kevin M wrote:
ot1138 wrote:I can tell you that on a monthly basis using data using data from April 1953 to Sept 2002, I get a R2=.51 when regressing PE10 against 10 year S&P 500. It's an ok relationship, not great. Sounds like I have a slightly better result than them but the results are consistent (perhaps they used a different timeframe... why don't you tell me?).
They use 10-year annualized total real returns for "the broad US stock market" from 1926-2011. R^2 for PE10 is 0.43 and for PE1 it's 0.38. The paper discusses different smoothed PEs quite a bit, but finds no one that is particularly more reliable than the others. I think it's an interesting discussion, and it contradicts your dismissal of PE 1, which is what I was interested in learning more about.
I calculated PE1 using the trailing 12 month earnings from Shiller's website and regressed it against 10 year returns for the period 4/1953 - 9/2002. R2 was .43. As I mentioned above, PE10 gives .51 for the same time period. Both are useful, but PE10 is significantly more so.

Seems my results are pretty consistent with Vanguard's, only difference being the timeframe. I think that's where they went wrong. Earnings data prior to 1934 was pretty sketchy and there's that messy matter of survivor bias in 1929. I avoided all of that using data from 1953 on.
Kevin M wrote: I didn't see a multi-variable model combining PE 10 with treasury rates, which I understand is what you are doing. Vanguard does mention that they use a proprietary forecasting model:
the Vanguard Capital Markets Model incorporates various valuation metrics (including those discussed here) blended in a
proprietary manner to produce a distribution of expected equity risk premiums.
This strikes me as a very intelligent approach which is grounded in research. We can't predict events but there's a lot of evidence that we can predict probability distributions. I'm actually working on something similar myself. My theory is that it will enhance returns but perhaps not enough so inclusive of fees to be attractive to most investors. Having my own proprietary model would give me the basis to found a durable family empire (that sounds really cool, doesn't it?). In other words, I don't expect a huge benefit in my lifetime but over the next 50-75 years, my heirs should be pretty well set.

mushyyy
Posts: 48
Joined: Tue Jan 13, 2015 1:34 pm

Re: The cult of PE10 - so what?

Post by mushyyy » Mon Feb 02, 2015 7:03 am

Apologies for bumping up this thread but I recently did my own research on the PE aspect and how can individual investors benefit from PE value.
I just want to share my findings and maybe you can help me find some flaws in my thinking.

did my own research to understand better the situation.
I took the S&P500 data from 1881 to 2013 and I calculated the return of 7 different strategies:

1) Buy and Hold
2) Sell our position when P/E10> 15 and buy when P/E10< 15
3) Sell our position when P/E10> 20 and buy when P/E10< 20
4) Sell our position when P/E10> 23 and buy when P/E10< 23
5) Sell our position when Monthly P/E > 15 and buy when monthly P/E10< 15
6) Sell our position when Monthly P/E > 20 and buy when monthly P/E10< 20
7) Sell our position when Monthly P/E > 23 and buy when monthly P/E10< 23

Here are the results of $10,000 invested in 1881 and with all the dividends re-invested at end of each year (I assume no fees, taxes etc)

Buy and Hold: $750,711,402 - CAGR: 8,9%
Sell our position when P/E10> 15 $2,942,434 - CAGR: 4,4%
Sell our position when P/E10> 20 $22,045,707 - CAGR: 6,0%
Sell our position when P/E10> 23 $165,773,291 - CAGR 7,6%
Sell our position when Monthly P/E > 15 $1,252,739 - CAGR 3,7%
Sell our position when Monthly P/E > 20 $244,122,209 - CAGR 8,0%
Sell our position when Monthly P/E > 23 $465,429,112 - CAGR 8,5%


At the end of the year, the dividends are used to purchase new shares at a December's price. If the holding period is less than 12months (because of PE10's value above/below the threshold), then only a proportion of yearly dividends have been used to purchase new stocks.

Example:
Dividend year X: $10
Holding period year X: From March to October (6 months)
Holding period dividend: $5
This $5 have been used to purchase new shares in October at October's price.


Here is the link to the spreadsheet, which was not meant to shared and I apologize for the poor format:
http://www.filedropper.com/pespreturns

randomguy
Posts: 6496
Joined: Wed Sep 17, 2014 9:00 am

Re: The cult of PE10 - so what?

Post by randomguy » Mon Feb 02, 2015 1:31 pm

You also need to look at volatility to determine if you are reducing risk by doing this. My other question is what are you investing that money in when not in stocks?

I am sure if you gave me enough time I could come up with some killer PE10 timing scheme (gradual shifts to and from stocks, using some moving average of PE10 to allow investment in the 90s/2000s, some hystersis, maybe look at rate of change,...) that gives very good results. I would have no faith in them going forward though.

Beliavsky
Posts: 690
Joined: Sun Jun 29, 2014 10:21 am

Re: The cult of PE10 - so what?

Post by Beliavsky » Mon Feb 02, 2015 1:35 pm

FFFollower wrote:I have been reading quite a bit here recently. I notice a rather odd paradox, and I would like to solicit thoughts. I will accept that PE10 has a degree of predictive capabilities, albeit with much skepticism. I will also accept that Market Timing has been proven, with more rigor then the proof of PE10's predictive capabilities, to be less effective long-term then buy and hold - i.e. it doesn't work. Why then the interest in debating PE10? What is actionable as a result of this calculation, whilst also considering the prove points of Market Timing? Or is this merely an academic discussion born from boredom and too much research?

- Jason
I have seen studies of how PE10 predicts long-term (for example 10 year) returns. Have there been studies of how PE10 predicts 1-year returns? Someone who thinks that valuation should guide their asset allocation would look at valuation measures at least annually.

User avatar
JoMoney
Posts: 6116
Joined: Tue Jul 23, 2013 5:31 am

Re: The cult of PE10 - so what?

Post by JoMoney » Mon Feb 02, 2015 2:58 pm

Beliavsky wrote:...I have seen studies of how PE10 predicts long-term (for example 10 year) returns. Have there been studies of how PE10 predicts 1-year returns? Someone who thinks that valuation should guide their asset allocation would look at valuation measures at least annually.
The Vanguard paper posted earlier in the thread does a little bit, showing correlation for both 1-year and 10-year...
https://personal.vanguard.com/pdf/s338.pdf
Image
One of the more interesting aspects of their study, is that they point out that the high correlation for the PE and PE10 (high relative to other models) is driven only at the extremes. When the market was somewhere in the middle of the extremes the correlation was very low... The fat-"tails" wagging the dog...
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

Leeraar
Posts: 4109
Joined: Tue Dec 10, 2013 8:41 pm
Location: Nowhere

Re: The cult of PE10 - so what?

Post by Leeraar » Mon Feb 02, 2015 3:19 pm

Forget PE10. The market is down in every year that New England has won the Super Bowl. Sell!!

L.
You can get what you want, or you can just get old. (Billy Joel, "Vienna")

Beliavsky
Posts: 690
Joined: Sun Jun 29, 2014 10:21 am

Re: The cult of PE10 - so what?

Post by Beliavsky » Mon Feb 02, 2015 3:22 pm

JoMoney wrote:
Beliavsky wrote:...I have seen studies of how PE10 predicts long-term (for example 10 year) returns. Have there been studies of how PE10 predicts 1-year returns? Someone who thinks that valuation should guide their asset allocation would look at valuation measures at least annually.
The Vanguard paper posted earlier in the thread does a little bit, showing correlation for both 1-year and 10-year...
https://personal.vanguard.com/pdf/s338.pdf
Image
Thank you. The graph actually shows the "proportion of variance explained" (R^2), which is the square of the correlation. A 0.2 correlation, which one may consider non-zero (but not very strong, either), amounts to only a 0.04 R^2.

User avatar
JoMoney
Posts: 6116
Joined: Tue Jul 23, 2013 5:31 am

Re: The cult of PE10 - so what?

Post by JoMoney » Mon Feb 02, 2015 3:23 pm

Leeraar wrote:Forget PE10. The market is down in every year that New England has won the Super Bowl. Sell!!

L.
What's the signal to start buying again? ... and I thought people should be buying while the market is going down, selling when it's going up?
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

mushyyy
Posts: 48
Joined: Tue Jan 13, 2015 1:34 pm

Re: The cult of PE10 - so what?

Post by mushyyy » Tue Feb 03, 2015 2:56 am

randomguy wrote:You also need to look at volatility to determine if you are reducing risk by doing this. My other question is what are you investing that money in when not in stocks?

I am sure if you gave me enough time I could come up with some killer PE10 timing scheme (gradual shifts to and from stocks, using some moving average of PE10 to allow investment in the 90s/2000s, some hystersis, maybe look at rate of change,...) that gives very good results. I would have no faith in them going forward though.
Actually, I was impressed by the superiority of the B&H strategy over the different market timing tactics, that's why I wanted to discuss it.
Because there are many threads on this forum discussing the value of PE strategies over B&H, value that I cannot see in my analysis.

cjking
Posts: 1815
Joined: Mon Jun 30, 2008 4:30 am

Re: The cult of PE10 - so what?

Post by cjking » Tue Feb 03, 2015 5:02 am

mushyyy wrote: Actually, I was impressed by the superiority of the B&H strategy over the different market timing tactics, that's why I wanted to discuss it.
In that case you need to find out what the average exposure of each strategy was to equities and compare each with a suitable benchmark. For example your strategy of switching in/out at a PE10 of 15 might need to be compared with a 50% equity buy-and-hold portfolio. Your buy-and-hold benchmark has higher average equity exposure/more risk than all the timing strategies you are comparing it with.

FWIW my vague recollection is that when I performed very similar analysis several years ago, using the Shiller data, the timing strategies added value, though less than 1% a year. The threads are probably still here. I think others have come to similar conclusions.

Having said that, this use of PE10 (to switch in and out of equities) is not really of great interest to anyone here, even if it would have added value in the past, as it is open to two criticisms:-
1. The future may be very different from the past, you run the risk of being out of equities permanently if the average PE multiple is changing.
2. Equities can still be the best-performing asset class, even when they are expensive by historical standards.

Someone has probably already said this (too lazy to check) but an example of a better way to use PE10 would be to adjust asset allocation to take into account that currently European shares have smoothed earnings yields roughly 3% higher then US shares. (I have Europe at 3.4% higher at the moment, also Emerging Markets at 2.6% higher and Asia-excluding-Japan at 2.8% higher.)

Leeraar
Posts: 4109
Joined: Tue Dec 10, 2013 8:41 pm
Location: Nowhere

Re: The cult of PE10 - so what?

Post by Leeraar » Tue Feb 03, 2015 5:19 am

cjking wrote:
mushyyy wrote: Actually, I was impressed by the superiority of the B&H strategy over the different market timing tactics, that's why I wanted to discuss it.
In that case you need to find out what the average exposure of each strategy was to equities and compare each with a suitable benchmark. For example your strategy of switching in/out at a PE10 of 15 might need to be compared with a 50% equity buy-and-hold portfolio. Your buy-and-hold benchmark has higher average equity exposure/more risk than all the timing strategies you are comparing it with.

FWIW my vague recollection is that when I performed very similar analysis several years ago, using the Shiller data, the timing strategies added value, though less than 1% a year. The threads are probably still here. I think others have come to similar conclusions.

Having said that, this use of PE10 (to switch in and out of equities) is not really of great interest to anyone here as it is open to two criticisms:-
1. The future may be very different from the past, you run the risk of being out of equities permanently if the average PE multiple is changing. Another version of this criticism is that thresholds that are optimal with hindsight can't be known in advance.
2. Equities can still be the best-performing asset class, even when they are expensive by historical standards.

Someone has probably already said this (too lazy to check) but an example of a better way to use PE10 would be to adjust asset allocation to take into account that currently European shares have smoothed earnings yields roughly 3% higher then US shares. (I have Europe at 3.4% higher at the moment, also Emerging Markets at 2.6% higher and Asia-excluding-Japan at 2.8% higher.)
Cool! let us know how it turns out. I'll bet that the Patriots win the Super Bowl forecast is more accurate: Vanguard Total Stock Market Index Fund Admiral Shares (VTSAX) will be down for calendar year 2015.

L.
You can get what you want, or you can just get old. (Billy Joel, "Vienna")

mushyyy
Posts: 48
Joined: Tue Jan 13, 2015 1:34 pm

Re: The cult of PE10 - so what?

Post by mushyyy » Tue Feb 03, 2015 5:42 am

cjking wrote:
mushyyy wrote: Actually, I was impressed by the superiority of the B&H strategy over the different market timing tactics, that's why I wanted to discuss it.
In that case you need to find out what the average exposure of each strategy was to equities and compare each with a suitable benchmark. For example your strategy of switching in/out at a PE10 of 15 might need to be compared with a 50% equity buy-and-hold portfolio. Your buy-and-hold benchmark has higher average equity exposure/more risk than all the timing strategies you are comparing it with.

FWIW my vague recollection is that when I performed very similar analysis several years ago, using the Shiller data, the timing strategies added value, though less than 1% a year. The threads are probably still here. I think others have come to similar conclusions.

Having said that, this use of PE10 (to switch in and out of equities) is not really of great interest to anyone here, even if it would have added value in the past, as it is open to two criticisms:-
1. The future may be very different from the past, you run the risk of being out of equities permanently if the average PE multiple is changing.
2. Equities can still be the best-performing asset class, even when they are expensive by historical standards.

Someone has probably already said this (too lazy to check) but an example of a better way to use PE10 would be to adjust asset allocation to take into account that currently European shares have smoothed earnings yields roughly 3% higher then US shares. (I have Europe at 3.4% higher at the moment, also Emerging Markets at 2.6% higher and Asia-excluding-Japan at 2.8% higher.)
Thanks for the reply. I agree that both methods should be compared on a risk-asjusted basis.
I thought that market timing strategies based on PE methods would generate nominal higher return than B&H. Yet, it seems they add value only if compared on a risk adjusted basis.

For for my curiosity. Has anyone found a strategy with the same risk of a B&H portfolio 100% equity, than generates higher return over a period of many years (50-100y)?

Leeraar
Posts: 4109
Joined: Tue Dec 10, 2013 8:41 pm
Location: Nowhere

Re: The cult of PE10 - so what?

Post by Leeraar » Tue Feb 03, 2015 8:40 am

mushyyy wrote:For for my curiosity. Has anyone found a strategy with the same risk of a B&H portfolio 100% equity, than generates higher return over a period of many years (50-100y)?
So,

You are asking, has anyone found a market timing strategy that generates a higher return with the same risk as my market timing strategy?

L.
You can get what you want, or you can just get old. (Billy Joel, "Vienna")

Caduceus
Posts: 1728
Joined: Mon Sep 17, 2012 1:47 am

Re: The cult of PE10 - so what?

Post by Caduceus » Tue Feb 03, 2015 8:59 am

P/E10 is not a metric like a company-size metric (which is essentially just P in aggregate).

The question I have for all these quantitatively-minded number-crunchers is if changes in GAAP over their period of study are material or not.

-- Did the move by American corporations to book stock options as an expense change the pattern of P/E10 during this period?
-- How significant were GAAP pronouncements on the re-classifications of operating vs. financial leases on corporate earnings?
-- What happened at the turn of the century to accounting earnings when the amortization of goodwill was removed, to be replaced by an impairment test?
-- How do persistent medium-term movements in foreign exchange rates (and therefore foreign currency translations) affect the E in any given period?

Have any of these statisticians actually also tried to apply some solid accounting to the data? It's funny how some folks in this thread immediately launch into discussions of Correlation and Variance when all indications suggest they know absolutely no accounting.
Someone who thinks that valuation should guide their asset allocation would look at valuation measures at least annually.
Someone who thinks that valuation should guide their asset allocation would worship not the false god of statistics, but pick up some real accounting - the natural language of business, of financial statements, and not incidentally, of P/E10. Accounting that would allow them to understand the utility and limits of numbers in investing, which are accounting artifacts, not natural artifacts.

mushyyy
Posts: 48
Joined: Tue Jan 13, 2015 1:34 pm

Re: The cult of PE10 - so what?

Post by mushyyy » Tue Feb 03, 2015 9:10 am

Leeraar wrote:
mushyyy wrote:For for my curiosity. Has anyone found a strategy with the same risk of a B&H portfolio 100% equity, than generates higher return over a period of many years (50-100y)?
So,

You are asking, has anyone found a market timing strategy that generates a higher return with the same risk as my market timing strategy?

L.
No my question is if anyone has found a market timing strategy that generates higher returns on a risk-adjusted basis, in comparison with a B&H 100% equity approach

Beliavsky
Posts: 690
Joined: Sun Jun 29, 2014 10:21 am

Re: The cult of PE10 - so what?

Post by Beliavsky » Tue Feb 03, 2015 9:59 am

Caduceus wrote:P/E10 is not a metric like a company-size metric (which is essentially just P in aggregate).

The question I have for all these quantitatively-minded number-crunchers is if changes in GAAP over their period of study are material or not.

-- Did the move by American corporations to book stock options as an expense change the pattern of P/E10 during this period?
-- How significant were GAAP pronouncements on the re-classifications of operating vs. financial leases on corporate earnings?
-- What happened at the turn of the century to accounting earnings when the amortization of goodwill was removed, to be replaced by an impairment test?
-- How do persistent medium-term movements in foreign exchange rates (and therefore foreign currency translations) affect the E in any given period?

Have any of these statisticians actually also tried to apply some solid accounting to the data? It's funny how some folks in this thread immediately launch into discussions of Correlation and Variance when all indications suggest they know absolutely no accounting.
How do you infer what other people know about accounting?

Beliavsky
Posts: 690
Joined: Sun Jun 29, 2014 10:21 am

Re: The cult of PE10 - so what?

Post by Beliavsky » Tue Feb 03, 2015 10:03 am

mushyyy wrote:Apologies for bumping up this thread but I recently did my own research on the PE aspect and how can individual investors benefit from PE value.
I just want to share my findings and maybe you can help me find some flaws in my thinking.

did my own research to understand better the situation.
I took the S&P500 data from 1881 to 2013 and I calculated the return of 7 different strategies:

1) Buy and Hold
2) Sell our position when P/E10> 15 and buy when P/E10< 15
3) Sell our position when P/E10> 20 and buy when P/E10< 20
4) Sell our position when P/E10> 23 and buy when P/E10< 23
5) Sell our position when Monthly P/E > 15 and buy when monthly P/E10< 15
6) Sell our position when Monthly P/E > 20 and buy when monthly P/E10< 20
7) Sell our position when Monthly P/E > 23 and buy when monthly P/E10< 23

Here are the results of $10,000 invested in 1881 and with all the dividends re-invested at end of each year (I assume no fees, taxes etc)

Buy and Hold: $750,711,402 - CAGR: 8,9%
Sell our position when P/E10> 15 $2,942,434 - CAGR: 4,4%
Sell our position when P/E10> 20 $22,045,707 - CAGR: 6,0%
Sell our position when P/E10> 23 $165,773,291 - CAGR 7,6%
Sell our position when Monthly P/E > 15 $1,252,739 - CAGR 3,7%
Sell our position when Monthly P/E > 20 $244,122,209 - CAGR 8,0%
Sell our position when Monthly P/E > 23 $465,429,112 - CAGR 8,5%
Thanks for your analysis. It would helpful to know what fraction of the time the various strategies were in the market. A strategy with slighlty lower returns than B&H that is out of the market a substantial fraction of the time may be better on a risk-adjusted basis.

Caduceus
Posts: 1728
Joined: Mon Sep 17, 2012 1:47 am

Re: The cult of PE10 - so what?

Post by Caduceus » Tue Feb 03, 2015 10:27 am

Beliavsky, what is your take on these questions?

-- Did the move by American corporations to book stock options as an expense change the pattern of P/E10 during this period?
-- How significant were GAAP pronouncements on the re-classifications of operating vs. financial leases on corporate earnings?
-- What happened at the turn of the century to accounting earnings when the amortization of goodwill was removed, to be replaced by an impairment test?
-- How do persistent medium-term movements in foreign exchange rates (and therefore foreign currency translations) affect the E in any given period?

pkcrafter
Posts: 13100
Joined: Sun Mar 04, 2007 12:19 pm
Location: CA
Contact:

Re: The cult of PE10 - so what?

Post by pkcrafter » Tue Feb 03, 2015 10:42 am

Keynes quote applies to using PE10 for AA adjustments--
Markets can remain irrational longer than you can remain solvent.
In other words, there may some predictive power to PE10 or any other measures, but the prediction may come about long enough in the future to cancel any benefit of acting on the indicator.
If an investor still insists on adjusting AA for any reason based on expectations, i.e. ignores BH philosophy, he/she should limit those adjustments to +/- 15% of equity allocation. Example = 80% equity down 15% to 68%. 40% down to 34%.


Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.

Beliavsky
Posts: 690
Joined: Sun Jun 29, 2014 10:21 am

Re: The cult of PE10 - so what?

Post by Beliavsky » Tue Feb 03, 2015 11:00 am

Caduceus wrote:Beliavsky, what is your take on these questions?

-- Did the move by American corporations to book stock options as an expense change the pattern of P/E10 during this period?
-- How significant were GAAP pronouncements on the re-classifications of operating vs. financial leases on corporate earnings?
-- What happened at the turn of the century to accounting earnings when the amortization of goodwill was removed, to be replaced by an impairment test?
-- How do persistent medium-term movements in foreign exchange rates (and therefore foreign currency translations) affect the E in any given period?
I don't know, although I will say that fluctuation in exchange rates occurred in the past, too. By how much and in what direction do you think these changes in accounting rules have affected reported earnings? A model is a simplification of reality but can still be useful.

User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: The cult of PE10 - so what?

Post by Clearly_Irrational » Tue Feb 03, 2015 11:56 am

Caduceus wrote:The question I have for all these quantitatively-minded number-crunchers is if changes in GAAP over their period of study are material or not.
I haven't yet found a consistent way to adjust for the two main changes (FAS 142 goodwill impairment vs. amortization & buybacks vs. dividends). I agree it's probably slightly overstated but I'm not sure exactly how much. Here is an article that discusses some of the issues: http://www.philosophicaleconomics.com/2013/12/shiller/

I'm afraid I don't agree with the authors solution or the size of the adjustment but it's at least one attempt at doing so. For me, since I only use it as a measure of market irrationality in combination with a number of other measures I don't stress about it too much. There is probably a good finance paper in there if someone comes up with an elegant solution.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: The cult of PE10 - so what?

Post by Rodc » Tue Feb 03, 2015 2:32 pm

JoMoney wrote:
Leeraar wrote:Forget PE10. The market is down in every year that New England has won the Super Bowl. Sell!!

L.
What's the signal to start buying again? ... and I thought people should be buying while the market is going down, selling when it's going up?
Market is at 100.

Buy 10 shares when it drops to 90.

Buy 10 more when it drops to 80.

It bottoms at 73 and since it did not hit 70 no more shares are bought.

You sell 10 shares at 80 on the way up.

You sell 10 more shares at 90.

You sell 10 more at 100.

How much money did you make?

Or if you wish, you invest $X at each step on the way down and sell a matching amount on the way up.

This if course is a simplification, but shows a basic flaw in the logic of buying on the way down and selling on the way up. To really make good money you have to end up buying very close to the bottom, then selling each lot at above the buying price, which does not automatically happen. For example, in the example, if you buy 10 shares when the market just happens to bottom at 70, then sell those at 80, sell the one you bought at 80 when the market hits 90, etc. But that takes some good luck.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

letsgobobby
Posts: 11647
Joined: Fri Sep 18, 2009 1:10 am

Re: The cult of PE10 - so what?

Post by letsgobobby » Tue Feb 03, 2015 3:20 pm

Rodc makes an important point that I realized after years of adhering to PE10. The only way to really benefit is by using momentum as well, which I guess turns out to be an independent factor, though I haven't read much about it.

I use momentum in its lay sense. I use PE10 to sell on the way up (above say 25) or buy on the way down (below 12), but then I do not buy (after selling) or sell (after buying) until PE10 gets back into its historic range (defined here as 12-18). I can explain more if necessary but that is basically what my IPS now says. Haven't had the opportunity to test it out yet, with PE10 refusing to move meaningfully above 26 (and I adjust current PE10 by about 4, per Larry's and others' calculations based on the change in GAAP in the early 2000s).

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: The cult of PE10 - so what?

Post by Rodc » Tue Feb 03, 2015 7:48 pm

letsgobobby wrote:Rodc makes an important point that I realized after years of adhering to PE10. The only way to really benefit is by using momentum as well, which I guess turns out to be an independent factor, though I haven't read much about it.

I use momentum in its lay sense. I use PE10 to sell on the way up (above say 25) or buy on the way down (below 12), but then I do not buy (after selling) or sell (after buying) until PE10 gets back into its historic range (defined here as 12-18). I can explain more if necessary but that is basically what my IPS now says. Haven't had the opportunity to test it out yet, with PE10 refusing to move meaningfully above 26 (and I adjust current PE10 by about 4, per Larry's and others' calculations based on the change in GAAP in the early 2000s).
Historically, prior to 2008, the best time to buy was when P/E10 was in single digits. If one waited in 2008 for single digits one is still waiting.

Nothing easy about this.

Using P/E10 might work; I'm not saying it can't. Just a general caution that while it is easy to use when backtesting, going forward is harder.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

User avatar
ray.james
Posts: 1040
Joined: Tue Jul 19, 2011 4:08 am

Re: The cult of PE10 - so what?

Post by ray.james » Tue Feb 03, 2015 8:12 pm

mushyyy wrote:
cjking wrote:
mushyyy wrote: Actually, I was impressed by the superiority of the B&H strategy over the different market timing tactics, that's why I wanted to discuss it.
In that case you need to find out what the average exposure of each strategy was to equities and compare each with a suitable benchmark. For example your strategy of switching in/out at a PE10 of 15 might need to be compared with a 50% equity buy-and-hold portfolio. Your buy-and-hold benchmark has higher average equity exposure/more risk than all the timing strategies you are comparing it with.

FWIW my vague recollection is that when I performed very similar analysis several years ago, using the Shiller data, the timing strategies added value, though less than 1% a year. The threads are probably still here. I think others have come to similar conclusions.

Having said that, this use of PE10 (to switch in and out of equities) is not really of great interest to anyone here, even if it would have added value in the past, as it is open to two criticisms:-
1. The future may be very different from the past, you run the risk of being out of equities permanently if the average PE multiple is changing.
2. Equities can still be the best-performing asset class, even when they are expensive by historical standards.

Someone has probably already said this (too lazy to check) but an example of a better way to use PE10 would be to adjust asset allocation to take into account that currently European shares have smoothed earnings yields roughly 3% higher then US shares. (I have Europe at 3.4% higher at the moment, also Emerging Markets at 2.6% higher and Asia-excluding-Japan at 2.8% higher.)
Thanks for the reply. I agree that both methods should be compared on a risk-asjusted basis.
I thought that market timing strategies based on PE methods would generate nominal higher return than B&H. Yet, it seems they add value only if compared on a risk adjusted basis.

For for my curiosity. Has anyone found a strategy with the same risk of a B&H portfolio 100% equity, than generates higher return over a period of many years (50-100y)?
I think this is incorrect way of looking at PE10 strategy used by most people.
From what I understand what Mr Bogle/ shiller did in 2000 and based on interviews they scaled the risk based on PE10

Lets say someone started at 60% equities/40% bonds asset allocation.(PE at 18)
If PE went under 12 rebalance to 80% equities/ 20%bonds.
If PE went under 10 rebalance to 90% equities/ 20% bonds.
If P/E went above 22 rebalance to 50% equities/ 50%bonds.
If P/E went above 26 rebalance to 40% equities/ 60%bonds.

The above numbers are made up. One can put the above as equation and find the optimal percentages and PE10 for historical data(not that it is going to repeat or that I am curious). But that is how I understood, people are using PE10 as a strategy for investments. There is a academic paper on this, posted sometime back by a forum member that I couldn't find now. But it turns out the bands of PE10 signalling such portfolio changes are clustered in 3 years periods and then it went quiet for few years. Business cycle I guess.
When in doubt, http://www.bogleheads.org/forum/viewtopic.php?f=1&t=79939

User avatar
siamond
Posts: 4219
Joined: Mon May 28, 2012 5:50 am

Re: The cult of PE10 - so what?

Post by siamond » Wed Feb 04, 2015 12:42 am

ot1138 wrote:
Rodc wrote: Also, it makes no sense to look at valuations in isolation. If bond returns are very low that also effects how the future may play out. I am very surprised we never seem to see a P/E10 AND bond yield market prediction. That would make more sense but I still doubt it would have much predictive power.
Actually, your intuition is right on. Those two variables combined predict 10 year average returns with an R2 of .67. Again, what you can do with the information is pretty limited but the relationship is most definitely real.
I do agree. Shifting the discussion to (expected) equity risk premium over bonds gets much more interesting than looking at equity valuation in a vacuum. I mean, the PE10 can be quite high (like now), if bond yields are at an historical low (like... now!), then it doesn't seem too likely that a lot of minimally rational investors will suddenly flock to bonds unless the PE10 truly goes to a 2000-like stratosphere... Everything is relative, in other words.

Now the trouble with such logic (which I played with at length) is that a bond yield is a nominal quantity and equity expected returns forecast (1/PE10 or DDM or whatever) is much better achieved in real terms. And then to compare the two, you need to make an assumption about inflation, and man, is the inflation random year over year... Sure, nowadays, we can use TIPS real yields, but they have been available since 2003, so no way one can do any meaningful backtesting with those. The oversized influence of inflation randomness gave me quite some grief in my attempts at combining those two variables, then backtest.

ot1138, would you mind elaborating on exactly how you combined those two variables?

mushyyy
Posts: 48
Joined: Tue Jan 13, 2015 1:34 pm

Re: The cult of PE10 - so what?

Post by mushyyy » Wed Feb 04, 2015 8:00 pm

Beliavsky wrote:
mushyyy wrote:Apologies for bumping up this thread but I recently did my own research on the PE aspect and how can individual investors benefit from PE value.
I just want to share my findings and maybe you can help me find some flaws in my thinking.

did my own research to understand better the situation.
I took the S&P500 data from 1881 to 2013 and I calculated the return of 7 different strategies:

1) Buy and Hold
2) Sell our position when P/E10> 15 and buy when P/E10< 15
3) Sell our position when P/E10> 20 and buy when P/E10< 20
4) Sell our position when P/E10> 23 and buy when P/E10< 23
5) Sell our position when Monthly P/E > 15 and buy when monthly P/E10< 15
6) Sell our position when Monthly P/E > 20 and buy when monthly P/E10< 20
7) Sell our position when Monthly P/E > 23 and buy when monthly P/E10< 23

Here are the results of $10,000 invested in 1881 and with all the dividends re-invested at end of each year (I assume no fees, taxes etc)

Buy and Hold: $750,711,402 - CAGR: 8,9%
Sell our position when P/E10> 15 $2,942,434 - CAGR: 4,4%
Sell our position when P/E10> 20 $22,045,707 - CAGR: 6,0%
Sell our position when P/E10> 23 $165,773,291 - CAGR 7,6%
Sell our position when Monthly P/E > 15 $1,252,739 - CAGR 3,7%
Sell our position when Monthly P/E > 20 $244,122,209 - CAGR 8,0%
Sell our position when Monthly P/E > 23 $465,429,112 - CAGR 8,5%
Thanks for your analysis. It would helpful to know what fraction of the time the various strategies were in the market. A strategy with slighlty lower returns than B&H that is out of the market a substantial fraction of the time may be better on a risk-adjusted basis.
I have only data for two strategies:
B&H strategy: Invested for 1595 months: 100%
PE10>23: Invested for 1398 months: 87%

How can I now evaluate which approach is better given the information above?

User avatar
William Million
Posts: 547
Joined: Wed May 05, 2010 4:41 am
Location: A Deep Mountain

Re: The cult of PE10 - so what?

Post by William Million » Sun Feb 08, 2015 2:24 am

mushyyy wrote:Apologies for bumping up this thread but I recently did my own research on the PE aspect and how can individual investors benefit from PE value.
I just want to share my findings and maybe you can help me find some flaws in my thinking.

did my own research to understand better the situation.
I took the S&P500 data from 1881 to 2013 and I calculated the return of 7 different strategies:

1) Buy and Hold
2) Sell our position when P/E10> 15 and buy when P/E10< 15
3) Sell our position when P/E10> 20 and buy when P/E10< 20
4) Sell our position when P/E10> 23 and buy when P/E10< 23
5) Sell our position when Monthly P/E > 15 and buy when monthly P/E10< 15
6) Sell our position when Monthly P/E > 20 and buy when monthly P/E10< 20
7) Sell our position when Monthly P/E > 23 and buy when monthly P/E10< 23

Here are the results of $10,000 invested in 1881 and with all the dividends re-invested at end of each year (I assume no fees, taxes etc)

Buy and Hold: $750,711,402 - CAGR: 8,9%
Sell our position when P/E10> 15 $2,942,434 - CAGR: 4,4%
Sell our position when P/E10> 20 $22,045,707 - CAGR: 6,0%
Sell our position when P/E10> 23 $165,773,291 - CAGR 7,6%
Sell our position when Monthly P/E > 15 $1,252,739 - CAGR 3,7%
Sell our position when Monthly P/E > 20 $244,122,209 - CAGR 8,0%
Sell our position when Monthly P/E > 23 $465,429,112 - CAGR 8,5%


At the end of the year, the dividends are used to purchase new shares at a December's price. If the holding period is less than 12months (because of PE10's value above/below the threshold), then only a proportion of yearly dividends have been used to purchase new stocks.

Example:
Dividend year X: $10
Holding period year X: From March to October (6 months)
Holding period dividend: $5
This $5 have been used to purchase new shares in October at October's price.


Here is the link to the spreadsheet, which was not meant to shared and I apologize for the poor format:
http://www.filedropper.com/pespreturns

One can critique many aspects of your methodology. However, that does not change the fact, supported by your research, that PE10-triggered market timing is pseudo scientific. It is likely to reduce, rather than enhance, returns.

heartandsoul
Posts: 82
Joined: Sat Jan 24, 2015 4:21 am

Re: The cult of PE10 - so what?

Post by heartandsoul » Sun Feb 08, 2015 5:26 am

Be careful of hindsight bias when looking for relationships between fundamental valuation measures, such as PE10, and future returns. The US economy and stock market have been an incredible success story for the last 100+ years which means that any period of turmoil, and perhaps low PE's, will have proven a good buying opportunity.
'Life is a strange thing. Just when you think you learned how to use it. It's gone' SS

User avatar
Clearly_Irrational
Posts: 3087
Joined: Thu Oct 13, 2011 3:43 pm

Re: The cult of PE10 - so what?

Post by Clearly_Irrational » Sun Feb 08, 2015 10:32 am

William Million wrote:PE10-triggered market timing is pseudo scientific.
I fail to see how. Using data to discover relationships between variables is a classic scientific pursuit. One has to be careful to be sure it isn't just curve fitting by using out of sample data but other than that I'm not sure I understand your argument.

User avatar
William Million
Posts: 547
Joined: Wed May 05, 2010 4:41 am
Location: A Deep Mountain

Re: The cult of PE10 - so what?

Post by William Million » Sun Feb 08, 2015 3:14 pm

Clearly_Irrational wrote:
William Million wrote:PE10-triggered market timing is pseudo scientific.
I fail to see how. Using data to discover relationships between variables is a classic scientific pursuit. One has to be careful to be sure it isn't just curve fitting by using out of sample data but other than that I'm not sure I understand your argument.
Selective back-testing. Switching money from stocks to bonds in years when the San Diego Padres have a winning record yields higher returns. There are many valuation metrics. Using a single one in isolation is recipe for disaster.

Post Reply