## My CAPE10 based IPS

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:24 am, edited 1 time in total.

qwertyjazz
Posts: 1132
Joined: Tue Feb 23, 2016 4:24 am

### Re: My CAPE10 based IPS

I am pretty sure about a future including death and taxes
http://freakonomics.com/2011/02/17/quot ... and-taxes/
G.E. Box "All models are wrong, but some are useful."

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

willthrill81 wrote:
Sun Sep 23, 2018 10:21 pm
Park wrote:
Sun Sep 23, 2018 10:17 pm
viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

Assume a PE10 of 10. That's an earnings yield of 10%. Assume the historical 1.5% annual growth in earnings. PE10 reflects average earnings over the last 10 years. The average time lag is 5 years. Earnings went up 1.5% each year for the those 5 years, and are 1.075 times greater than they were 5 years ago. So multiply 10% by 1.075 and you get 10.75% earnings yield. The model predicts a real return of 10.75%, or about double the historical average. Your margin of safety is large; even if you're off by 50% due to changes in accounting rules, you'll get the historical return. And historically, bear market risk is less. Risk adjusted returns have been every better.

Assume a PE10 of 40. When you do the same calculations as in the last paragraph, the predicted real return is 2.69%, or about half the historical average. Where is your margin of safety? Stocks have to do about double what the model predicts to get the historical average. And the risk of exposure to a bear market has gone up considerably. It's not return free risk, but it's headed that way.

Is an expected real return of 2.69% worth the possible risk of losing more than 50% of your investment? In other words, what expected return justifies the equity risk premium? I might/might not consider 2.69% enough, but it's getting near the edge. At a PE10 of 50, that's a 2.15% earnings yield. Only in the absence of reasonable alternatives would I want to be invested in stocks with such an earnings yield
The problem with that is that the last 27 years of U.S. stock history fly in the face of it (since 1992, U.S. stocks crossed the historic CAPE average and essentially never looked back). PE10 (i.e. CAPE) has been higher than its historic average over virtually the entire period, yet real returns over the same period have been higher than the historic average as well.

A high CAPE does not necessarily preclude good returns, nor does a low CAPE ensure good returns. There are many other factors at work.
https://www.marketwatch.com/story/this- ... 2017-12-28

I couldn't find the average CAPE over the last 27 years. But this article gives average CAPE over 15 years ending Nov 30/17, and it's 25. I'm not saying, and I doubt the OP is saying, that one should use a CAPE of 25 to time the market. The predictive value of a CAPE of 25 is such that I wouldn't include it in an investment strategy.

But over your 27 year period, CAPE has reached a max of about 44 in December 1999 and a minimum of about 13 in March 2009. At those extremes, CAPE did have considerable predictive value.

About Russia not giving very good 10-15 year returns despite a CAPE of less than 10, average CAPE for Russian market for 1997-2017 is 10.8. In 2017 Russian CAPE was 5.4. In 2007, it was 21.9. That's from the link below.

https://www.theemergingmarketsinvestor.com/2018/01/

The link below gives a graph of the Russian stock market index since 1997. With the exception of the period starting Jan 2008, it doesn't look bad. And that bad period started with a comparatively high CAPE.

Perhaps you have better data than I do. If so, show me the data that when the Russian stock market had a CAPE of less than 10, real returns over the next 10-15 years were poor.

michaeljc70
Posts: 6071
Joined: Thu Oct 15, 2015 3:53 pm

### Re: My CAPE10 based IPS

It seems rather vague. "up to" and "Modest tilt (25% or less of equities) to value, small", etc.. What determines whether the "25% or less" is 0%, 5%, 10%, 15%, etc.? What determines what goes to value and what goes to small? Seems like a lot of decisions to be made and a lot of judgement/emotion can enter in.

Valuethinker
Posts: 39254
Joined: Fri May 11, 2007 11:07 am

### Re: My CAPE10 based IPS

letsgobobby wrote:
Sun Sep 23, 2018 11:04 am
Here it is, for those who have asked (in summary form, omitting elements of the IPS which are specific to my situation or otherwise personal):

* 60/40 stocks/bonds
* Bonds 50/50 total bond/stable value until 10 year treasury > 6%
* Stocks 50/50 US/international, can go up 75% international based on relative valuations (change no more than once per year)
* Modest tilt (25% or less of equities) to value, small
* US stocks: if CAPE10 > 25 then reduce by 10%. If > 30 reduce by 20%. If > 35 reduce by 30%. If CAPE < 12 increase by 10%. If < 8 increase by 20%. If < 5 increase by 30%. Err on the side of doing nothing.

I've had this exact IPS since 2012, it fluctuated a bit before then but the CAPE element has been pretty stable since the early 2000s.

fire away...
Using a nominal value (6% interest rate) in a real calculation (PE ratio) is likely to lead you astray. You probably should use real rates (TIPS).

It's basically "the Fed Model" and if you google that, you'll find it has been pretty strongly criticized.

Put it another way, what if interest rates never get to 6%? Through most of the history of interest rates, they do not appear to have been that high (thinking say UK govt debt since 1694 as a long debt series).

Valuethinker
Posts: 39254
Joined: Fri May 11, 2007 11:07 am

### Re: My CAPE10 based IPS

Park wrote:
Wed Sep 26, 2018 12:44 am
willthrill81 wrote:
Sun Sep 23, 2018 10:21 pm
Park wrote:
Sun Sep 23, 2018 10:17 pm
viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

Assume a PE10 of 10. That's an earnings yield of 10%. Assume the historical 1.5% annual growth in earnings. PE10 reflects average earnings over the last 10 years. The average time lag is 5 years. Earnings went up 1.5% each year for the those 5 years, and are 1.075 times greater than they were 5 years ago. So multiply 10% by 1.075 and you get 10.75% earnings yield. The model predicts a real return of 10.75%, or about double the historical average. Your margin of safety is large; even if you're off by 50% due to changes in accounting rules, you'll get the historical return. And historically, bear market risk is less. Risk adjusted returns have been every better.

Assume a PE10 of 40. When you do the same calculations as in the last paragraph, the predicted real return is 2.69%, or about half the historical average. Where is your margin of safety? Stocks have to do about double what the model predicts to get the historical average. And the risk of exposure to a bear market has gone up considerably. It's not return free risk, but it's headed that way.

Is an expected real return of 2.69% worth the possible risk of losing more than 50% of your investment? In other words, what expected return justifies the equity risk premium? I might/might not consider 2.69% enough, but it's getting near the edge. At a PE10 of 50, that's a 2.15% earnings yield. Only in the absence of reasonable alternatives would I want to be invested in stocks with such an earnings yield
The problem with that is that the last 27 years of U.S. stock history fly in the face of it (since 1992, U.S. stocks crossed the historic CAPE average and essentially never looked back). PE10 (i.e. CAPE) has been higher than its historic average over virtually the entire period, yet real returns over the same period have been higher than the historic average as well.

A high CAPE does not necessarily preclude good returns, nor does a low CAPE ensure good returns. There are many other factors at work.
https://www.marketwatch.com/story/this- ... 2017-12-28

I couldn't find the average CAPE over the last 27 years. But this article gives average CAPE over 15 years ending Nov 30/17, and it's 25. I'm not saying, and I doubt the OP is saying, that one should use a CAPE of 25 to time the market. The predictive value of a CAPE of 25 is such that I wouldn't include it in an investment strategy.

But over your 27 year period, CAPE has reached a max of about 44 in December 1999 and a minimum of about 13 in March 2009. At those extremes, CAPE did have considerable predictive value.

About Russia not giving very good 10-15 year returns despite a CAPE of less than 10, average CAPE for Russian market for 1997-2017 is 10.8. In 2017 Russian CAPE was 5.4. In 2007, it was 21.9. That's from the link below.

https://www.theemergingmarketsinvestor.com/2018/01/

The link below gives a graph of the Russian stock market index since 1997. With the exception of the period starting Jan 2008, it doesn't look bad. And that bad period started with a comparatively high CAPE.

Perhaps you have better data than I do. If so, show me the data that when the Russian stock market had a CAPE of less than 10, real returns over the next 10-15 years were poor.
I should be amazed if CAPE10 works well in Russia.

Russia there is one main factor - oil prices. Oil prices are what drives the Russian market (plus semi random geopolitical events). If expectations of future oil prices are high the stocks will move to higher multiples and if the market thinks we are near a cyclical peak, the PE of the energy stocks will be very low.

(you buy deep cyclicals when their PEs are high to not meaningful, ie when they are not earning an economic return on their assets. You sell them when they are near to peak earnings. Mining, oil & gas, pulp & paper, petrochemicals, steel - all the deep cycle stocks).

The world's second largest energy company (Gazprom) by revenues and reserves is listed in Moscow (the largest is Saudi Aramco, where the IPO appears to be paused if not killed at this moment). Vladimir Putin reputedly has the Gazprom accounts delivered directly to his office for review -- Gazprom is the Russian state, and the Russian state is Gazprom.

You might be able to use CAPE10 on the Emerging Markets as a whole. My guess is you'd find them to be quite expensive due to the multiples of the Chinese internet stocks (Tencent, Alibaba & ?Naspers?)?

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

willthrill81 wrote:
Mon Sep 24, 2018 3:10 pm
3funder wrote:
Mon Sep 24, 2018 2:50 pm
HEDGEFUNDIE wrote:
Sun Sep 23, 2018 1:52 pm
Where is the evidence that CAPE10 is predictive of short-medium run stock prices?

I suppose the worst that could happen is you are underinvested in equities during a long bull run.
I definitely think CAPE 10 can be useful for getting a general sense of intermediate-term stock returns. Of course, anyone can cherry pick based on this date or that, but if you have a better approach, please share.
The average stock allocation for investors has a much better backtested track record than CAPE. It's currently predicting 10 year returns of 3.13% for the S&P 500.
The average stock allocation does have better backtested track record than CAPE.

http://www.philosophicaleconomics.com/2 ... t-returns/

As has been mentioned, valuations aren't the only reason you get bear markets. I wouldn't be surprised if average stock allocation is better than valuations, when it comes to predicting returns not related to valuation. But I don't think it changes the fact that extremes of valuation have predictive power.

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:24 am, edited 1 time in total.

Valuethinker
Posts: 39254
Joined: Fri May 11, 2007 11:07 am

### Re: My CAPE10 based IPS

Park wrote:
Wed Sep 26, 2018 8:37 am
willthrill81 wrote:
Mon Sep 24, 2018 3:10 pm
3funder wrote:
Mon Sep 24, 2018 2:50 pm
HEDGEFUNDIE wrote:
Sun Sep 23, 2018 1:52 pm
Where is the evidence that CAPE10 is predictive of short-medium run stock prices?

I suppose the worst that could happen is you are underinvested in equities during a long bull run.
I definitely think CAPE 10 can be useful for getting a general sense of intermediate-term stock returns. Of course, anyone can cherry pick based on this date or that, but if you have a better approach, please share.
The average stock allocation for investors has a much better backtested track record than CAPE. It's currently predicting 10 year returns of 3.13% for the S&P 500.
The average stock allocation does have better backtested track record than CAPE.

http://www.philosophicaleconomics.com/2 ... t-returns/

As has been mentioned, valuations aren't the only reason you get bear markets. I wouldn't be surprised if average stock allocation is better than valuations, when it comes to predicting returns not related to valuation. But I don't think it changes the fact that extremes of valuation have predictive power.
That is really interesting, thank you.

I am wondering though what it shows?

Does it simply show that as equity returns improve, investors *drift* upwards in their equity allocations. i.e. that there is some fundamental underlying target equity allocation but as investors we let it drift with performance? I.e. that we are pro cyclical.

Can we use this data to forecast future returns?

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

Valuethinker wrote:
Wed Sep 26, 2018 8:34 am
Park wrote:
Wed Sep 26, 2018 12:44 am
willthrill81 wrote:
Sun Sep 23, 2018 10:21 pm
Park wrote:
Sun Sep 23, 2018 10:17 pm
viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

Assume a PE10 of 10. That's an earnings yield of 10%. Assume the historical 1.5% annual growth in earnings. PE10 reflects average earnings over the last 10 years. The average time lag is 5 years. Earnings went up 1.5% each year for the those 5 years, and are 1.075 times greater than they were 5 years ago. So multiply 10% by 1.075 and you get 10.75% earnings yield. The model predicts a real return of 10.75%, or about double the historical average. Your margin of safety is large; even if you're off by 50% due to changes in accounting rules, you'll get the historical return. And historically, bear market risk is less. Risk adjusted returns have been every better.

Assume a PE10 of 40. When you do the same calculations as in the last paragraph, the predicted real return is 2.69%, or about half the historical average. Where is your margin of safety? Stocks have to do about double what the model predicts to get the historical average. And the risk of exposure to a bear market has gone up considerably. It's not return free risk, but it's headed that way.

Is an expected real return of 2.69% worth the possible risk of losing more than 50% of your investment? In other words, what expected return justifies the equity risk premium? I might/might not consider 2.69% enough, but it's getting near the edge. At a PE10 of 50, that's a 2.15% earnings yield. Only in the absence of reasonable alternatives would I want to be invested in stocks with such an earnings yield
The problem with that is that the last 27 years of U.S. stock history fly in the face of it (since 1992, U.S. stocks crossed the historic CAPE average and essentially never looked back). PE10 (i.e. CAPE) has been higher than its historic average over virtually the entire period, yet real returns over the same period have been higher than the historic average as well.

A high CAPE does not necessarily preclude good returns, nor does a low CAPE ensure good returns. There are many other factors at work.
https://www.marketwatch.com/story/this- ... 2017-12-28

I couldn't find the average CAPE over the last 27 years. But this article gives average CAPE over 15 years ending Nov 30/17, and it's 25. I'm not saying, and I doubt the OP is saying, that one should use a CAPE of 25 to time the market. The predictive value of a CAPE of 25 is such that I wouldn't include it in an investment strategy.

But over your 27 year period, CAPE has reached a max of about 44 in December 1999 and a minimum of about 13 in March 2009. At those extremes, CAPE did have considerable predictive value.

About Russia not giving very good 10-15 year returns despite a CAPE of less than 10, average CAPE for Russian market for 1997-2017 is 10.8. In 2017 Russian CAPE was 5.4. In 2007, it was 21.9. That's from the link below.

https://www.theemergingmarketsinvestor.com/2018/01/

The link below gives a graph of the Russian stock market index since 1997. With the exception of the period starting Jan 2008, it doesn't look bad. And that bad period started with a comparatively high CAPE.

Perhaps you have better data than I do. If so, show me the data that when the Russian stock market had a CAPE of less than 10, real returns over the next 10-15 years were poor.
I should be amazed if CAPE10 works well in Russia.

Russia there is one main factor - oil prices. Oil prices are what drives the Russian market (plus semi random geopolitical events). If expectations of future oil prices are high the stocks will move to higher multiples and if the market thinks we are near a cyclical peak, the PE of the energy stocks will be very low.

(you buy deep cyclicals when their PEs are high to not meaningful, ie when they are not earning an economic return on their assets. You sell them when they are near to peak earnings. Mining, oil & gas, pulp & paper, petrochemicals, steel - all the deep cycle stocks).

The world's second largest energy company (Gazprom) by revenues and reserves is listed in Moscow (the largest is Saudi Aramco, where the IPO appears to be paused if not killed at this moment). Vladimir Putin reputedly has the Gazprom accounts delivered directly to his office for review -- Gazprom is the Russian state, and the Russian state is Gazprom.

You might be able to use CAPE10 on the Emerging Markets as a whole. My guess is you'd find them to be quite expensive due to the multiples of the Chinese internet stocks (Tencent, Alibaba & ?Naspers?)?
It might be true that CAPE10 doesn't works well in Russia. But it wouldn't change my thinking that extremes of valuation have predictive ability. That predictive ability is based on reversion to mean. Reversion to mean is a noisy signal. It won't work everywhere. If it did, it would stop working, because it would be taken into account in pricing.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

Park wrote:
Wed Sep 26, 2018 12:44 am
willthrill81 wrote:
Sun Sep 23, 2018 10:21 pm
The problem with that is that the last 27 years of U.S. stock history fly in the face of it (since 1992, U.S. stocks crossed the historic CAPE average and essentially never looked back). PE10 (i.e. CAPE) has been higher than its historic average over virtually the entire period, yet real returns over the same period have been higher than the historic average as well.
I couldn't find the average CAPE over the last 27 years.
Here is an easy way to do it, go to http://multpl.com, click on Shiller PE, click on Table, and with a simple copy and paste in a spreadsheet, you can compute all the averages you want...

The average between Jan 1991 and Dec 2017 is a little over 25. I don't see that this shows very much though, 27 years isn't such a long period of time in the midst of stock market vagaries.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

A few posts ago, I promised to run a solid backtest of an algorithm somewhat close to what was proposed in the OP.

Given the lack of International data, I solely focused on US history (monthly values), using data from Prof. Shiller for S&P 500 CAPE, for stocks and bonds returns until the early 1970s. Then I used well-known indices for S&P 500 and Agg Bonds returns until Jan 2018. Note that this makes the bonds history a tad discontinuous as I used short-term interest rates to simulate short-term bond returns before the 70s, and then Agg. Bonds is more centered on IT bonds. This shouldn't have much impact on the test itself though.

First off, let's look at CAPE data WITHOUT any foresight (ex ante). The first chart shows the monthly CAPE values, compared to the rolling average of the past 50 years. There are non negligible variations over time, the most commonly discussed being the upward trend from the past few decades. The second chart breaks down the historical data known from 1881 until a certain date in percentiles (e.g. the 50% percentile is the median value -which is a tad different from the average-; the 2% percentile is the upper cap of the 2% lowest values; the 98% percentile is the upper cap for the 98% lowest values, which is the same thing as the lower cap for the 2% highest values). As you can see, the great depression (30s) and the Internet crisis (2002) changed things in a rather significant manner. History is great, but it doesn't always tell us what the future will be made of...

Note that I chose to illustrate the 90%, 95% and 98% percentiles (and their counterparts on the low side), as those seem close enough to the caps chosen by the OP to drive his changes of Asset Allocation.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

Using the historical data described in the previous post, I looked at various retirement cycles of 30 years starting from a given year with \$1M, and withdrawing \$40k at the beginning of each year, inflation-adjusted. A first strategy keeps the good old fixed 60/40 asset allocation. The second strategy adjusts the asset allocation in a tactical manner based on CAPE and the percentile caps known so far (no cheating, without foresight). I tried to do something close enough to the OP, adding 10%, 20%, 30% to the stocks allocation for the lowest percentiles (10%, 5%, 2%), and subtracting 10%, 20%, 30% for the highest percentiles (90%, 95%, 98%). Here is the portfolio balance trajectory when starting in 1925, 1965, 1985.

The red line (and rightmost vertical axis) illustrates the stocks % of the CAPE-based strategy over time.

As suggested by the OP, it would be nice to add some hysteresis refinement in the algorithm, but I don't think this would have changed the outcome very much (besides transaction costs, which I didn't model anyway).

Overall, this isn't quite satisfying, to say the least... There are multiple problems. First, the percentile caps can't foresee the future and didn't quite account for the amplitude of the Internet crisis, as a case in point; it also may not adjust fast enough to possible secular changes like might (?!) be happening nowadays. Next, the oil crisis in the 70s was NOT a valuation crisis, nothing would have triggered an AA change ahead of the crash. But more importantly, CAPE has decent mid/long-term predictive ability, but no short-term predictive ability. It just does not tell you WHEN a crash is coming. So the algorithm kind of over-reacts to the impeding 1929 and 2002 crisis, the portfolio is rebalanced way too early, significant subsequent stock gains are missed, and this impacts the 30-years trajectory in quite a significant way.

Interestingly enough, when I ran a similar test a couple of years ago using annual CAPE/stocks/bonds returns, the results were better. I think this is because making annual decisions would have mitigated the over-reaction. This was still not convincing, mind you...

Tonight, I'll post a chart with a more systematic test, showing all possible starting years. You can guess the outcome though.

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

Siamond, thanks for all your hard work.

One conclusion that I draw is that there is almost always an equity risk premium. So I'm not entirely surprised by the results starting in 1925 and 1985, as they're associated with overall lower stock exposure. The results starting in 1965 do surprise me somewhat, as the market timing strategy resulted in increased stock allocation when valuations were low. However, that was in the late 70s/early 80s. How much equity risk premium was there?

Your results reinforce the conclusion that market timing between stocks and bonds is difficult.

However, that is not my plan, when it comes to using valuation based market timing. I'm more interested in using such a strategy between stock subasset classes (American, developed, emerging markets etc.). I've never seen a study of that, perhaps because it is difficult to do.

Once again, thanks for your work.

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:29 am, edited 1 time in total.

dcabler
Posts: 1195
Joined: Wed Feb 19, 2014 11:30 am

### Re: My CAPE10 based IPS

nedsaid wrote:
Mon Sep 24, 2018 10:26 pm
siamond wrote:
Mon Sep 24, 2018 9:42 am
About Swedroe's math, the pre-GAAP earnings allowed companies to unduly defer losses, not to permanently hide them. A permanent downward adjustment to CAPE is... weird. He also picked one varying factor in the middle of 10,000 other historical changes in corporate life, and assigned an arbitrary weight on it. Which makes little sense, even if the GAAP thing was indeed a pretty big one-time thing. In the same paper, he also double-counted the effect of the payout ratio historical shift, and when I pointed that out, he had no meaningful answer. This whole paper was basically gut feeling that *something* needed to be done to adjust CAPE calibration. The gut feeling might not be wrong, mind you, but that is no basis for any algebraic decision logic.

Personally, I see things in a much simpler manner. Earnings are split between dividends (impacts total return right away), stock buybacks (impacts price right away), and the rest of what is reinvested should drive future growth, hence price. This remains true irrespective of how earnings are computed. If accountants play games, well, the total returns should follow accordingly, up and down (Enron as an extreme case in point!). Therefore, I see no reason to adjust CAPE. What I would question though is the sheer existence of a constant mean to return to... And THAT is the \$1M question...
Yes, Siamond, it seems that the market reverts to an ever increasing mean. Another way of saying it is that the goal posts keep moving.
Let's call it a variable constant!

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

letsgobobby wrote:
Wed Sep 26, 2018 2:03 pm
I appreciate your work on this. I agree, as I hinted at upthread, that there is a weakness to the model and you have hit on it: namely, that it lowers stock exposure "too early" in cases of extremely high valuations, which then puts the investor behind the buy and hold strategy.

That said, remember my goal is not to maximize gains, but to minimize risk, ie, prevent the generational catastrophes to my portfolio that would result from 50+% declines. Therefore would you be kind enough to compare a 60/40 CAPE model to a 50/50 buy and hold model? I think that is a better comparator.
Actually, let's take a closer look at risk before running other asset allocation comparisons. The following chart shows the portfolio end balance for all cycles of 30 years starting in Jan-1926, Feb-1926, Mar-1926, etc. Note that I used a 3.5% withdrawal in this case, contrary to the previous test, to avoid the few portfolio failures scenarios which were making my 2nd chart below hard to read!

The pattern is impressively clear, the Fixed AA wins in each and every case (I double-checked!). Note that if we define risk as very low portfolio balance (and corresponding panic attacks!), then the Fixed AA is undoubtedly superior. Some of the CAPE AA outcomes are quite iffy (and remember, I was being super conservative with a 3.5% withdrawal rate). Personally, I find such measure of risk more palatable than the following discussion, but that's just me.

Now let's focus on other definitions of risk. Let's equate risk to standard-deviation of the portfolio balance (inflation-adjusted). Same thing, all cycles of 30 years.

The result surprised me. Then I looked more closely. When the cycle of 30 years includes either the Great Depression (1929/37) or the Internet crisis (2002), then yeah, the CAPE AA appears less risky (although results are actually mixed for the great depression). But when the cycle of 30 years is more centered on the oil crisis (1973), the outcome is reversed. Which actually makes sense, given that the oil crisis was NOT a valuation crisis.

I ran a quick test using other (better imho) measures of risk than std-deviation, meaning max drawdown and Ulcer index. Results were very similar to the std-deviation chart.

OP, what do you think? The case isn't terribly convincing, isn't it?

Then when you add all the doubts and uncertainties due to the moving goal posts (as nedsaid put it), then I can't see that it's worth it. Oh, and just to be clear, when I started my own study a couple of years ago (using annual returns), I was VERY open-minded about such ideas, as this seemed remarkably intuitive. Then I grew more and more skeptical as I ran more numbers, and the moving goal posts thing put the nail on the coffin. I am happy to have run the monthly experiment though, as I've always wondered if the top-level conclusion would be similar. I believe it is.

nedsaid
Posts: 12976
Joined: Fri Nov 23, 2012 12:33 pm

### Re: My CAPE10 based IPS

letsgobobby wrote:
Tue Sep 25, 2018 8:32 pm
nedsaid wrote:
Mon Sep 24, 2018 10:26 pm

Yes, Siamond, it seems that the market reverts to an ever increasing mean. Another way of saying it is that the goal posts keep moving have moved for the last twenty years.
I've fixed your quote for you. No thanks necessary.

If you, or anyone else here, knows the future, please let me know.
Below is what I actually said. Didn't see any note that I had edited it.
Yes, Siamond, it seems that the market reverts to an ever increasing mean. Another way of saying it is that the goal posts keep moving.
A fool and his money are good for business.

nedsaid
Posts: 12976
Joined: Fri Nov 23, 2012 12:33 pm

### Re: My CAPE10 based IPS

dcabler wrote:
Wed Sep 26, 2018 2:16 pm
nedsaid wrote:
Mon Sep 24, 2018 10:26 pm
siamond wrote:
Mon Sep 24, 2018 9:42 am
About Swedroe's math, the pre-GAAP earnings allowed companies to unduly defer losses, not to permanently hide them. A permanent downward adjustment to CAPE is... weird. He also picked one varying factor in the middle of 10,000 other historical changes in corporate life, and assigned an arbitrary weight on it. Which makes little sense, even if the GAAP thing was indeed a pretty big one-time thing. In the same paper, he also double-counted the effect of the payout ratio historical shift, and when I pointed that out, he had no meaningful answer. This whole paper was basically gut feeling that *something* needed to be done to adjust CAPE calibration. The gut feeling might not be wrong, mind you, but that is no basis for any algebraic decision logic.

Personally, I see things in a much simpler manner. Earnings are split between dividends (impacts total return right away), stock buybacks (impacts price right away), and the rest of what is reinvested should drive future growth, hence price. This remains true irrespective of how earnings are computed. If accountants play games, well, the total returns should follow accordingly, up and down (Enron as an extreme case in point!). Therefore, I see no reason to adjust CAPE. What I would question though is the sheer existence of a constant mean to return to... And THAT is the \$1M question...
Yes, Siamond, it seems that the market reverts to an ever increasing mean. Another way of saying it is that the goal posts keep moving.
Let's call it a variable constant!
Market P/E ratios have been creeping up over the years. This has been discussed a lot on this forum. We also talk a lot about mean reversion. The question is this, what mean is the market reverting to?
A fool and his money are good for business.

CraigTester
Posts: 115
Joined: Wed Aug 08, 2018 6:34 am

### Re: My CAPE10 based IPS

siamond wrote:
Wed Sep 26, 2018 11:37 pm
letsgobobby wrote:
Wed Sep 26, 2018 2:03 pm
I appreciate your work on this. I agree, as I hinted at upthread, that there is a weakness to the model and you have hit on it: namely, that it lowers stock exposure "too early" in cases of extremely high valuations, which then puts the investor behind the buy and hold strategy.

That said, remember my goal is not to maximize gains, but to minimize risk, ie, prevent the generational catastrophes to my portfolio that would result from 50+% declines. Therefore would you be kind enough to compare a 60/40 CAPE model to a 50/50 buy and hold model? I think that is a better comparator.
Actually, let's take a closer look at risk before running other asset allocation comparisons. The following chart shows the portfolio end balance for all cycles of 30 years starting in Jan-1926, Feb-1926, Mar-1926, etc. Note that I used a 3.5% withdrawal in this case, contrary to the previous test, to avoid the few portfolio failures scenarios which were making my 2nd chart below hard to read!

The pattern is impressively clear, the Fixed AA wins in each and every case (I double-checked!). Note that if we define risk as very low portfolio balance (and corresponding panic attacks!), then the Fixed AA is undoubtedly superior. Some of the CAPE AA outcomes are quite iffy (and remember, I was being super conservative with a 3.5% withdrawal rate). Personally, I find such measure of risk more palatable than the following discussion, but that's just me.

Now let's focus on other definitions of risk. Let's equate risk to standard-deviation of the portfolio balance (inflation-adjusted). Same thing, all cycles of 30 years.

The result surprised me. Then I looked more closely. When the cycle of 30 years includes either the Great Depression (1929/37) or the Internet crisis (2002), then yeah, the CAPE AA appears less risky (although results are actually mixed for the great depression). But when the cycle of 30 years is more centered on the oil crisis (1973), the outcome is reversed. Which actually makes sense, given that the oil crisis was NOT a valuation crisis.

I ran a quick test using other (better imho) measures of risk than std-deviation, meaning max drawdown and Ulcer index. Results were very similar to the std-deviation chart.

OP, what do you think? The case isn't terribly convincing, isn't it?

Then when you add all the doubts and uncertainties due to the moving goal posts (as nedsaid put it), then I can't see that it's worth it. Oh, and just to be clear, when I started my own study a couple of years ago (using annual returns), I was VERY open-minded about such ideas, as this seemed remarkably intuitive. Then I grew more and more skeptical as I ran more numbers, and the moving goal posts thing put the nail on the coffin. I am happy to have run the monthly experiment though, as I've always wondered if the top-level conclusion would be similar. I believe it is.
Nice work on this Siamond.

I’m actually surprised how correlated the results appear using this simple algorithm.

It suggests to me that the OP might mitigate the misery of fully participating in the next 50% drawdown - and still wind up with a similar final outcome. If that’s his real goal, he’s probably encouraged.

Further, it also suggests with some tweaking, one might even stumble on a set of rules that perform even better, while avoiding the severe drawdowns.

PS. My reading of the OPs definition of “risk”, is simply avoiding the next +50% drawdown. Probably less concerned with interim volatility.

It would be interesting to set your objective function to “maintain parity to buy-n-hold final outcome while avoiding excessive drawdowns”. You could then tweak your allocation thresholds to find the optimal

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

You mention appropriately that the oil crisis, when CAPE was low, was not a valuation crisis. Would you disagree that 1929 and 1999, when CAPE was high, were valuation crises?

My impression is that the stock market has greater trouble dealing with overvaluation than undervaluation. When stocks are underpriced, you buy. When stocks are overpriced, you short. But shorting is not the opposite of buying. Shorting is the opposite of leveraged long, and has more risk than buying. Also, stock markets tend to go up with time. When you short, you've got a headwind.

My conclusion is that valuation based market timing, may work better when stocks are overpriced than when they are underpriced. When stocks are cheap by valuation metrics, they may be more rationally priced than when they are expensive.

Siamond, thanks for your work. However, your work is based on relative valuation and not absolute valuation.

See my previous posts in this thread for more detail on what is written below.

See Fig. 6 from above. When CAPE greater than 50, returns have been negative on average. If you exclude the Danish and Swedish stock markets, returns have been modest at best, when CAPE greater than 40. For a well diversified investor, I question how relevant the data from the Danish and Swedish stock markets is. For them, the S&P500 data is more appropriate. When S&P500 CAPE is more than 40, there hasn't been a positive return.

Assume a PE10 of 40. When you calculate expected return based on PE10, the predicted real return is 2.69%. The risk of exposure to a bear market has gone up considerably. It's not return free risk, but it's headed that way.

Is an expected real return of 2.69% worth the possible risk of losing more than 50% of your investment? In other words, what expected return justifies the equity risk premium? I might/might not consider 2.69% enough, but it's getting near the edge. At a PE10 of 50, that's a 2.15% earnings yield. Only in the absence of reasonable alternatives would I want to be invested in stocks with such an earnings yield.

So I haven't given up on valuation based market timing. I don't think I'll use it when CAPE etc is low. But when CAPE is high, I will seriously consider using it.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

CraigTester wrote:
Thu Sep 27, 2018 8:09 am
It suggests to me that the OP might mitigate the misery of fully participating in the next 50% drawdown - and still wind up with a similar final outcome. If that’s his real goal, he’s probably encouraged.

Further, it also suggests with some tweaking, one might even stumble on a set of rules that perform even better, while avoiding the severe drawdowns.

PS. My reading of the OPs definition of “risk”, is simply avoiding the next +50% drawdown. Probably less concerned with interim volatility.
When it comes to portfolio balance, it seems to me that absolute values (amplitude) are much more important than correlation (directionality). But focusing on your point about drawdowns, yes, this would be nice, except that... it didn't work in any meaningful manner. For the oil crisis, the algorithm didn't help whatsoever with the drawdown (also note that bonds also cratered in the following years, no safe haven for sure). For the Internet crisis, it helped a bit (which the OP probably benefited from). For the Financial crisis, it didn't help. For the Great Depression, it didn't help.

We saw it in the 1925/1965/1985 scenarios I illustrated earlier (visually, this might not be 100% obvious, but I checked the corresponding values). The test I ran with the Ulcer Index (which is a combined measure of depth and width of drawdowns, small and large) showed a pattern very similar to the standard deviation chart.

To make things fully palatable to you guys, I switched the risk metric to max drawdowns, and sheesh, this isn't exactly encouraging for the OP, I think. Note that I don't see that a difference of a few points would bring any comfort when facing a steep 30% to 60% drawdown...

As to tweaking the algorithm (eg. thresholds, hysteresis, etc), I'll play around later today. I suspect the OP's thresholds are too high if the point is to minimize drawdowns. But... I am not terribly hopeful. Plus it might quickly become an exercise in data mining as we don't have that many data points when it comes to big drawdowns (we essentially had three major stock crises -- and also two major bonds crises).

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

Park wrote:
Thu Sep 27, 2018 9:00 am
You mention appropriately that the oil crisis, when CAPE was low, was not a valuation crisis. Would you disagree that 1929 and 1999, when CAPE was high, were valuation crises?
Totally agreed, both crises were clearly due to speculative mania. The recent financial crisis is less clear, but clearly the subprime market went amok with pricing. The Japanese asset bubble crisis (late 80s) was clearly a speculative crisis. Still, the biggest threat to a retiree was undoubtedly the oil crisis (actually retiring in the mid-60s), this is where we find the worst SWRs, and this is true for most countries, not only the US.
Park wrote:
Thu Sep 27, 2018 9:00 am
Siamond, thanks for your work. However, your work is based on relative valuation and not absolute valuation. [...]
Welcome. Yes, I am well aware of Star Capital's excellent work, and the (relatively short) history of International CAPEs. Trouble is I don't know how to calibrate absolute valuation without falling in the trap of hindsight and ex post reasonings. Setting aside the Japanese asset bubble, by the late 80s, nobody had a clue that CAPE could go as high as it would a decade later, which totally recalibrated our perception of 'absolute' valuations. And if one thinks that the Internet crisis was close to the worst possible speculative crisis, then do take a look at the Japanese asset bubble. Never say never in this financial world...
Park wrote:
Thu Sep 27, 2018 9:00 am
So I haven't given up on valuation based market timing. I don't think I'll use it when CAPE etc is low. But when CAPE is high, I will seriously consider using it.
I agree with you that acting on high CAPE seems more attractive than acting on low CAPE. And yes, it is so hard to give up on those TAA ideas, because in hindsight, things seem so clear and intuitive. I've been there, done it, took me a while to give up. But please think hard to the charts I posted, and try to force yourself to ignore hindsight. And the fundamental contradiction of using a long-term (blunt) indicator for short-term decisions. And the moving goal posts. It just doesn't add up. All this being said, if CAPE goes over 40, I'll probably have another round of debate with myself, gut feeling against analytical thinking...

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:31 am, edited 1 time in total.

willthrill81
Posts: 15165
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

### Re: My CAPE10 based IPS

siamond wrote:
Wed Sep 26, 2018 11:37 pm
letsgobobby wrote:
Wed Sep 26, 2018 2:03 pm
I appreciate your work on this. I agree, as I hinted at upthread, that there is a weakness to the model and you have hit on it: namely, that it lowers stock exposure "too early" in cases of extremely high valuations, which then puts the investor behind the buy and hold strategy.

That said, remember my goal is not to maximize gains, but to minimize risk, ie, prevent the generational catastrophes to my portfolio that would result from 50+% declines. Therefore would you be kind enough to compare a 60/40 CAPE model to a 50/50 buy and hold model? I think that is a better comparator.
Actually, let's take a closer look at risk before running other asset allocation comparisons. The following chart shows the portfolio end balance for all cycles of 30 years starting in Jan-1926, Feb-1926, Mar-1926, etc. Note that I used a 3.5% withdrawal in this case, contrary to the previous test, to avoid the few portfolio failures scenarios which were making my 2nd chart below hard to read!

The pattern is impressively clear, the Fixed AA wins in each and every case (I double-checked!). Note that if we define risk as very low portfolio balance (and corresponding panic attacks!), then the Fixed AA is undoubtedly superior. Some of the CAPE AA outcomes are quite iffy (and remember, I was being super conservative with a 3.5% withdrawal rate). Personally, I find such measure of risk more palatable than the following discussion, but that's just me.

Now let's focus on other definitions of risk. Let's equate risk to standard-deviation of the portfolio balance (inflation-adjusted). Same thing, all cycles of 30 years.

The result surprised me. Then I looked more closely. When the cycle of 30 years includes either the Great Depression (1929/37) or the Internet crisis (2002), then yeah, the CAPE AA appears less risky (although results are actually mixed for the great depression). But when the cycle of 30 years is more centered on the oil crisis (1973), the outcome is reversed. Which actually makes sense, given that the oil crisis was NOT a valuation crisis.

I ran a quick test using other (better imho) measures of risk than std-deviation, meaning max drawdown and Ulcer index. Results were very similar to the std-deviation chart.

OP, what do you think? The case isn't terribly convincing, isn't it?

Then when you add all the doubts and uncertainties due to the moving goal posts (as nedsaid put it), then I can't see that it's worth it. Oh, and just to be clear, when I started my own study a couple of years ago (using annual returns), I was VERY open-minded about such ideas, as this seemed remarkably intuitive. Then I grew more and more skeptical as I ran more numbers, and the moving goal posts thing put the nail on the coffin. I am happy to have run the monthly experiment though, as I've always wondered if the top-level conclusion would be similar. I believe it is.
Thanks Siamond for continuing to produce excellent work.

This is, in my view, another nail in the coffin for the idea that valuation driven investing just doesn't work well over time.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:31 am, edited 1 time in total.

michaelsieg
Posts: 582
Joined: Mon Jan 07, 2013 11:02 am

### Re: My CAPE10 based IPS

It is interesting, that in portfoliovisualizer, the Schiler PE market timing strategy looks better than the static 60/40, maybe it is due to the starting date, it goes back to 1985:

https://www.portfoliovisualizer.com/tes ... sisResults

I hope the link works, otherwise, if not, it can be found on the portfolovisualizer webpage, in the "timing strategies, market valuation" section.

Letsgobobby, I am still not sure how your portfolio is allocated today based on your IPS, could you please tell us how much international you hold, and what your stock vs. bond allocation is? Sorry I am confused that your IPS seems to dictate that changes need to be made, but you mentioned that you have not made any adjustments.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

michaelsieg wrote:
Thu Sep 27, 2018 9:58 pm
It is interesting, that in portfoliovisualizer, the Schiler PE market timing strategy looks better than the static 60/40, maybe it is due to the starting date, it goes back to 1985
Yes, if the Internet crisis is included (2002), then a CAPE strategy appears to have reduced volatility. Otherwise, not so much.

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

deleted
Last edited by letsgobobby on Fri Apr 26, 2019 9:31 am, edited 1 time in total.

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

siamond wrote:
Thu Sep 27, 2018 9:21 am
And the fundamental contradiction of using a long-term (blunt) indicator for short-term decisions.
CAPE is considered to have its optimum predictive value at 10-15 years. That means that if you use it to market time, you should stick with that decision for at least 10 years, and not change it on a yearly basis.

The above may not be completely true. See Fig 2 from the link below:

https://www.etf.com/sections/features-a ... nopaging=1

"In Figure 2 we compare the starting CAPE ratio with the S&P 500 Index return at each 1-, 5-, 10-, and 20-year horizon beginning in 1881 through October 2017"

As you go from 20 to 10 to 5 to 1 year, the regression slope becomes progressively steeper. As the time horizon shortens, the more sensitive returns are to CAPE. However, as the time horizon shortens, the dispersion of returns becomes greater. Factors other than CAPE play a relatively greater role in returns as the time horizon shortens. At a time horizon of 10-15 years, returns are less sensitive to CAPE than at 1 year. However, returns are even less sensitive to other factors. This means the maximal signal to noise ratio for CAPE ends up being 10-15 years.

So the relative importance of CAPE versus other factors plays an important role in the predictive ability of CAPE.

The above analysis assumes that the relative importance of CAPE versus other factors only depends on time horizon. But what if it also depends on the value of CAPE itself. In Fig 2, take at look at the dispersion of returns with a 5 year, 10 year and 20 year horizon. When CAPE is extremely high, the dispersion of returns is less than when CAPE is not extremely high.

I'm reasonably confident that a high CAPE ratio has predictive value at 5 years. Now I don't think it has predictive value at 1 year. Does it have predictive value at 2-3 years? I don't know, although I wish I did.

It's interesting to note that William Bernstein advocates rebalancing every 2-3 years, to allow reversion to mean to occur. If you rebalance more frequently than once a year, you're fighting momentum.

So If you're going to engage in valuation based market timing, wait a minimum of 3 years between rebalancing.

CraigTester
Posts: 115
Joined: Wed Aug 08, 2018 6:34 am

### Re: My CAPE10 based IPS

Park wrote:
Fri Sep 28, 2018 1:29 am
siamond wrote:
Thu Sep 27, 2018 9:21 am
And the fundamental contradiction of using a long-term (blunt) indicator for short-term decisions.
CAPE is considered to have its optimum predictive value at 10-15 years. That means that if you use it to market time, you should stick with that decision for at least 10 years, and not change it on a yearly basis.

The above may not be completely true. See Fig 2 from the link below:

https://www.etf.com/sections/features-a ... nopaging=1

"In Figure 2 we compare the starting CAPE ratio with the S&P 500 Index return at each 1-, 5-, 10-, and 20-year horizon beginning in 1881 through October 2017"

As you go from 20 to 10 to 5 to 1 year, the regression slope becomes progressively steeper. As the time horizon shortens, the more sensitive returns are to CAPE. However, as the time horizon shortens, the dispersion of returns becomes greater. Factors other than CAPE play a relatively greater role in returns as the time horizon shortens. At a time horizon of 10-15 years, returns are less sensitive to CAPE than at 1 year. However, returns are even less sensitive to other factors. This means the maximal signal to noise ratio for CAPE ends up being 10-15 years.

So the relative importance of CAPE versus other factors plays an important role in the predictive ability of CAPE.

The above analysis assumes that the relative importance of CAPE versus other factors only depends on time horizon. But what if it also depends on the value of CAPE itself. In Fig 2, take at look at the dispersion of returns with a 5 year, 10 year and 20 year horizon. When CAPE is extremely high, the dispersion of returns is less than when CAPE is not extremely high.

I'm reasonably confident that a high CAPE ratio has predictive value at 5 years. Now I don't think it has predictive value at 1 year. Does it have predictive value at 2-3 years? I don't know, although I wish I did.

It's interesting to note that William Bernstein advocates rebalancing every 2-3 years, to allow reversion to mean to occur. If you rebalance more frequently than once a year, you're fighting momentum.

So If you're going to engage in valuation based market timing, wait a minimum of 3 years between rebalancing.

It does nice job of corralling much of the to-and-fro on this "religious" topic.

When its all said and done, despite how much we may not like the answer, if you buy anything when it is historically expensive (houses, tulip-bulbs, stocks, et al) you will one day regret the price you paid. But until that day comes, as a species, we have a remarkable ability to rationalize just about anything.

P.S. William Bengen (father of the 4% rule) did a bunch of research on optimal rebalancing frequency. He concluded that 8 years is best for maximizing long-term survivability of a retirement account.

I think the "trick" to using any type of valuation tool for market timing is to understand that long-term signals do not satisfy our hunger for instant-gratification. Market timing is only mastered by those who are extremely patient (e.g. Warren Buffet, who is currently sitting on over \$100M in cash, has a history of exploding on the scene with a fat war chest AFTER everyone else got talked into premature investing).

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

Park wrote:
Fri Sep 28, 2018 1:29 am
siamond wrote:
Thu Sep 27, 2018 9:21 am
And the fundamental contradiction of using a long-term (blunt) indicator for short-term decisions.
CAPE is considered to have its optimum predictive value at 10-15 years. That means that if you use it to market time, you should stick with that decision for at least 10 years, and not change it on a yearly basis.

The above may not be completely true. See Fig 2 from the link below:

https://www.etf.com/sections/features-a ... nopaging=1

"In Figure 2 we compare the starting CAPE ratio with the S&P 500 Index return at each 1-, 5-, 10-, and 20-year horizon beginning in 1881 through October 2017"

As you go from 20 to 10 to 5 to 1 year, the regression slope becomes progressively steeper. As the time horizon shortens, the more sensitive returns are to CAPE. However, as the time horizon shortens, the dispersion of returns becomes greater. Factors other than CAPE play a relatively greater role in returns as the time horizon shortens. At a time horizon of 10-15 years, returns are less sensitive to CAPE than at 1 year. However, returns are even less sensitive to other factors. This means the maximal signal to noise ratio for CAPE ends up being 10-15 years.

So the relative importance of CAPE versus other factors plays an important role in the predictive ability of CAPE.

The above analysis assumes that the relative importance of CAPE versus other factors only depends on time horizon. But what if it also depends on the value of CAPE itself. In Fig 2, take at look at the dispersion of returns with a 5 year, 10 year and 20 year horizon. When CAPE is extremely high, the dispersion of returns is less than when CAPE is not extremely high.

I'm reasonably confident that a high CAPE ratio has predictive value at 5 years. Now I don't think it has predictive value at 1 year. Does it have predictive value at 2-3 years? I don't know, although I wish I did.

It's interesting to note that William Bernstein advocates rebalancing every 2-3 years, to allow reversion to mean to occur. If you rebalance more frequently than once a year, you're fighting momentum.

So If you're going to engage in valuation based market timing, wait a minimum of 3 years between rebalancing.
I'd like to revise the above. CAPE reached a peak of about 44 in December 1999. By October 2002, it was about 22. In 34 months, it had been cut in half.

http://www.multpl.com/shiller-pe/table?f=m

The Japanese stock market reached a peak of about 90 at the end of 1989. It was about 70 more months, until it got to less than 40. From the peak of 1989, it was about 260 more months until it got to less than 30.

The following is my thinking about valuation based market timing. If CAPE is above 40, time the market, and then be prepared to wait. It may be longer than 10-15 years (Japanese stock market), but it may also be shorter (US stock market).

As to when to get back into the market, I'll have to think about that.

watchnerd
Posts: 2669
Joined: Sat Mar 03, 2007 11:18 am
Location: Seattle, WA, USA

### Re: My CAPE10 based IPS

nedsaid wrote:
Sun Sep 23, 2018 3:37 pm

It does drive me a bit crazy when people try to make these tactical documents. I think this goes against the whole reason for writing an IPS. It is okay to say something like this, "When markets get to valuation and sentiment extremes, I will consider taking perhaps 15% to 20% off the top of my stock investments at market highs and will consider adding 15% to 20% to my stocks at market lows."
Seems logical, but how do you know when you hit a market high?

Basing it on indicators like "Dow now at highest ever" would be one way, but those highs are often sustained for long periods, or surpassed.
70% Global Market Weight Equities | 15% Long Treasuries 15% short TIPS & cash || RSU + ESPP

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

letsgobobby wrote:
Thu Sep 27, 2018 12:27 pm
I'd love to see a comparison between a 60/40 CAPE strategy and a 50/50 buy and hold, for an accumulator.
It seems a little bit apple and orange to compare in such a way, but ok, I ran the numbers... I assumed a 30 years accumulation period, starting from a portfolio balance of \$1000, and adding \$1000 of savings every month. All numbers are adjusted for inflation -- so the nominal amount of monthly savings actually increases).

Before I post charts corresponding to your request, let me quickly comment on the comparison between a 60/40 CAPE strategy and a 60/40 buy and hold, for an accumulator. If the starting AA is the same (60/40), then for all accumulation cycles I tested, the end balance is better with the fixed AA. As to max. drawdowns, they are NOT meaningfully different (even for the cycles including the 2002 crisis). Volatility (std-deviation) isn't meaningfully different, with a small advantage for the fixed AA by the late 60s.

siamond
Posts: 5138
Joined: Mon May 28, 2012 5:50 am

### Re: My CAPE10 based IPS

siamond wrote:
Fri Sep 28, 2018 9:41 pm
letsgobobby wrote:
Thu Sep 27, 2018 12:27 pm
I'd love to see a comparison between a 60/40 CAPE strategy and a 50/50 buy and hold, for an accumulator.
It seems a little bit apple and orange to compare in such a way, but ok, I ran the numbers... I assumed a 30 years accumulation period, starting from a portfolio balance of \$1000, and adding \$1000 of savings every month. All numbers are adjusted for inflation -- so the nominal amount of monthly savings actually increases).
Here is the outcome of the corresponding test, with the CAPE 60/40 vs Fixed 50/50 AA. First, the end balance (after 30 years) per starting month. It's a mixed bag.

Next, volatility (std-deviation). Quite remarkably, it's a tie.

Next, the maximum drawdown. Well, no question which AA wins here...

nedsaid
Posts: 12976
Joined: Fri Nov 23, 2012 12:33 pm

### Re: My CAPE10 based IPS

watchnerd wrote:
Fri Sep 28, 2018 10:42 am
nedsaid wrote:
Sun Sep 23, 2018 3:37 pm

It does drive me a bit crazy when people try to make these tactical documents. I think this goes against the whole reason for writing an IPS. It is okay to say something like this, "When markets get to valuation and sentiment extremes, I will consider taking perhaps 15% to 20% off the top of my stock investments at market highs and will consider adding 15% to 20% to my stocks at market lows."
Seems logical, but how do you know when you hit a market high?

Basing it on indicators like "Dow now at highest ever" would be one way, but those highs are often sustained for long periods, or surpassed.
It is a good comment, you raised a good point. The whole reason we invest in stocks are for those new market highs. We hope those new market highs keep coming.

But notice that I also talked about extremes in market valuation and sentiment. My belief is that valuations are perhaps stretched here but still reasonable, I am not seeing euphoria here either. Sentiment and valuation are not at extremes here. It depends upon whether or not you believe that recently great earnings are sustainable. I believe they are. The new normal was 2% GDP growth, the new, new normal is 3% plus GDP growth. My view is that faster GDP growth makes recently great earnings sustainable.
A fool and his money are good for business.

Financial Engineer
Posts: 12
Joined: Tue Jun 06, 2017 10:01 pm

### Re: My CAPE10 based IPS

I have tried and failed (after significant effort) to find a systematic backtest using CAPE that would have resulted in a higher return over the full data history available. Even with perfect hindsight the right path is not clear. This method of market timing may make one feel better but I see no indication it will pay off financially. Please post on your results once every year for the next 25 years.
Do you think the market is pricey? S&P 500 CAGR from 1929 peak (CAPE = 32.6) to 1982 low (CAPE = 6.7) was 3.8% real (7.3% nominal). Stay the course.

willthrill81
Posts: 15165
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

### Re: My CAPE10 based IPS

Financial Engineer wrote:
Sat Sep 29, 2018 8:13 am
Please post on your results once every year for the next 25 years.
Even that would be insufficient because we cannot evaluate a strategy strictly on the basis of results. For instance, a dice may be weighted to come up heads 51% of the time, in which case we should bet on heads every time. But there is still a significant probability that tails will come up several times in a row.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

https://www.morningstar.com/videos/3539 ... roach.html

Below is from a discussion between Frank Kinniry of Vanguard and Christine Benz of Morningstar in 2010, and is taken from the above link:

Benz: ...But the tactical umbrella has a lot of different strategies beneath it, and I'm wondering are there any strategies that make more sense to you than others within the tactical realm.

Kinniry: We've looked at it, long and hard. And actually, we have a real reason to look at this and spend a lot of time looking at it, and really the only thing we've found that has some merit is at the stock/bond mix, meaning we don't see a lot between U.S. and international or growth and value or size, but we do see some ability of tactical allocation in the extremes and really the extremes are limited periods in time where this occurs. So, 1998-1999, when you saw a valuations at unprecedented levels, one could have said the equity market was set up for potentially lower returns, and in 2009, believe it or not, at the bottom, the equity market looked about as attractive as it had in 20 to 25-plus years.

And so there is a lot of noise in the middle, but there are some small windows in time where the stock/bond mix, you may want to shade, and we would say only small shades if you were 60/40 stock/bonds, maybe it would go up 5% or 10% in your equities or down 5% or 10%, but really never making wholesale moves.

Benz: And arguably a rebalancing strategy would kind of get you there.

Kinniry: Absolutely, a rebalancing strategy is probably the best way to take advantage of some of these opportunities without making wholesale changes; that's exactly right."

My conclusions is that when it comes to market timing, Kinniry is implying that the only one that works is a valuation based strategy at the stock/bond level. And he's also saying that such timing only works during the limited time periods, when valuations are at extremes.

His comment about market timing between US and international not working is interesting. Is that based on binary testing between US or international? Or is it based on testing at a more detailed level, such as US and Asia and Europe and emerging markets?

http://www.efficientfrontier.com/ef/197/rebal197.htm

William Bernstein:

"Over very long time horizons there is usually relatively little difference in the returns in most national equity markets; under such circumstances rebalanced portfolios dominate. For example, when looking at the 1970-94 period, rebalancing various MSCI national asset pairs almost always provides returns superior to nonrebalanced national pairs. Only when long term return differences among asssets exceed 5 percent do nonrebalanced portfolios provide superior returns, and then only at the cost of increased risk. (The exceptions which prove the rule are the very high returns of Japanese equity, and the very low returns of Australian and Italian equity.) Care should be taken to note that the superiority of rebalancing as a long term strategy pertains only to national and regional markets as a whole, and not among different industry groups. Over the course of decades entire industries often shrivel while others prosper mightily; rebalancing the pharmacutecal and petroleum segments of the S&P500 over the past three decades would have been a disastrous strategy."

William Bernstein makes a case for overbalancing, which takes rebalancing one step further. I see now how overbalancing within stock asset subclasses, at the national or regional level, might improve returns. One can make the case that overbalancing is a market timing strategy.

james22
Posts: 1495
Joined: Tue Aug 21, 2007 2:22 pm

### Re: My CAPE10 based IPS

Investors would love to find an easy way to tell them when to get out of the market to avoid an oncoming train on the tracks, especially at today’s elevated valuation levels.

Unfortunately, valuation is not going to do that for you because expensive markets can always get more expensive, fundamentals can change, and no one really knows what level of valuations will cause investors to change their perception of risk until after the fact.

Valuations are useful for setting expectations but not timing the market.

https://awealthofcommonsense.com/2018/0 ... g-is-hard/

watchnerd
Posts: 2669
Joined: Sat Mar 03, 2007 11:18 am
Location: Seattle, WA, USA

### Re: My CAPE10 based IPS

james22 wrote:
Sun Sep 30, 2018 12:48 am
Investors would love to find an easy way to tell them when to get out of the market to avoid an oncoming train on the tracks, especially at today’s elevated valuation levels.

Unfortunately, valuation is not going to do that for you because expensive markets can always get more expensive, fundamentals can change, and no one really knows what level of valuations will cause investors to change their perception of risk until after the fact.

Valuations are useful for setting expectations but not timing the market.

https://awealthofcommonsense.com/2018/0 ... g-is-hard/
That was a great article, thank you for that.
70% Global Market Weight Equities | 15% Long Treasuries 15% short TIPS & cash || RSU + ESPP

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

The following is my market timing strategy. For further details and rationale, please see my previous posts in this thread.

I like William Bernstein's overbalancing strategy. If you're invested in national or regional stock funds, instead of rebalancing every 3 years, overbalance. Go slightly more than you otherwise would. For example, if you were going to decreased US stock exposure by 5%, decrease it by 5.5%. This is a dynamic asset allocation strategy that is valuation based. If you want, you could use expected returns based on CAPE instead. And the recommended changes are small. When it comes to market timing, everyone that I respect says that if you're going to do it, do it modestly. Market timing is a low return activity, with a high dispersion of results.

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.

watchnerd
Posts: 2669
Joined: Sat Mar 03, 2007 11:18 am
Location: Seattle, WA, USA

### Re: My CAPE10 based IPS

Park wrote:
Sun Sep 30, 2018 9:18 am

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.
Sell how much?

If I hold total US and total International, and US CAPE >40, are you saying sell 100% of US and put it all in international?
70% Global Market Weight Equities | 15% Long Treasuries 15% short TIPS & cash || RSU + ESPP

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

watchnerd wrote:
Sun Sep 30, 2018 9:38 am
Park wrote:
Sun Sep 30, 2018 9:18 am

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.
Sell how much?

If I hold total US and total International, and US CAPE >40, are you saying sell 100% of US and put it all in international?
Yes. If international has a CAPE of 39.9 at same time that US CAPE is greater than 40, I don't know what to do.

I'd like to emphasize that this is my plan. I'm not recommending that anyone else use this strategy.

For this strategy, it would probably be better to divide international into Europe, Asia and emerging markets etc. That would have been a better way to avoid the Japanese stock market bubble.

watchnerd
Posts: 2669
Joined: Sat Mar 03, 2007 11:18 am
Location: Seattle, WA, USA

### Re: My CAPE10 based IPS

Park wrote:
Sun Sep 30, 2018 9:43 am
watchnerd wrote:
Sun Sep 30, 2018 9:38 am
Park wrote:
Sun Sep 30, 2018 9:18 am

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.
Sell how much?

If I hold total US and total International, and US CAPE >40, are you saying sell 100% of US and put it all in international?
Yes. If international has a CAPE of 39.9 at same time that US CAPE is greater than 40, I don't know what to do.

I'd like to emphasize that this is my plan. I'm not recommending that anyone else use this strategy.

For this strategy, it would probably be better to divide international into Europe, Asia and emerging markets etc. That would have been a better way to avoid the Japanese stock market bubble.
Total international market indexes are 90% correlated wish US; developed market subsegments are also highly correlated.

I don't see how your plan will work unless you end up investing mostly in emerging markets most of the time.
70% Global Market Weight Equities | 15% Long Treasuries 15% short TIPS & cash || RSU + ESPP

Park
Posts: 691
Joined: Sat Nov 06, 2010 4:56 pm

### Re: My CAPE10 based IPS

watchnerd wrote:
Sun Sep 30, 2018 9:48 am
Park wrote:
Sun Sep 30, 2018 9:43 am
watchnerd wrote:
Sun Sep 30, 2018 9:38 am
Park wrote:
Sun Sep 30, 2018 9:18 am

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.
Sell how much?

If I hold total US and total International, and US CAPE >40, are you saying sell 100% of US and put it all in international?
Yes. If international has a CAPE of 39.9 at same time that US CAPE is greater than 40, I don't know what to do.

I'd like to emphasize that this is my plan. I'm not recommending that anyone else use this strategy.

For this strategy, it would probably be better to divide international into Europe, Asia and emerging markets etc. That would have been a better way to avoid the Japanese stock market bubble.
Total international market indexes are 90% correlated wish US; developed market subsegments are also highly correlated.

I don't see how your plan will work unless you end up investing mostly in emerging markets most of the time.
At the end of 1989, the Japanese stock market had a CAPE of around 90. On Jan 1 1990, the CAPE of the S&P500 was around 17. I don't disagree with your 90% correlation of total international with the US. But what is the correlation at the uncommon times when valuations are at their extremes?

Topic Author
letsgobobby
Posts: 12074
Joined: Fri Sep 18, 2009 1:10 am

### Re: My CAPE10 based IPS

Deleted
Last edited by letsgobobby on Fri Apr 26, 2019 2:20 pm, edited 1 time in total.

watchnerd
Posts: 2669
Joined: Sat Mar 03, 2007 11:18 am
Location: Seattle, WA, USA

### Re: My CAPE10 based IPS

Park wrote:
Sun Sep 30, 2018 9:58 am
watchnerd wrote:
Sun Sep 30, 2018 9:48 am
Park wrote:
Sun Sep 30, 2018 9:43 am
watchnerd wrote:
Sun Sep 30, 2018 9:38 am
Park wrote:
Sun Sep 30, 2018 9:18 am

The following is the second part of my market timing strategy. When the CAPE of a national or regional stock fund is more than 40, sell. Don't go to cash, but put it in other stock markets with CAPE lower than 40. If there are no other such stock markets, I don't know what to do.

Should you get back in, when CAPE goes lower than 40? You could end up getting whipsawed doing that.

When to get back in?

viewtopic.php?f=10&t=251522

" Stocks have had remarkably consistent 5%-6% real returns over four centuries."

The following is one possible interpretation. The historical real return of cash is about 0%. With stocks, there is the possiblity that 2 years from now, they'll have lost half their value. So stock investors have demanded a real premium of 5-6% to compensate for that risk. Based on the CAPE model, 5-6% real return corresponds to CAPE of 17.9-21.5.

Assume that investors expect a real return of 3%. That corresponds to a CAPE of 35.8.

So get back in when CAPE less than 35.
Sell how much?

If I hold total US and total International, and US CAPE >40, are you saying sell 100% of US and put it all in international?
Yes. If international has a CAPE of 39.9 at same time that US CAPE is greater than 40, I don't know what to do.

I'd like to emphasize that this is my plan. I'm not recommending that anyone else use this strategy.

For this strategy, it would probably be better to divide international into Europe, Asia and emerging markets etc. That would have been a better way to avoid the Japanese stock market bubble.
Total international market indexes are 90% correlated wish US; developed market subsegments are also highly correlated.

I don't see how your plan will work unless you end up investing mostly in emerging markets most of the time.
At the end of 1989, the Japanese stock market had a CAPE of around 90. On Jan 1 1990, the CAPE of the S&P500 was around 17. I don't disagree with your 90% correlation of total international with the US. But what is the correlation at the uncommon times when valuations are at their extremes?
Do you want a portfolio that optimizes for best performance once in a decade?

Or do you want a portfolio that gives best performance, on average, over time?
70% Global Market Weight Equities | 15% Long Treasuries 15% short TIPS & cash || RSU + ESPP

nedsaid
Posts: 12976
Joined: Fri Nov 23, 2012 12:33 pm

### Re: My CAPE10 based IPS

Financial Engineer wrote:
Sat Sep 29, 2018 8:13 am
I have tried and failed (after significant effort) to find a systematic backtest using CAPE that would have resulted in a higher return over the full data history available. Even with perfect hindsight the right path is not clear. This method of market timing may make one feel better but I see no indication it will pay off financially. Please post on your results once every year for the next 25 years.
A big reason for this is that market P/E ratios have been creeping up for a long time. We talk a lot here about "reversion to the mean", but in the environment of rising P/E ratios, how do you define reversion to the mean? What mean are you reverting to?
A fool and his money are good for business.

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

### Re: My CAPE10 based IPS

Do as you want as it is your money (OP and others), but do want to point out an issue with using previous data (such as PE10) and projecting results based on past results. I have brought this up MANY times, but do NOT use the past data to be predictive. For example, Vanguard paper on predictive use of different valuation metrics showed they are all almost useless. The only real exception is they did look at the tail ends (12.% on this side and 12.5% on that side) of PE10 and saw it did have an R2 of 0.8. Crunching the data myself came out with roughly a PE10 <10 or >25 as the lines in the sand on either side.

Now here comes the problem. Those tails are derived from knowing the results of the ENTIRE data set to now. The guy in late 1990's only had the data UP to his time period. He did not have the resulting fall out AFTER the late 1990's showing PE>30's was a disaster. So the question remains is does that data help prospectively going forward since each new data point may change those tail numbers OR only useful after the fact which is useless for a real investor? Did the Great Depression investor even if they did know what PE10 was (more likely PE7 that Ben Graham popularized) have a time period before theirs showing a PE<10 as a great buying opportunity? Just something to ponder.

BTW, I think if one is going to market time I think the OP approach is very reasonable. You are not betting the farm on it so even if you are wrong it isn't a disaster. I would likely do it myself, but have a gut feeling since I have large taxable account the friction of selling at taking LTCG will likely wipe out a lot of the advantage (if any) when it is time to cut back. Also, don't know if the drag that is stuck in bonds/ cash for extended periods would negate the eventual upswing of equities nullifies some/ if any of the advantage.

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