My trend following strategy and experience

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Locked
User avatar
Forester
Posts: 1348
Joined: Sat Jan 19, 2019 2:50 pm
Location: UK

Re: My trend following strategy and experience

Post by Forester » Mon Oct 14, 2019 12:31 pm

IMHO if doing a simple strategy, use a low/min vol index. Less junky companies = better signal, less troughs & less peaks = less whipsaws. And even on a false signal you're selling higher & buying lower.

jayhawkerbeef
Posts: 263
Joined: Tue Jul 22, 2014 11:10 am

Re: My trend following strategy and experience

Post by jayhawkerbeef » Mon Oct 14, 2019 1:40 pm

BlueEars wrote:
Mon Oct 14, 2019 11:45 am
jayhawkerbeef wrote:
Mon Oct 14, 2019 11:37 am
My strategy was just a simple 3mo dual momentum trading month end price, hate seeing/experiencing how terribly I’ve lagged the S&P over the last 4 years since I started.

Know that’s not long enough to follow any strategy. However, I did best the S&p in 2018, -1% to -4.5%. But like you I know I will hate drawdowns, and will not just sit and do nothing. And this year, S&P up ~20% and I’m 4%. Really really second guessing timing thing. Haha
I suspect your strategy is suffering because of the 3 month timing selection. I would imagine this is too short a period. How does 3 month work on a back test over a long time period? One period to look at would be the 1990's when few trend following methods beat buy-hold.
Actually checked that on PV and it would have been worse the longer the holding period; below is based upon a 12mo.

Year Inflation Timing VG 500 Timing Balance VG 500 Balance
2012 1.74% 19.07% 15.82% $11,907 $11,582
2013 1.50% 21.22% 32.18% $14,433 $15,309
2014 0.76% 9.64% 13.51% $15,825 $17,377
2015 0.73% -12.71% 1.25% $13,813 $17,594
2016 2.07% -2.13% 11.82% $13,519 $19,673
2017 2.11% 25.81% 21.67% $17,008 $23,936
2018 1.91% -0.22% -4.52% $16,971 $22,853
2019 2.12% 8.12% 20.44% $18,349 $27,525

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Mon Oct 14, 2019 3:20 pm

Forester wrote:
Mon Oct 14, 2019 12:31 pm
IMHO if doing a simple strategy, use a low/min vol index. Less junky companies = better signal, less troughs & less peaks = less whipsaws. And even on a false signal you're selling higher & buying lower.
I think that a target volatility approach using broad market indices and bonds, preferably long-term bonds, is a fine approach. It's definitely simpler than the one I'm using here.
“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

TheDDC
Posts: 1035
Joined: Mon Jan 08, 2018 11:11 am

Re: My trend following strategy and experience

Post by TheDDC » Tue Oct 15, 2019 10:13 pm

willthrill81 wrote:
Mon Oct 14, 2019 3:20 pm
Forester wrote:
Mon Oct 14, 2019 12:31 pm
IMHO if doing a simple strategy, use a low/min vol index. Less junky companies = better signal, less troughs & less peaks = less whipsaws. And even on a false signal you're selling higher & buying lower.
I think that a target volatility approach using broad market indices and bonds, preferably long-term bonds, is a fine approach. It's definitely simpler than the one I'm using here.
Will- Would you mind "thrilling" us by updating your signature with a "buy into stocks/move to bonds" signal? Maybe "Blue horseshoe loves stocks" would be appropriate. Seriously, though, I would be interested in tracking your trend following strategy and correlating with what I might notice using [what I think is] your method. Understanding how quickly you react to a change in the criteria (or one of two when buying back in) would be helpful to all of us I think.

-TheDDC
Rules to wealth building: 90-100% VTSAX piled high and deep, 0-10% VIGAX tilt, 0% given away to banks, minimize amount given to medical-industrial complex

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Wed Oct 16, 2019 9:47 am

TheDDC wrote:
Tue Oct 15, 2019 10:13 pm
willthrill81 wrote:
Mon Oct 14, 2019 3:20 pm
Forester wrote:
Mon Oct 14, 2019 12:31 pm
IMHO if doing a simple strategy, use a low/min vol index. Less junky companies = better signal, less troughs & less peaks = less whipsaws. And even on a false signal you're selling higher & buying lower.
I think that a target volatility approach using broad market indices and bonds, preferably long-term bonds, is a fine approach. It's definitely simpler than the one I'm using here.
Will- Would you mind "thrilling" us by updating your signature with a "buy into stocks/move to bonds" signal? Maybe "Blue horseshoe loves stocks" would be appropriate. Seriously, though, I would be interested in tracking your trend following strategy and correlating with what I might notice using [what I think is] your method. Understanding how quickly you react to a change in the criteria (or one of two when buying back in) would be helpful to all of us I think.

-TheDDC
I am updating when I move from stocks into bonds in this thread. I am not updating every time I make a change from something like LCG to MCG, but I will provide a summary report at the end of each year. Right now, I'm still mostly in LCG.
“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

User avatar
tadamsmar
Posts: 9024
Joined: Mon May 07, 2007 12:33 pm

Re: My trend following strategy and experience

Post by tadamsmar » Wed Oct 16, 2019 11:33 am

TheDDC wrote:
Tue Oct 15, 2019 10:13 pm
Will- Would you mind "thrilling" us by updating your signature with a "buy into stocks/move to bonds" signal? Maybe "Blue horseshoe loves stocks" would be appropriate. Seriously, though, I would be interested in tracking your trend following strategy and correlating with what I might notice using [what I think is] your method. Understanding how quickly you react to a change in the criteria (or one of two when buying back in) would be helpful to all of us I think.

-TheDDC
I don't think you have the information required to track his trend following strategy.

I don't think there is one "buy into stocks/move to bonds" signal. Instead, there is one such signal per fiduciary, I think.

If you had a list of the investments for each fiduciary then I think you could track and model his trend following strategy. Specifically, the investments that willthrill81 is willing to invest in at each fiduciary as part of his strategy. That might be enough unless some of the investments are such that extensive public information is not available.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Wed Oct 16, 2019 12:08 pm

tadamsmar wrote:
Wed Oct 16, 2019 11:33 am
TheDDC wrote:
Tue Oct 15, 2019 10:13 pm
Will- Would you mind "thrilling" us by updating your signature with a "buy into stocks/move to bonds" signal? Maybe "Blue horseshoe loves stocks" would be appropriate. Seriously, though, I would be interested in tracking your trend following strategy and correlating with what I might notice using [what I think is] your method. Understanding how quickly you react to a change in the criteria (or one of two when buying back in) would be helpful to all of us I think.

-TheDDC
I don't think you have the information required to track his trend following strategy.

I don't think there is one "buy into stocks/move to bonds" signal. Instead, there is one such signal per fiduciary, I think.

If you had a list of the investments for each fiduciary then I think you could track and model his trend following strategy. Specifically, the investments that willthrill81 is willing to invest in at each fiduciary as part of his strategy. That might be enough unless some of the investments are such that extensive public information is not available.
That's correct. My investment choices vary tremendously from one account to the next. I have far more options in our IRAs than in my 457 plan (only six there). But most of my performance could be tracked with broad market indices since the stock/bond allocation has a much greater impact on one's returns than one's choice of stock class.
“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

User avatar
Forester
Posts: 1348
Joined: Sat Jan 19, 2019 2:50 pm
Location: UK

Re: My trend following strategy and experience

Post by Forester » Fri Oct 18, 2019 3:18 pm

"Momentum is a data-mined anomaly" - article by a guy who dislikes trend following. He cites the performance of trend following performed on EEM, the volatile emerging markets index.

https://www.priceactionlab.com/Blog/20 ... et-timing/

Lee_WSP
Posts: 2700
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: My trend following strategy and experience

Post by Lee_WSP » Fri Oct 18, 2019 3:47 pm

Forester wrote:
Fri Oct 18, 2019 3:18 pm
"Momentum is a data-mined anomaly" - article by a guy who dislikes trend following. He cites the performance of trend following performed on EEM, the volatile emerging markets index.

https://www.priceactionlab.com/Blog/20 ... et-timing/
It may well be data mined, but to debunk it, you need to compare it to a fund with the same volatility. EEM is extraordinarily volatile.

User avatar
BlueEars
Posts: 3845
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: My trend following strategy and experience

Post by BlueEars » Fri Oct 18, 2019 5:10 pm

Yes for algorithms I've tried, having a mildly moving asset is better then the volatile ones. Haven't figured out how to do EEM or gold or sector funds.

User avatar
tadamsmar
Posts: 9024
Joined: Mon May 07, 2007 12:33 pm

Re: My trend following strategy and experience

Post by tadamsmar » Mon Oct 21, 2019 7:45 am

Lee_WSP wrote:
Fri Oct 18, 2019 3:47 pm
Forester wrote:
Fri Oct 18, 2019 3:18 pm
"Momentum is a data-mined anomaly" - article by a guy who dislikes trend following. He cites the performance of trend following performed on EEM, the volatile emerging markets index.

https://www.priceactionlab.com/Blog/20 ... et-timing/
It may well be data mined, but to debunk it, you need to compare it to a fund with the same volatility. EEM is extraordinarily volatile.
Not sure that pure momentum strategies need to be debunked. Did they have credence in the first place?

Asness' dissertation apparently gave some respectability to a combination of value plus momentum, but that is not what we are talking about on this thread.

Tdubs
Posts: 1054
Joined: Tue Apr 24, 2018 7:50 pm

Re: My trend following strategy and experience

Post by Tdubs » Mon Oct 21, 2019 7:57 am

willthrill81 wrote:
Wed Oct 16, 2019 9:47 am
TheDDC wrote:
Tue Oct 15, 2019 10:13 pm
willthrill81 wrote:
Mon Oct 14, 2019 3:20 pm
Forester wrote:
Mon Oct 14, 2019 12:31 pm
IMHO if doing a simple strategy, use a low/min vol index. Less junky companies = better signal, less troughs & less peaks = less whipsaws. And even on a false signal you're selling higher & buying lower.
I think that a target volatility approach using broad market indices and bonds, preferably long-term bonds, is a fine approach. It's definitely simpler than the one I'm using here.
Will- Would you mind "thrilling" us by updating your signature with a "buy into stocks/move to bonds" signal? Maybe "Blue horseshoe loves stocks" would be appropriate. Seriously, though, I would be interested in tracking your trend following strategy and correlating with what I might notice using [what I think is] your method. Understanding how quickly you react to a change in the criteria (or one of two when buying back in) would be helpful to all of us I think.

-TheDDC
I am updating when I move from stocks into bonds in this thread. I am not updating every time I make a change from something like LCG to MCG, but I will provide a summary report at the end of each year. Right now, I'm still mostly in LCG.
Sorry for wandering into this thread so late. Are your updates buried among the 20 pages of discussion? Or, is there one place to find it all the moves?

User avatar
tadamsmar
Posts: 9024
Joined: Mon May 07, 2007 12:33 pm

Re: My trend following strategy and experience

Post by tadamsmar » Mon Oct 21, 2019 8:00 am

Tdubs wrote:
Mon Oct 21, 2019 7:57 am

Sorry for wandering into this thread so late. Are your updates buried among the 20 pages of discussion? Or, is there one place to find it all the moves?
Updates are at the bottom of the original post of this thread.

Tdubs
Posts: 1054
Joined: Tue Apr 24, 2018 7:50 pm

Re: My trend following strategy and experience

Post by Tdubs » Mon Oct 21, 2019 8:02 am

tadamsmar wrote:
Mon Oct 21, 2019 8:00 am
Tdubs wrote:
Mon Oct 21, 2019 7:57 am

Sorry for wandering into this thread so late. Are your updates buried among the 20 pages of discussion? Or, is there one place to find it all the moves?
Updates are at the bottom of the original post of this thread.
Thanks!

Galton's Ghost
Posts: 4
Joined: Mon Oct 21, 2019 7:43 pm

Re: My trend following strategy and experience

Post by Galton's Ghost » Mon Oct 21, 2019 7:50 pm

Will,

Do you still use a single look-back period? The research that I've seen and that I've done myself seems to indicate that multiple look-backs are better in terms of eliminating specification risk, i.e. bad (or good) luck.

Also, have you ever checked out Paul Novell's work. He does something very similar but with multiple economic indicators.

Best,
Galton

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Mon Oct 21, 2019 8:52 pm

Galton's Ghost wrote:
Mon Oct 21, 2019 7:50 pm
Will,

Do you still use a single look-back period? The research that I've seen and that I've done myself seems to indicate that multiple look-backs are better in terms of eliminating specification risk, i.e. bad (or good) luck.

Also, have you ever checked out Paul Novell's work. He does something very similar but with multiple economic indicators.

Best,
Galton
Yes, I use a 12 month look-back for the UER and a 7 month for the assets. From what I've seen, the long-term results of most look-back periods ranging from 3 to 12 months (or longer) are pretty similar, so the complications introduced by multiple look-back periods are not worthwhile to me.

No, I'm not familiar with Paul Novell.
“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

Always passive
Posts: 554
Joined: Fri Apr 14, 2017 4:25 am
Location: Israel

Re: My trend following strategy and experience

Post by Always passive » Tue Oct 22, 2019 8:24 am

Question: why do you use moving average and not momentum on 7 month (or other period) return, like the dual momentum strategy

Galton's Ghost
Posts: 4
Joined: Mon Oct 21, 2019 7:43 pm

Re: My trend following strategy and experience

Post by Galton's Ghost » Tue Oct 22, 2019 10:28 am

willthrill81 wrote:
Mon Oct 21, 2019 8:52 pm
Galton's Ghost wrote:
Mon Oct 21, 2019 7:50 pm
Will,

Do you still use a single look-back period? The research that I've seen and that I've done myself seems to indicate that multiple look-backs are better in terms of eliminating specification risk, i.e. bad (or good) luck.

Also, have you ever checked out Paul Novell's work. He does something very similar but with multiple economic indicators.

Best,
Galton
Yes, I use a 12 month look-back for the UER and a 7 month for the assets. From what I've seen, the long-term results of most look-back periods ranging from 3 to 12 months (or longer) are pretty similar, so the complications introduced by multiple look-back periods are not worthwhile to me.

No, I'm not familiar with Paul Novell.
Regarding lookbacks, I was talking about multiple look-backs for the assets, i.e. instead of using just one - the 7-month in this case - use several - maybe 3,6,9 and 12. I realize that over long periods of time (20 or 30 years), they all give similar results; however, over shorter periods of time, they can give dramatically different results, which can be hard to swallow. Using multiple look-backs helps you avoid luck - good and bad - leaving you with a more pure access to trend. As either the ReSolve guys or Corey Hoffstein say, " We'd rather be generally right, than specifically wrong."

Regarding Paul Novell, he has written about this strategy quite a bit over at Investing for a Living. He uses several economic indicators instead of just unemployment. If any go "red," he flips on the trend screen. Again, very similar to you, he just adds economic indicators to have a bit more breadth. He's written quite a bit about this strategy so I figured you might want to check him out. (If it matters, I'm not associated with him in any way, other than occasionally having had an email chat about this strategy.)

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Tue Oct 22, 2019 10:49 am

Galton's Ghost wrote:
Tue Oct 22, 2019 10:28 am
willthrill81 wrote:
Mon Oct 21, 2019 8:52 pm
Galton's Ghost wrote:
Mon Oct 21, 2019 7:50 pm
Will,

Do you still use a single look-back period? The research that I've seen and that I've done myself seems to indicate that multiple look-backs are better in terms of eliminating specification risk, i.e. bad (or good) luck.

Also, have you ever checked out Paul Novell's work. He does something very similar but with multiple economic indicators.

Best,
Galton
Yes, I use a 12 month look-back for the UER and a 7 month for the assets. From what I've seen, the long-term results of most look-back periods ranging from 3 to 12 months (or longer) are pretty similar, so the complications introduced by multiple look-back periods are not worthwhile to me.

No, I'm not familiar with Paul Novell.
Regarding lookbacks, I was talking about multiple look-backs for the assets, i.e. instead of using just one - the 7-month in this case - use several - maybe 3,6,9 and 12. I realize that over long periods of time (20 or 30 years), they all give similar results; however, over shorter periods of time, they can give dramatically different results, which can be hard to swallow. Using multiple look-backs helps you avoid luck - good and bad - leaving you with a more pure access to trend. As either the ReSolve guys or Corey Hoffstein say, " We'd rather be generally right, than specifically wrong."

Regarding Paul Novell, he has written about this strategy quite a bit over at Investing for a Living. He uses several economic indicators instead of just unemployment. If any go "red," he flips on the trend screen. Again, very similar to you, he just adds economic indicators to have a bit more breadth. He's written quite a bit about this strategy so I figured you might want to check him out. (If it matters, I'm not associated with him in any way, other than occasionally having had an email chat about this strategy.)
When I was forming my strategy, I thought about using multiple look-back periods, but the added complexity of implementation didn't appeal to me, and I'm not that concerned with the short-term differences stemming from the use of one vs. several.

There are other economic indicators or indices that could certainly be used as well. The Conference Board LEI has had a very good track record, as has the PMI.
“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

User avatar
tadamsmar
Posts: 9024
Joined: Mon May 07, 2007 12:33 pm

Re: My trend following strategy and experience

Post by tadamsmar » Tue Oct 22, 2019 12:08 pm

Always passive wrote:
Tue Oct 22, 2019 8:24 am
Question: why do you use moving average and not momentum on 7 month (or other period) return, like the dual momentum strategy
Willthill81 does not use moving average he uses relative strength. (He uses moving average only for unemployment)

As far as I can tell, each month for each fiduciary, he chooses the asset with the highest relative strength from the set of equity assets available at that fiduciary. (Or maybe he chooses from a subset, not the full set, not sure). Then he invests in that asset if and only if it has a higher relative strength than does VBMFX or unemployment is not above it's 12-month moving average. Otherwise, he invests in VBMFX or something else. See here:

viewtopic.php?f=10&t=270035&p=4560031&h ... h#p4560031

Interestingly, he implies that he does not necessarily move to VBMFX even if it is available, and I don't think he has ever explained this. But it may be that in practice he does use VBMFX when it is available at any of his limited list of fiduciaries.

If you go here, and you click on Methodology then you will see a discussion of what "relative strength" is:

https://www.portfoliovisualizer.com/faq

I am assuming he does not use any of the bells and whistles mentioned there.

But they don't provide an explicit formula. I think it is just (E-B)/B where B if the beginning price and E is the end price of the 7-month period.

The title of this thread is "My trend following strategy and experience" but (unless I am mistaken) after 20 pages his strategy is still somewhat vaguely specified.

Edit: added the stipulation about unemployment
Last edited by tadamsmar on Tue Oct 22, 2019 4:04 pm, edited 2 times in total.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Tue Oct 22, 2019 12:16 pm

tadamsmar wrote:
Tue Oct 22, 2019 12:08 pm
Interestingly, he implies that he does not necessarily move to VBMFX even if it is available, and I don't think he has ever explained this. But it may be that in practice he does use VBMFX when it is available at any of his limited list of fiduciaries.
VBMFX is only available in our IRAs. And recall that per my strategy, I can only move into any type of fixed income if the UER is above its 12 month moving average.
tadamsmar wrote:
Tue Oct 22, 2019 12:08 pm
But they don't provide an explicit formula. I think it is just (E-B)/B where B if the beginning price and E is the end price of the 7-month period.
Yes, that's correct.
“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

Always passive
Posts: 554
Joined: Fri Apr 14, 2017 4:25 am
Location: Israel

Re: My trend following strategy and experience

Post by Always passive » Sat Oct 26, 2019 1:13 am

Hi WILLTHRILL81:
To ask my question, let me quote from your original post
"this system calls me to own stocks unless both (1) the U.S. unemployment rate (UER) is above its 12 month moving average and (2) all available stock indexes have had lower performance than that of bond indexes over the prior 7 months"
My question is the following: US unemployment is a potential indicator of recession in the US. However, in your list of options, you include non US stock indexes. Does that mean that you believe that a US recession always impacts negatively foreign equities? Is it not possible that foreign equities may do well in a US recession?
Erwin

Busdrvr
Posts: 173
Joined: Fri Jan 29, 2016 11:56 am

Re: My trend following strategy and experience

Post by Busdrvr » Thu Oct 31, 2019 8:47 am

I see Sept. UE rate @ bls is lowest number on chart at 3.5. Which is nice!

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Thu Oct 31, 2019 9:19 am

Always passive wrote:
Sat Oct 26, 2019 1:13 am
My question is the following: US unemployment is a potential indicator of recession in the US. However, in your list of options, you include non US stock indexes. Does that mean that you believe that a US recession always impacts negatively foreign equities? Is it not possible that foreign equities may do well in a US recession?
There is a quote from the first article I linked to that directly addresses your question.
If we use U.S. economic data as a filter to time foreign securities, the performance turns out to be excellent. But if we use economic data from the foreign countries themselves, then the strategy ends up underperforming a simple unfiltered trend-following strategy. Among other things, this tells us something that we could probably have already deduced from observation: the health of our economy and our equity markets is more relevant to the performance of foreign equity markets than the health of their own economies. This is especially true with respect to large downward moves–the well-known global “crises” that drag all markets down in unison, and that make trend-following a historically profitable strategy.
https://www.philosophicaleconomics.com/2016/02/uetrend/
“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

User avatar
BlueEars
Posts: 3845
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: My trend following strategy and experience

Post by BlueEars » Thu Oct 31, 2019 10:24 am

If we look at Vanguard Total World index we see that North America is 58% of the total. https://investor.vanguard.com/mutual-fu ... olio/vtwax

I think as long as this persists the use of US stats to time broad foreign funds will be good.

Green Suit Investor
Posts: 10
Joined: Wed Feb 05, 2014 9:27 am

Re: My trend following strategy and experience

Post by Green Suit Investor » Sat Nov 23, 2019 2:56 pm

willthrill,

Kudos to you for having the long-term patience to explain the system to everyone. I have also done a lot of work along these lines so I’d like to add my 2 cents. There seem to be a lot of misconceptions out there. First, those unfamiliar with “trend following” and “relative strength” may simply not understand this strategy in its entirety. Researching those topics will clarify, and may help broaden perspectives.
1. A timing system does not replicate a portfolio of stocks and bonds. Even though it holds both stocks and bonds at different times, the key is the timing. Volatility clusters, so by avoiding stocks below the long term average it is ducking the clusters of negative volatility while still capturing the clusters of upside volatility. The result is a portfolio with returns more similar to 100% equities with a drawdown profile more similar to 100% bonds. That does not make it the silver bullet, since returns will underperform pure equities, but it offers an attractive trade-off.
2. A lot of people here seem to criticize this approach for not being 100% successful in timing past downturns. They are missing the point. This approach is to use unemployment as an edge that puts the odds in favor. No approach will be 100% right, but it can still be effective if it is 70-80% right (as this one is).
3. I agree perfectly with the insurance analogy, I think that is the best way to think about it. To be clear, trend-following is strictly a defensive strategy. Whipsaws are the premium paid to mitigate the risk of severe drawdowns borne by buy and hold approach. However, a recession-timing approach is like timing insurance, trying to only pay the premiums around the time you expect to have a payout, and dropping the insurance at all other times. This relates to the central risk of this approach as I see it, see below.
4. I agree with the philosophy of your approach - trend following alleviates the worst drawdowns, so is likely easier to execute. Many people try BH but fail to stick to it through the massive drawdowns, especially because correlations all move toward 1 during a bear market so a diversified portfolio is no refuge. The biggest opportunity for behavioral biases to creep in with this approach is when comparing to the benchmark. However, you have a firm grasp of the strategy so it seems you have the best inoculation against tracking bias.
5. Some people have commented on how this relates to the Boglehead philosophy of buy and hold. I’ll be somewhat provocative here with this part: buy and hold is purely theoretical; no one can actually realize pure buy and hold. One, no one has their lifetime of investable assets to invest at year zero. Everyone contributes over time, so it is buying and buying more. Two, the point of investing is to eventually live off these assets someday, which means selling. So it is actually buying, buying more, selling, selling more over one’s lifetime. The only difference between this buy and sell pattern and a trend-following buy and sell pattern is the time between the buys and sells and the triggers to buy and sell. Everyone is timing the market, the only difference is some use rules and do it more frequently, other use life circumstances and do it less frequently.
6. A lot of posters here seem to underappreciated the drawdown mitigating impact of this approach. A simple timing system based on the 200 day moving average an a global basket of assets had a max drawdown of 11% over the past 40 years. When trying to live off one’s assets, whether by using a safe withdrawal rate or other means, major drawdowns at the start of retirement are the biggest risk in retirement planning. Any strategy that minimizes drawdowns is therefore valuable, especially if it preserves equity-like returns. The “insurance cost” of whipsaws is worth it in this case. You pay the insurance to remove risk that you can’t afford to take. Anyone doubting the real risk of massive drawdowns should review 1929-1954, a period where it took decades for equities to get back to their pre-drawdown levels. Adding bonds to the portfolio also reduces the severity of drawdowns, but to reduce the drawdown to similar levels (around 11%) you have to hold 80% in bonds. Even Bogle said 80% is too much in bonds to support 40+ years of retirement, or decades of equity recovery in the case of the Great Depression, unless someone is already extremely wealthy.
7. For the non-retired, reducing drawdowns is still valuable from a behavioral standpoint. Again, few people are actually able to execute buy and hold, as evidenced by Vanguard’s research showing the vast majority of investors dramatically underperform their benchmarks. I would be surprised if that didn’t include some Bogleheads as well.

Similar to Blue Ears I also created a multi “indicator” approach using Yield Curve and CAPE. I backtested this system back to 1871. Nearly 150 years of data gave me confidence in the system. In response to the 1929 question, yes there was unemployment data before 1929, I had to dig to find it. If you use the 12 month unemployment average, then it would have triggered in October 1929. Obviously the data that far back is certainly questionable, but I think it is reasonable to conclude an unemployment-based timing system would be out by at least early 1930 if not earlier.

Drawbacks:
One of the central drawbacks to this strategy is the time involved. Not in managing it but understanding it at a deep enough level to actually implement it when it is underperforming the market. That means hours of working through backtests to see how the strategy has performed in different periods and doing scenario analysis with different variables to see how they interact. Anything less is unlikely to convey the deep understanding needed to withstand the rough periods (like this year). A lot of work upfront which many people wouldn’t be willing to do. The plus side is once understood, it is very easy to manage in a few minutes per month.

BlueEars pointed out the central risk in this strategy, which was only evident to me months after I developed it. It relies on the correlation between unemployment and downturns continuing into the future. Two scenarios result in permanent capital loss: One, if the recession indicators trigger (in your case unemployment) and the market goes sideways with significant volatility, resulting in many whipsaws. Two, if there is a major downturn without the unemployment triggering prior, then you will have incurred the major drawdown since your timing insurance was turned off. If both of these happened sequentially, it would result in a large loss which may never be recovered. This is the central risk of this strategy in my mind. Despite having backtested this strategy over the past 150 years, there is nothing guaranteeing this correlation will persist into the future.

Given above, I’ve deferred the recession timing aspect of my portfolio for now, and only carry the timing insurance overlay. I’m in the process of working out how significant this risk is and how I can mitigate it to make it a viable strategy again.

Again, thank you for the post and enduring effort to engage on these ideas. Reading this (long!) thread was very insightful. Thank you.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sat Nov 23, 2019 4:31 pm

Green Suit Investor wrote:
Sat Nov 23, 2019 2:56 pm
That does not make it the silver bullet, since returns will underperform pure equities, but it offers an attractive trade-off.
This is the only thing in your post that I disagree with. For one, we don't know that trend following's returns will underperform equities, since (1) we don't know that will be true in the future and (2) in the past, that was often not the case.
Green Suit Investor wrote:
Sat Nov 23, 2019 2:56 pm
2. A lot of people here seem to criticize this approach for not being 100% successful in timing past downturns. They are missing the point. This approach is to use unemployment as an edge that puts the odds in favor. No approach will be 100% right, but it can still be effective if it is 70-80% right (as this one is).
3. I agree perfectly with the insurance analogy, I think that is the best way to think about it. To be clear, trend-following is strictly a defensive strategy. Whipsaws are the premium paid to mitigate the risk of severe drawdowns borne by buy and hold approach. However, a recession-timing approach is like timing insurance, trying to only pay the premiums around the time you expect to have a payout, and dropping the insurance at all other times. This relates to the central risk of this approach as I see it, see below.
4. I agree with the philosophy of your approach - trend following alleviates the worst drawdowns, so is likely easier to execute. Many people try BH but fail to stick to it through the massive drawdowns, especially because correlations all move toward 1 during a bear market so a diversified portfolio is no refuge. The biggest opportunity for behavioral biases to creep in with this approach is when comparing to the benchmark. However, you have a firm grasp of the strategy so it seems you have the best inoculation against tracking bias.
5. Some people have commented on how this relates to the Boglehead philosophy of buy and hold. I’ll be somewhat provocative here with this part: buy and hold is purely theoretical; no one can actually realize pure buy and hold. One, no one has their lifetime of investable assets to invest at year zero. Everyone contributes over time, so it is buying and buying more. Two, the point of investing is to eventually live off these assets someday, which means selling. So it is actually buying, buying more, selling, selling more over one’s lifetime. The only difference between this buy and sell pattern and a trend-following buy and sell pattern is the time between the buys and sells and the triggers to buy and sell. Everyone is timing the market, the only difference is some use rules and do it more frequently, other use life circumstances and do it less frequently.
6. A lot of posters here seem to underappreciated the drawdown mitigating impact of this approach. A simple timing system based on the 200 day moving average an a global basket of assets had a max drawdown of 11% over the past 40 years. When trying to live off one’s assets, whether by using a safe withdrawal rate or other means, major drawdowns at the start of retirement are the biggest risk in retirement planning. Any strategy that minimizes drawdowns is therefore valuable, especially if it preserves equity-like returns. The “insurance cost” of whipsaws is worth it in this case. You pay the insurance to remove risk that you can’t afford to take. Anyone doubting the real risk of massive drawdowns should review 1929-1954, a period where it took decades for equities to get back to their pre-drawdown levels. Adding bonds to the portfolio also reduces the severity of drawdowns, but to reduce the drawdown to similar levels (around 11%) you have to hold 80% in bonds. Even Bogle said 80% is too much in bonds to support 40+ years of retirement, or decades of equity recovery in the case of the Great Depression, unless someone is already extremely wealthy.
7. For the non-retired, reducing drawdowns is still valuable from a behavioral standpoint. Again, few people are actually able to execute buy and hold, as evidenced by Vanguard’s research showing the vast majority of investors dramatically underperform their benchmarks. I would be surprised if that didn’t include some Bogleheads as well.

Similar to Blue Ears I also created a multi “indicator” approach using Yield Curve and CAPE. I backtested this system back to 1871. Nearly 150 years of data gave me confidence in the system. In response to the 1929 question, yes there was unemployment data before 1929, I had to dig to find it. If you use the 12 month unemployment average, then it would have triggered in October 1929. Obviously the data that far back is certainly questionable, but I think it is reasonable to conclude an unemployment-based timing system would be out by at least early 1930 if not earlier.

Drawbacks:
One of the central drawbacks to this strategy is the time involved. Not in managing it but understanding it at a deep enough level to actually implement it when it is underperforming the market. That means hours of working through backtests to see how the strategy has performed in different periods and doing scenario analysis with different variables to see how they interact. Anything less is unlikely to convey the deep understanding needed to withstand the rough periods (like this year). A lot of work upfront which many people wouldn’t be willing to do. The plus side is once understood, it is very easy to manage in a few minutes per month.

BlueEars pointed out the central risk in this strategy, which was only evident to me months after I developed it. It relies on the correlation between unemployment and downturns continuing into the future. Two scenarios result in permanent capital loss: One, if the recession indicators trigger (in your case unemployment) and the market goes sideways with significant volatility, resulting in many whipsaws. Two, if there is a major downturn without the unemployment triggering prior, then you will have incurred the major drawdown since your timing insurance was turned off. If both of these happened sequentially, it would result in a large loss which may never be recovered. This is the central risk of this strategy in my mind. Despite having backtested this strategy over the past 150 years, there is nothing guaranteeing this correlation will persist into the future.

Given above, I’ve deferred the recession timing aspect of my portfolio for now, and only carry the timing insurance overlay. I’m in the process of working out how significant this risk is and how I can mitigate it to make it a viable strategy again.

Again, thank you for the post and enduring effort to engage on these ideas. Reading this (long!) thread was very insightful. Thank you.
That's an excellent summary of the basic ideas involved. I agree that there is a real risk of a significant loss with this strategy, but I view that risk as being much smaller than the risk's resulting from a buy-and-hold approach, especially for me personally. And thank you for your thanks! :beer
Last edited by willthrill81 on Sat Nov 23, 2019 9:58 pm, edited 1 time in total.
“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

Green Suit Investor
Posts: 10
Joined: Wed Feb 05, 2014 9:27 am

Re: My trend following strategy and experience

Post by Green Suit Investor » Sat Nov 23, 2019 9:57 pm

If you look at a pure trend-following approach, most studies I've seen show a slight impairment in returns compared with buy and hold over the long run (40+ years). Your system is different, but it carries another sort of risk.

I think it is very important to be crystal clear on the risks involved in any strategy (including buy and hold). In the case of buy and hold, the risk is assuming deep drawdowns that don't recover before an investor needs to begin withdrawing capital or which impair an investor's goals. In the case of recession timing trend following, the risk is a loss of correlation between unemployment and market downturns, which exposes the strategy to whipsaws AND market downturns. The picture in the Philosophical Economics article made this risk really stark for me:

Image
http://www.philosophicaleconomics.com/2016/01/gtt/

That doesn't quite capture it since that only looks at whipsaws, but it conveys the idea. A timing strategy could conceivably go to zero if it incurred whipsaws in excess of the compound growth rate, which is more likely if it was exposed to deep drawdowns as well.

To use an analogy, it is very much like car insurance. Driving without car insurance and saving the premium every month might work if the accidents were occasional and relatively minor so overall the money saved is able to pay for the accidents and still grow over time. This is buy and hold. Another option is to carry insurance all the time. This means overall less capital at the end (assuming carrying no insurance continues to have positive expectancy), but it smooths out the risk a lot better. This is trend following. Still another approach is to try to time the insurance to only cover when accidents are most likely to happen. This might look like only carrying insurance during rush hour periods for a few hours per day. Overall, most accidents happen at that time so the strategy would cover losses when needed and not cost as much in the intervening periods. Best of both worlds so long as accidents continue to be correlated with rush hour periods. If something in the environment fundamentally changed, for example parking lot vandals, then the whole strategy would break down and you could incur the worst of both worlds: paying premiums AND suffering the losses. If the risk was severe enough it could have a significant impairment on capital which could be life-changing.

How could our environment change to decouple unemployment and market performance? No idea, but since it is not obvious to me why this relationship occurs, other than the really loose relationship between economic performance and market performance, I think it is a conceivable scenario for the future. If you believe that all correlations eventually breakdown, then this would be a likely scenario at some point in the future.

Of course, it is up to each of us to decide what we think the severity and appropriateness of the risks we want to assume. This was enough to give me pause at least, but perhaps you've already thought through these issues and come to different conclusions.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sat Nov 23, 2019 10:20 pm

Green Suit Investor wrote:
Sat Nov 23, 2019 9:57 pm
How could our environment change to decouple unemployment and market performance? No idea, but since it is not obvious to me why this relationship occurs, other than the really loose relationship between economic performance and market performance, I think it is a conceivable scenario for the future. If you believe that all correlations eventually breakdown, then this would be a likely scenario at some point in the future.

Of course, it is up to each of us to decide what we think the severity and appropriateness of the risks we want to assume. This was enough to give me pause at least, but perhaps you've already thought through these issues and come to different conclusions.
Yes, I'm aware that it is theoretically possible for something like this to occur. However, I believe that there is a stronger relationship between the economy and the stock market than you seem to, especially when the economy is doing poorly, and I also believe that this relationship is likely to remain in the future.
“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

freewill404
Posts: 9
Joined: Mon Apr 01, 2019 7:35 am

Re: My trend following strategy and experience

Post by freewill404 » Mon Nov 25, 2019 5:46 am

The bottom line is that in absentia of effective recession timing, I am not sure this strategy will produce a superior risk-adjusted return compared to B&H. It looks like it has more to do with market timing than bonafide trend following.

User avatar
tadamsmar
Posts: 9024
Joined: Mon May 07, 2007 12:33 pm

Re: My trend following strategy and experience

Post by tadamsmar » Mon Nov 25, 2019 9:29 am

freewill404 wrote:
Mon Nov 25, 2019 5:46 am
My problem is that ultimately, this selective trend following strategy basically boils down to being able to predict recessions (the key here is ultimately being able to tell whipsaws from genuine year-long dumps)... and we all know how that ends.
You don't seem to know where Dublin is.

Ultimately, it boils down to being able to retire at a respectable age. The OP says "I frankly don't need great returns to become financially independent at a very respectable age".

Willthrill81 does not need to beat buy and hold. As far as I can tell from the OP he just wants an investment strategy that he thinks he can stick with and I don't know of any financial analysis that would decide that matter.

Edit: But just for grins, let's assume beating buy hold is Dublin. It that case it boils down to a monthly prediction game about the relative performance of two investment strategies for the next month. The game is scored based on cumulative returns of montly returns. No real need to predict recessions per se.

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Fri Nov 29, 2019 4:45 pm

Barsoom wrote:
Wed Jun 19, 2019 10:36 am
willthrill81 wrote:
Thu Jan 17, 2019 7:16 pm
I use a nearly identical version of that advocated by the Philosophical Economist and detailed here. (He also provides an excellent overview to why trend following works here.)
When I first came across these articles about six months ago, I was interested because I was going to retire later this year and was concerned about retiring into a down market. I wasn't so much interested in timing the market per se, as I was about getting into the market in the first place with a significant lump sum pension distribution. Secondarily, I had a large position in Megacorp stock that has served me well but would need to be diversified as well. So, for me it was about holding onto my cash, and timing when to first enter the market if storms were signaling on the horizon. Go now or wait.

To that end, I began tracking the trigger metrics listed in the linked articles, as well as a few more that I came across from other publications. For reference and discussion purposes, I thought I'd share the current values of these triggers. Most metrics are sourced from FRED.

Update for July 2019...

S&P 500
I am using a 9-month moving average as a proxy for a 200 day moving average. The trigger is when the S&P 500 moves below the 9MMA AND other triggers fire. In July, the S&P 500 was approximately 7.52% above the 9MMA. Today, it is 2.83% above the 9MMA.

Unemployment
According to the articles, the unemployment signal contributed 1.4% return over the B&H strategy. The trigger is when the unemployment rate moves above its trailing twelve month average. In July, the unemployment rate was 1.99% below the TTM average (trending towards trigger). This trigger fired in January 2019.

YoY Job Growth
This trigger contributed 1.2% return over B&H. It fires when the YoY Job Growth % is negative. In July, YoY job growth was 0.80% (trending towards trigger).

YoY Industrial Production Growth
This trigger contributed 1.1% return over B&H. It fires when YoY Industrial Production Growth is negative. In July, YoY Industrial Production Growth was 0.48% (trending towards trigger).

YoY Retail Sales Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Retail Sales Growth is negative. In July, YoY Retail Sales Growth 1.60% (trending away from trigger/steady). This trigger fired in December 2018.

YoY Personal Income Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Personal Income Growth is less than 3%. In June, YoY Personal Income Growth was 3.54% (trending away from trigger).

YoY New Housing Starts
This trigger contributed 0.8% return over B&H. It fires when YoY New Housing Starts is below -10%. In June, YoY New Housing Starts was 5.63%. This trigger is volitile, and fired in February and March 2019.

In no intentional order, below are the additional interesting "early warning signs" I began tracking after reading other articles on recession leading indicators. These were not correlated to B&H strategy returns.

YoY Small Business Hiring Sentiment
This metric supposedly leads unemployment by about four months. It is a survey index measure, so YoY percent change is used, suggesting that a negative result indicates an upcoming hiring downturn. In July, the result was -8.7% (trending negative). It was negative in January, February, March, and June 2019 due to stronger sentiment over the same months in 2018.

New Jobless Claims and Growth from Lowest Point
This is a 4WMA of weekly new jobless claims. This is supposedly a leading indicator of unemployment when it is above 300K. Currently, it 213,750 and holding fairly steady.

Growth in New Jobless Claims is a supposed leading indicator of unemployment when it reaches 69K over its lowest point. The lowest claims in the past two years was 212,000, which just occurred at the end of July 2019. It is currently 1,750 above the two-year low point.

"Temp" Workers Employment
This tracks the YoY change in employment of workers in temporary job positions. The idea is that temporary workers are the first to be laid off in a downturn, sort of the proverbial canary in the coal mine, and triggers when temporary hire growth goes negative. In July, this metric was 0.98% (trending towards trigger).

"Breadwinner" Employment
This is a YoY change in employment for people between 25-54 years old. It triggers when "breadwinner" growth goes negative. In July, this metric was -0.05% (triggered).

Industrial Capacity Utilization
Along with Industrial Production, this metric indicates how full the industrial pipeline is. It is a supposedly leading indicator of recession when it is above 80%, suggesting an overheating economy. In July, Industrial Capacity Utilization was 77.49% and decreasing. In the past two years, it peaked at 79.57% in November 2018.

PMI
This is a survey of manufacturing Purchasing Managers sentiment, based on expectations of new purchase orders for manufacturing supplies, raw materials, etc. Scores above 50 indicate optimism, below 50 indicates pessimism in future ordering.

In essence, this is a leading indicator of the manufacturing supply chain. Supposedly, Taiwan PMI is a six-month leading indicator of USA PMI, and Singapore PMI is a four-month leading indicator as these countries are at the front-end of the global supply chain. When these countries turn pessimistic, it suggests that USA manufacturing will show a downturn in four to six months, with Europe somewhere in between.

Here are the July PMI results for key indicator countries:
  • China (#2 economy) - 49.9 (48.3 six months ago)
  • Japan (#3 economy) - 49.4 (50.3 six months ago)
  • Taiwan - 48.1 (47.5 six months ago)
  • South Korea - 47.3 (48.3 six months ago)
  • Singapore - 49.8 (50.7 six months ago)
  • Germany (#4 economy) - 43.2 (49.7 six months ago)
  • France - 49.7 (51.2 six months ago)
  • Italy - 48.5 (47.8 six months ago)
  • Eurozone (#2 economy) - 46.5 (50.5 six months ago)
  • USA (#1 economy) - 50.4 (56.6 six months ago)
Inflation
Inflation has been fairly steady. YoY CPI in July was 1.81%, down from a high of 2.95% in July 2018. Core Inflation (ex food and energy) is 2.21%, down from 2.35% last July.

Yield Curve
An inverted yield curve sustained for three months supposedly leads a recession by 9-18 months. The 1-month and 3-month yield curves have been inverted since May. The 1-year yield curve inverted in May and again in July. Currently the yields are:
  • 1Mo: -0.37%
  • 3Mo: -0.32%
  • 1Yr: -0.18%
  • 2Yr: +0.02%
  • 5Yr: +0.11%
Bottom Line
Take this FWIW...

The linked articles' leading indicators are currently NOT signaling a retreat, but trending in that direction. Jobless claims are still low, but breadwinners are contracting and temps are close. This is corroborated by small business hiring sentiment, which has been negative from a year ago for the past two months.

Manufacturing is slowing down. Industrial production growth is close to zero and capacity utilization is falling. Asia-Pac manufacturing sentiment is still pessimistic and Europe is still declining. The United States is still optimistic, but USA PMI is borderline.

The big news is the inverted yield curve. The 1Yr yield inverted and the 2Yr yield is practically zero. The short-term yields have been inverted since May.

-B
Time for an update for Oct 2019...

S&P 500
I am using a 9-month moving average as a proxy for a 200 day moving average. The trigger is when the S&P 500 moves below the 9MMA AND other triggers fire. In October, the S&P 500 was approximately 2.83% above the 9MMA. Today, it is 6.97% above the 9MMA.

Unemployment
According to the articles, the unemployment signal contributed 1.4% return over the B&H strategy. The trigger is when the unemployment rate moves above its trailing twelve month average. In October, the unemployment rate was 3.57% below the TTM average (down and up since July). This trigger fired in January 2019.

YoY Job Growth
This trigger contributed 1.2% return over B&H. It fires when the YoY Job Growth % is negative. In October, YoY job growth was 0.37% (close to, but slowly moving away from trigger).

YoY Industrial Production Growth
This trigger contributed 1.1% return over B&H. It fires when YoY Industrial Production Growth is negative. In October, YoY Industrial Production Growth was -1.13% (trigger fired in September at -0.07%).

YoY Retail Sales Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Retail Sales Growth is negative. In October, YoY Retail Sales Growth 1.31% (trending towards trigger). This trigger fired in December 2018.

YoY Personal Income Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Personal Income Growth is less than 3%. In September, YoY Personal Income Growth was 3.52% (trending away from trigger).

YoY New Housing Starts
This trigger contributed 0.8% return over B&H. It fires when YoY New Housing Starts is below -10%. In October, YoY New Housing Starts was 7.40%. This trigger is volitile, and fired in February and March 2019.

In no intentional order, below are the additional interesting "early warning signs" I began tracking after reading other articles on recession leading indicators. These were not correlated to B&H strategy returns.

YoY Small Business Hiring Sentiment
This metric supposedly leads unemployment by about four months. It is a survey index measure, so YoY percent change is used, suggesting that a negative result indicates an upcoming hiring downturn. In October, the result was -18.2% (trending negative). It was -23.1% in August, and -26.1% in September.

New Jobless Claims and Growth from Lowest Point
This is a 4WMA of weekly new jobless claims. This is supposedly a leading indicator of unemployment when it is above 300K. Currently, it 219,750 and slowly increasing.

Growth in New Jobless Claims is a supposed leading indicator of unemployment when it reaches 69K over its lowest point. The lowest claims in the past two years was 212,000, which occurred at the end of July 2019. It is currently 7,750 above the two-year low point.

"Temp" Workers Employment
This tracks the YoY change in employment of workers in temporary job positions. The idea is that temporary workers are the first to be laid off in a downturn, sort of the proverbial canary in the coal mine, and triggers when temporary hire growth goes negative. In October, this metric was -0.15% (triggered).

"Breadwinner" Employment
This is a YoY change in employment for people between 25-54 years old. It triggers when "breadwinner" growth goes negative. In October, this metric was 0.52% (triggered briefly in July).

Industrial Capacity Utilization
Along with Industrial Production, this metric indicates how full the industrial pipeline is. It is a supposedly leading indicator of recession when it is above 80%, suggesting an overheating economy. In October, Industrial Capacity Utilization was 76.74% and decreasing. In the past two years, it peaked at 79.57% in November 2018. This corroborates the negative YoY Industrial Production trigger that is currently fired.

PMI
This is a survey of manufacturing Purchasing Managers sentiment, based on expectations of new purchase orders for manufacturing supplies, raw materials, etc. Scores above 50 indicate optimism, below 50 indicates pessimism in future ordering.

In essence, this is a leading indicator of the manufacturing supply chain. Supposedly, Taiwan PMI is a six-month leading indicator of USA PMI, and Singapore PMI is a four-month leading indicator as these countries are at the front-end of the global supply chain. When these countries turn pessimistic, it suggests that USA manufacturing will show a downturn in four to six months, with Europe somewhere in between.

Here are the October PMI results for key indicator countries:
  • China (#2 economy) - 51.7 (50.2 six months ago)
  • Japan (#3 economy) - 48.4 (50.2 six months ago)
  • Taiwan - 49.8 (48.2 six months ago)
  • South Korea - 48.4 (50.2 six months ago)
  • Singapore - 49.8 (50.7 six months ago)
  • Germany (#4 economy) - 41.9 (44.4 six months ago)
  • France - 50.5 (50.0 six months ago)
  • Italy - 47.7 (49.1 six months ago)
  • Eurozone (#2 economy) - 45.7 (47.9 six months ago)
  • USA (#1 economy) - 51.3 (52.6 six months ago)
Inflation
Inflation has been fairly steady. YoY CPI in October was 1.77%, down from a high of 2.95% in July 2018. Core Inflation (ex food and energy) is 2.31%, down from 2.39% last August.

Yield Curve
An inverted yield curve sustained for three months supposedly leads a recession by 9-18 months. The 1-month, 3-month, and 1-year yield curves have been inverted since May, but reverted in October. Currently the yields are:
  • 1Mo: +0.11%
  • 3Mo: +0.14%
  • 1Yr: +0.15%
  • 2Yr: +0.16%
  • 5Yr: +0.16%
Shiller CAPE10
A new interesting trigger I just started tracking is the Shiller CAPE10. An analyst posted an article at Forbes indicating that in the past 50 years, when the 10-month moving average of the Shiller CAPE10 falls more than 10% from its recent high, a recession is likely within the next 12-24 months. (See Forbes: Another S&P 500 Warning Sign Appears, How To Deal With It). I'm using the Shiller Total Return CAPE10, not his original one.

This indicator triggered in December 2018, January 2019, and August 2019 (-11.63%, -11.31%, -10.00% respectively). In October, this indicator was -9.65%, suggesting that a recession is coming in the next year.

Bottom Line
Take this FWIW...

The linked articles' leading indicators are currently NOT signaling a retreat, but trending in that direction. Jobless claims are slowly rising, breadwinners are contracting growth, and temp hires are reducing. This is corroborated by small business hiring sentiment, which has been negative from a year ago for the past four months. The overall unemployment is holding steady, though.

Manufacturing is slowing down. Industrial production is reducing and capacity utilization is falling. Asia-Pac manufacturing sentiment is still pessimistic, Europe is still declining but bottoming. The United States is still optimistic, USA PMI is borderline but slowing advancing.

The yield curves have corrected since their inversion a few months ago, but they and the Shiller CAPE10 long-term indicator suggests that we're about a year into a two-year recession warning.

Historical Context

For comparison, in advance of the financial crisis of 2008, most of the triggers fired by November 2007 when the S&P 500 was at 1,463. The recent high was the prior month of October 2007 when it was 1,539. In subsequent months:
  • Oct 2007: 1,539
  • Nov 2007: 1,463
  • Jan 2008: 1,378
  • Mar 2008: 1,316
  • May 2008: 1,403
  • Jul 2008: 1,257
  • Sep 2008: 1,216
  • Oct 2008: 968
  • Nov 2008: 883
  • Dec 2008: 877
  • Jan 2009: 865
  • Feb 2009: 805
  • Mar 2009: 757
  • Apr 2009: 848
  • May 2009: 902
  • Jun 2009: 926
The main trigger turned off in June 2009 when the S&P 500 crossed above its 9-month moving average, although all the other indicators were still on. The indicators began clearing by December 2009, with the last of them turning off by June 2010.

By following these triggers, a trend follower would have left the market when the S&P 500 was at 1,463 and reentered the market when the S&P 500 was at 926, avoiding a 37% drop. By December 2011, the S&P 500 had grown back to 1,241 (34% growth).

I should also point out that these indicators whip-sawed in Jun-Aug 2010 (3% loss) and Aug-Dec 2011 (8% loss), likely residual effects of the lowered S&P 500 in the prior year. It whip-sawed again in Aug 2015-Feb 2016 (1% growth), and Dec 2018-Jan 2019 (7% loss). If these whip-saws were ignored, growth from reentering after the financial crisis is 235%.

-B
Last edited by Barsoom on Sat Nov 30, 2019 1:22 am, edited 5 times in total.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Fri Nov 29, 2019 5:00 pm

Barsoom wrote:
Fri Nov 29, 2019 4:45 pm
The linked articles' leading indicators are currently NOT signaling a retreat, but trending in that direction. Jobless claims are slowly rising, breadwinners are contracting growth, and temp hires are reducing. This is corroborated by small business hiring sentiment, which has been negative from a year ago for the past four months. The overall unemployment is holding stead, though.
Thanks for the update. It's always interesting to see others' trend following strategies. Subjectively, I would agree that the indicators I've seen seem to point to the economy continuing to be robust but softening somewhat. I've been pleasantly surprised that stocks have advanced as much as they have the last few months.
“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

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Fri Nov 29, 2019 5:33 pm

willthrill81 wrote:
Fri Nov 29, 2019 5:00 pm
It's always interesting to see others' trend following strategies.
Just to be clear, the trend following strategy I'm showing is the full strategy from the article linked in your OP. It includes all the indicators, not just the S&P500/unemployment pair that I think you're using. I added additional indicators I picked up along the way of my research.
willthrill81 wrote:
Fri Nov 29, 2019 5:00 pm
Subjectively, I would agree that the indicators I've seen seem to point to the economy continuing to be robust but softening somewhat. I've been pleasantly surprised that stocks have advanced as much as they have the last few months.
Me, too.

Looking at the main trigger S&P 500 likelihood of falling below its 9-month moving average, it got a big bump with today's close. In July, it was 7.42% above its 10MMA, in August it was +3.17%, in September it was +4.47%, in October it was +2.83%, and today it is +6.97%. So, the S&P 500 was approaching its 10MMA, which is required for the other triggers to matter.

The other triggers have time horizons, and a cascade effect. For instance, Asia PMI has a 4-6 month lead for impacting USA PMI. USA PMI hasn't fallen below 50 yet, but YoY Industrial Production growth (a primary indicator) went negative beginning in September and Industrial Capacity Utilization is beginning to fall.

Small business hiring sentiment has a 4-6 month lead on unemployment changes. This metric was negative from Jan-Mar, and again from Jul-present. July was also the recent low point in new jobless claims (4-week moving average), which has grown by 7,500 since then (69,000 is the trigger). Overall unemployment against its trailing twelve month moving average is still holding at slightly below (roughly 2%-3% below the TTM). However, movement is showing up in the negative growth in temporary hires and average YoY Job Growth in the past 3 months at +0.26% (another primary indicator, trigger is less than 0%). Main "breadwinner" (ages 25-54) job growth is at risk of going negative very soon.

Next week will be interesting. The November PMI is due out on the first of December, unemployment data is updated on the 7th, and small business hiring comes out on the second Tuesday (Dec 12, which is late). We will soon know if the employment data will hold.

Remember that it was last December-January that whip-sawed, although I suspect some of that was the impact of the Fed raising interest rates against the wishes of many in the lead-up to the 2018 mid-term election. The Fed rapidly retracted the rate increase early in 2019 and the market recovered. I don't see the same forces in play this time around.

-B

rascott
Posts: 2150
Joined: Wed Apr 15, 2015 10:53 am

Re: My trend following strategy and experience

Post by rascott » Sat Nov 30, 2019 12:56 am

Barsoom wrote:
Fri Nov 29, 2019 4:45 pm
[

I should also point out that these indicators whip-sawed in Jun-Aug 2010 (3% loss) and Aug-Dec 2011 (8% loss), likely residual effects of the lowered S&P 500 in the prior year. It whip-sawed again in Aug 2015-Feb 2016 (1% growth), and Dec 2018-Jan 2019 (7% loss). If these whip-saws were ignored, growth from reentering after the financial crisis is 235%.

-B

Good stuff and enjoyed reading it...... that all said..... wouldn't most all of this data be price incorporated? So one could just as easily follow the old moving average? These are all widely reported and digested economic reports that typically drive equity prices.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sat Nov 30, 2019 1:19 am

rascott wrote:
Sat Nov 30, 2019 12:56 am
Barsoom wrote:
Fri Nov 29, 2019 4:45 pm
[

I should also point out that these indicators whip-sawed in Jun-Aug 2010 (3% loss) and Aug-Dec 2011 (8% loss), likely residual effects of the lowered S&P 500 in the prior year. It whip-sawed again in Aug 2015-Feb 2016 (1% growth), and Dec 2018-Jan 2019 (7% loss). If these whip-saws were ignored, growth from reentering after the financial crisis is 235%.

-B

Good stuff and enjoyed reading it...... that all said..... wouldn't most all of this data be price incorporated? So one could just as easily follow the old moving average? These are all widely reported and digested economic reports that typically drive equity prices.
You might want to read the OP and the linked articles for a discussion of why using price alone can be a problem and how using it in conjunction with other signals may be useful.
“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

User avatar
BlueEars
Posts: 3845
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: My trend following strategy and experience

Post by BlueEars » Sat Nov 30, 2019 11:35 am

Barsoom wrote:
Fri Nov 29, 2019 4:45 pm
...

Time for an update for Oct 2019...

S&P 500
I am using a 9-month moving average as a proxy for a 200 day moving average. The trigger is when the S&P 500 moves below the 9MMA AND other triggers fire. In October, the S&P 500 was approximately 2.83% above the 9MMA. Today, it is 6.97% above the 9MMA.

Unemployment
According to the articles, the unemployment signal contributed 1.4% return over the B&H strategy. The trigger is when the unemployment rate moves above its trailing twelve month average. In October, the unemployment rate was 3.57% below the TTM average (down and up since July). This trigger fired in January 2019.

YoY Job Growth
This trigger contributed 1.2% return over B&H. It fires when the YoY Job Growth % is negative. In October, YoY job growth was 0.37% (close to, but slowly moving away from trigger).

YoY Industrial Production Growth
This trigger contributed 1.1% return over B&H. It fires when YoY Industrial Production Growth is negative. In October, YoY Industrial Production Growth was -1.13% (trigger fired in September at -0.07%).

YoY Retail Sales Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Retail Sales Growth is negative. In October, YoY Retail Sales Growth 1.31% (trending towards trigger). This trigger fired in December 2018.

YoY Personal Income Growth
This trigger contributed 0.9% return over B&H. It fires when YoY Personal Income Growth is less than 3%. In September, YoY Personal Income Growth was 3.52% (trending away from trigger).

YoY New Housing Starts
This trigger contributed 0.8% return over B&H. It fires when YoY New Housing Starts is below -10%. In October, YoY New Housing Starts was 7.40%. This trigger is volitile, and fired in February and March 2019.

In no intentional order, below are the additional interesting "early warning signs" I began tracking after reading other articles on recession leading indicators. These were not correlated to B&H strategy returns.

YoY Small Business Hiring Sentiment
This metric supposedly leads unemployment by about four months. It is a survey index measure, so YoY percent change is used, suggesting that a negative result indicates an upcoming hiring downturn. In October, the result was -18.2% (trending negative). It was -23.1% in August, and -26.1% in September.

New Jobless Claims and Growth from Lowest Point
This is a 4WMA of weekly new jobless claims. This is supposedly a leading indicator of unemployment when it is above 300K. Currently, it 219,750 and slowly increasing.

Growth in New Jobless Claims is a supposed leading indicator of unemployment when it reaches 69K over its lowest point. The lowest claims in the past two years was 212,000, which occurred at the end of July 2019. It is currently 7,750 above the two-year low point.

"Temp" Workers Employment
This tracks the YoY change in employment of workers in temporary job positions. The idea is that temporary workers are the first to be laid off in a downturn, sort of the proverbial canary in the coal mine, and triggers when temporary hire growth goes negative. In October, this metric was -0.15% (triggered).

"Breadwinner" Employment
This is a YoY change in employment for people between 25-54 years old. It triggers when "breadwinner" growth goes negative. In October, this metric was 0.52% (triggered briefly in July).

Industrial Capacity Utilization
Along with Industrial Production, this metric indicates how full the industrial pipeline is. It is a supposedly leading indicator of recession when it is above 80%, suggesting an overheating economy. In October, Industrial Capacity Utilization was 76.74% and decreasing. In the past two years, it peaked at 79.57% in November 2018. This corroborates the negative YoY Industrial Production trigger that is currently fired.

PMI
This is a survey of manufacturing Purchasing Managers sentiment, based on expectations of new purchase orders for manufacturing supplies, raw materials, etc. Scores above 50 indicate optimism, below 50 indicates pessimism in future ordering.

In essence, this is a leading indicator of the manufacturing supply chain. Supposedly, Taiwan PMI is a six-month leading indicator of USA PMI, and Singapore PMI is a four-month leading indicator as these countries are at the front-end of the global supply chain. When these countries turn pessimistic, it suggests that USA manufacturing will show a downturn in four to six months, with Europe somewhere in between.

Here are the October PMI results for key indicator countries:
  • China (#2 economy) - 51.7 (50.2 six months ago)
  • Japan (#3 economy) - 48.4 (50.2 six months ago)
  • Taiwan - 49.8 (48.2 six months ago)
  • South Korea - 48.4 (50.2 six months ago)
  • Singapore - 49.8 (50.7 six months ago)
  • Germany (#4 economy) - 41.9 (44.4 six months ago)
  • France - 50.5 (50.0 six months ago)
  • Italy - 47.7 (49.1 six months ago)
  • Eurozone (#2 economy) - 45.7 (47.9 six months ago)
  • USA (#1 economy) - 51.3 (52.6 six months ago)
Inflation
Inflation has been fairly steady. YoY CPI in October was 1.77%, down from a high of 2.95% in July 2018. Core Inflation (ex food and energy) is 2.31%, down from 2.39% last August.

Yield Curve
An inverted yield curve sustained for three months supposedly leads a recession by 9-18 months. The 1-month, 3-month, and 1-year yield curves have been inverted since May, but reverted in October. Currently the yields are:
  • 1Mo: +0.11%
  • 3Mo: +0.14%
  • 1Yr: +0.15%
  • 2Yr: +0.16%
  • 5Yr: +0.16%
Shiller CAPE10
A new interesting trigger I just started tracking is the Shiller CAPE10. An analyst posted an article at Forbes indicating that in the past 50 years, when the 10-month moving average of the Shiller CAPE10 falls more than 10% from its recent high, a recession is likely within the next 12-24 months. (See Forbes: Another S&P 500 Warning Sign Appears, How To Deal With It). I'm using the Shiller Total Return CAPE10, not his original one.

This indicator triggered in December 2018, January 2019, and August 2019 (-11.63%, -11.31%, -10.00% respectively). In October, this indicator was -9.65%, suggesting that a recession is coming in the next year.

Bottom Line
Take this FWIW...

The linked articles' leading indicators are currently NOT signaling a retreat, but trending in that direction. Jobless claims are slowly rising, breadwinners are contracting growth, and temp hires are reducing. This is corroborated by small business hiring sentiment, which has been negative from a year ago for the past four months. The overall unemployment is holding steady, though.

Manufacturing is slowing down. Industrial production is reducing and capacity utilization is falling. Asia-Pac manufacturing sentiment is still pessimistic, Europe is still declining but bottoming. The United States is still optimistic, USA PMI is borderline but slowing advancing.

The yield curves have corrected since their inversion a few months ago, but they and the Shiller CAPE10 long-term indicator suggests that we're about a year into a two-year recession warning.

Historical Context

....
Barsoom, interesting list you have there. I wonder how many of the above have available monthly data that cover several recessions. Generally I don't use economic data indicators that don't go back to about 1950. I have rejected some data series that appear to be less robust or not useful including some mentioned in that long ago Philosophical Economics article. The ones I think I have robust data for and appear to be fairly coincident with major declines are (1) yield curve, (2) unemployment, and (3) CAPE. And then there is the SP500 plus dividends.

How far back can you go with the other economic data you mentioned above?

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sat Nov 30, 2019 1:57 pm

BlueEars wrote:
Sat Nov 30, 2019 11:35 am
Barsoom, interesting list you have there. I wonder how many of the above have available monthly data that cover several recessions. Generally I don't use economic data indicators that don't go back to about 1950. I have rejected some data series that appear to be less robust or not useful including some mentioned in that long ago Philosophical Economics article. The ones I think I have robust data for and appear to be fairly coincident with major declines are (1) yield curve, (2) unemployment, and (3) CAPE. And then there is the SP500 plus dividends.

How far back can you go with the other economic data you mentioned above?
Almost all of my data is from FRED and goes back to various start dates. I run the FRED report and download the dataset as an Excel sheet. Only a few dataset are manually entered each month.

For reference and transparency, here are my sources (Link, Frequency & date available, Earliest Data):
-B

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sat Nov 30, 2019 2:54 pm

rascott wrote:
Sat Nov 30, 2019 12:56 am
Good stuff and enjoyed reading it...... that all said..... wouldn't most all of this data be price incorporated? So one could just as easily follow the old moving average? These are all widely reported and digested economic reports that typically drive equity prices.
I'm a novice at this and don't have deep knowledge... that all said...

I think that people act on imperfect information, emotion, recency bias and optimism bias, slowly acclimate to the current state, and begin to take on more and more risk, calling it the "new normal."

Price, at the S&P 500 index level, is an aggregate; individual stock price is set by the respective company's current market conditions and company health. Many of these leading indicators are cross-industry/market segments, so I think are somewhat removed from actual product price set by actual companies.

Regarding the 2008 financial crisis and subsequent whip-saws...

The OP article indicators began triggering at different dates prior to the correction. Remember, the rule is that S&P 500 moves below its 9-month moving average AND one or more of the below triggers fires:
  • S&P 500 triggered in November 2007 (main switch)
  • Unemployment triggered in June 2007
  • Job Growth triggered briefly in November 2006 and steadily in March 2008
  • Industrial Production Growth triggered in April 2008
  • Retail Sales Growth triggered in December 2007
  • Personal Income Growth triggered in September 2007
  • New Housing Starts triggered in June 2006
My additional "secondary" (or confirming triggers) triggered (where I have data) at these dates:
  • Breadwinner Employment Growth triggered steadily in January 2008
  • Temp Worker Employment Growth triggered in April 2007
  • Industrial Capacity Utilization triggered in February 2005 (capacity was running at 80%-81% and turned off in May 2008 after Industrial Production Growth declined)
Personally, I wouldn't act on just one of the above triggers firing; I'd wait for a preponderance of the evidence to indicate a recession. Fortunately, in 2008 all of the primary triggers fired before the S&P 500 trigger did.

For the whip-saws...

The Jun-Aug 2010 whip-saw was triggered by Personal Income Growth. However, this indicator only fired for one month. My secondary indicator of Breadwinner employment was also triggered, keeping this active until the S&P 500 re-crossed its 9-month moving average in September 2010. Perhaps this shouldn't be counted as a whip-saw because it was kept going by an indicator not included in the OP article?

The Aug-Dec 2011 whip-saw was also triggered by Personal Income Growth (which remained active until March 2012). Breadwinner employment and New Jobless Claims were also triggered, but were parallel to Personal Income Growth.

The Aug 2015-Feb 2016 whip-saw was triggered by Industrial Production Growth. This indicator fired in April 2015, but the S&P 500 didn't cross its 9-month moving average until August 2015. The S&P 500 re-crossed its 9MMA in March 2016, although the Industrial Production Growth indicator didn't clear until November 2016.

The Dec 2018-Jan 2019 whip-saw was actually triggered earlier, but I turned off some of the more volatile triggers. The S&P 500 actually crossed its 9-month moving average in November 2018. At that time, the Personal Income Growth indicator triggered for only November. In December, the Retail Sales Growth triggered for just that month. The Unemployment indicator also triggered for just January, before going back under its trailing 12 month average again in February. New Housing Starts triggered in March and April, but the S&P 500 re-crossed its 9MMA in February, nullifying these triggers.

As you can see from the whip-saws, this is why I would look for a combination (majority?) of indicators to trigger before making a move.

-B

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sun Dec 01, 2019 11:06 am

Barsoom wrote:
Sat Nov 30, 2019 2:54 pm
For the whip-saws...

The Jun-Aug 2010 whip-saw was triggered by Personal Income Growth. However, this indicator only fired for one month. My secondary indicator of Breadwinner employment was also triggered, keeping this active until the S&P 500 re-crossed its 9-month moving average in September 2010. Perhaps this shouldn't be counted as a whip-saw because it was kept going by an indicator not included in the OP article?

The Aug-Dec 2011 whip-saw was also triggered by Personal Income Growth (which remained active until March 2012). Breadwinner employment and New Jobless Claims were also triggered, but were parallel to Personal Income Growth.

The Aug 2015-Feb 2016 whip-saw was triggered by Industrial Production Growth. This indicator fired in April 2015, but the S&P 500 didn't cross its 9-month moving average until August 2015. The S&P 500 re-crossed its 9MMA in March 2016, although the Industrial Production Growth indicator didn't clear until November 2016.

The Dec 2018-Jan 2019 whip-saw was actually triggered earlier, but I turned off some of the more volatile triggers. The S&P 500 actually crossed its 9-month moving average in November 2018. At that time, the Personal Income Growth indicator triggered for only November. In December, the Retail Sales Growth triggered for just that month. The Unemployment indicator also triggered for just January, before going back under its trailing 12 month average again in February. New Housing Starts triggered in March and April, but the S&P 500 re-crossed its 9MMA in February, nullifying these triggers.

As you can see from the whip-saws, this is why I would look for a combination (majority?) of indicators to trigger before making a move.
Whipsaws, which are an inevitable aspect of any trend following strategy, are part of what really puts off a lot of folks from adopting such a strategy, but this seems to be more emotionally than empirically driven. It's true that whipsaws reduce returns compared to perfect timing, but that's impossible. The alternative, holding a significant fixed income allocation, has historically reduced returns as well, but this aspect seems to be overlooked by its adherents. In the end, both holding fixed income and trend following, at least as most people seem to implement them, are essentially risk mitigation strategies. The former tends to result in frequent opportunity costs (i.e. if stocks outperform fixed income, which they generally have), while the latter tends to result in less frequent but potentially bigger opportunity costs (i.e. whipsaws). However, this isn't to say that both strategies have had equal long-term performance because they clearly have not.

The issues of specificity (i.e. the percentage of actual negatives identified as such) and sensitivity (i.e. the percentage of actual positives identified as such) are relevant to this discussion. While I'm sure that you're familiar with these terms after reading the PE posts, those unfamiliar with these terms should be aware of them. The first PE post linked in the OP says this:
To use an example, suppose that there are 100 recessionary months in a given data set. In 86 of those months, a recessionary indicator comes back positive, correctly indicating the recession. The indicator’s sensitivity to recession would then be 86 / 100 = 86%.

Alternatively, suppose that there are 700 non-recessionary months in a given data set. In 400 of those non-recessionary months, a recessionary indicator comes back negative, correctly indicating no recession. The indicator’s specificity to recession would then be 400 / 700 = 57%.
If you will only move out of stocks if both (1) stocks are below their 9 MMA and (2) at least two of the indicators (I'm not sure if this is the case or if you want a majority of the indicators to be 'firing' before you attend to stocks MMA) you follow are 'firing', then your strategy may have a high degree of sensitivity, which is desirable in this situation as the need for specificity is less since you require that stocks be below their 9 MMA before you move out of them (i.e. the 9 MMA is helping to protect you against false negatives). Have you examined this across other bear markets in the past?
“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

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sun Dec 01, 2019 2:11 pm

willthrill81 wrote:
Sun Dec 01, 2019 11:06 am
If you will only move out of stocks if both (1) stocks are below their 9 MMA and (2) at least two of the indicators (I'm not sure if this is the case or if you want a majority of the indicators to be 'firing' before you attend to stocks MMA) you follow are 'firing', then your strategy may have a high degree of sensitivity, which is desirable in this situation as the need for specificity is less since you require that stocks be below their 9 MMA before you move out of them (i.e. the 9 MMA is helping to protect you against false negatives). Have you examined this across other bear markets in the past?
The article in the OP did a sensitivity analysis of the indicators. In his case, he measured the overall return by selling stocks when a single indicator triggered vs the return from an equivalent B&H strategy. He identified the equivalent B&H strategy as an asset allocation of stocks/bonds equal to the percent of time the indicator kept the investor in/out of the market.

The example would be this: if the indicator kept the investor in the market 86% of the time (and out of the market 14% of the time), then the B&H portfolio would be 86% stock/14% bonds. The difference in return between the trend following strategy and this hypothetical B&H portfolio is the extra return attributed to following that single indicator alone. This analysis was done for each indicator independently, and the results are included in my description of each of the article's indicators (sorted descending by contribution).

The two lowest indicators (Personal Income Growth and New Housing Starts) seem volatile, and I wouldn't act on these indicators triggering alone. The highest value indicator (Unemployment) is the one that you are using. I fear that even using just the highest trigger alone is susceptible to whip-saws. Fortunately, in 2008 all the indicators triggered before the S&P 500 triggered.

In the case of the 2000 recession...

The S&P 500 crossed its 9MMA in Oct 2000, signaling an exit 1,390. The Unemployment indicator (1.4% contribution to value) didn't trigger until Jan 2001 at 1,335 (-0.4% from October). The Job Growth indicator (1.2% contribution) began triggering in Feb 2000, the Industrial Production indicator (1.1% contribution) began triggering in Feb 2001, the Retail Sales Growth indicator (0.9% contribution) began triggering in Dec 2000, the Personal Income Growth indicator (0.9% contribution) didn't trigger until May 2001, and the New Housing Starts indicator (0.8% contribution) did not trigger in a sustained way at all.

The strategy signaled re-entering the market in Apr 2003 when the S&P 500 re-crossed its 9MMA at 890 (-36% from exiting in Oct 2000). The Unemployment indicator didn't clear until Oct 2003, the Job Growth indicator cleared in Mar 2004, the Industrial Production trigger cleared in Jun 2002, Retail Sales Growth cleared in Oct 2001, and Personal Income Growth cleared in Feb 2004.

A whip-saw occurred from Apr 1994-Jul 1994, triggered by Personal Income Growth, for no real loss.

A whip-saw occurred from Oct 1987-May 1988, triggered by Personal Income Growth and New Housing Starts.

1981-1982 Recession...

The S&P 500 crossed below its 9MMA in May 1981 at 131. The Unemployment indicator had been triggering since Aug 1979, signaling a brief whip-saw in Mar-Apr 1980. The Job Growth indicator also began triggering in Dec 1979. Industrial Production Growth began triggering in Mar 1980, Retail Sales Growth began triggering steadily in Oct 1979, Personal Income Growth in Apr 1979, and New Housing Starts in Feb 1979. The triggers were waiting for the S&P 500 in this recession.

The S&P 500 re-crossed its 9MMA in Sep 1982 at 122 for a 7% drop. There was a 12% gain in just Aug-Sep 1982 that was missed by waiting to re-enter in September. Re-entering in August would avoided a loss of 20%. The Unemployment indicator cleared in May 1983. The Job Growth indicator cleared in Sep 1983. The Industrial Production Growth indicator cleared in May 1983. The Retail Sales Growth indicator cleared in Nov 1982. The Personal Income Growth indicator cleared in Oct 1983. The New Housing Starts indicator cleared in Jul 1982.

1980 Recession...

Wikipedia says a recession occurred from Jan-Jul 1980. All of the indicators were triggered during this period, but the S&P 500 only crossed its 9MMA for the two months of Mar-Apr 1980.

1973 Recession...

The S&P 500 crossed below its 9MMA in Mar 1973, but no indicators had triggered. New Housing Starts triggered for one month in Jun 1973, and then steadily in Aug 1973. The S&P 500 briefly re-crossed its 9MMA in Oct 1973, but crossed again in Nov 1973 until Feb 1975. Let's call the strategy start of this recession as Nov 1973 at 102.

The Unemployment indicator didn't trigger until Dec 1973. The Job Growth indicator triggered in Jan 1974. Industrial Production Growth didn't trigger until Sep 1974. Retail Sales Growth and Personal Income Growth both triggered in Dec 1973.

The S&P 500 re-crossed its 9MMA for good in Feb 1975 at 80 (avoiding a 22% drop). All the indicators were still triggered. Unemployment cleared in Dec 1975. Job Growth cleared in Feb 1976. Industrial Production Growth cleared in Dec 1975. Retail Sales Growth cleared in Sep 1975. Personal Income Growth cleared in Jan 1976. New Housing Starts cleared in Jul 1975.

Let me know if you want more details or have more questions.

-B

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sun Dec 01, 2019 2:30 pm

Barsoom wrote:
Sun Dec 01, 2019 2:11 pm
Let me know if you want more details or have more questions.
Thanks for the added information. One more thing: have you compared the long-term returns of this strategy to just using the S&P plus UER combination that PE did?
“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

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sun Dec 01, 2019 2:50 pm

willthrill81 wrote:
Sun Dec 01, 2019 2:30 pm
Barsoom wrote:
Sun Dec 01, 2019 2:11 pm
Let me know if you want more details or have more questions.
Thanks for the added information. One more thing: have you compared the long-term returns of this strategy to just using the S&P plus UER combination that PE did?
I think I see the disconnect. Your OP linked articles is missing the middle article in the series. This is the one where the indicators are each discussed (it is referenced in the second article you linked). See this article (second half of it): Growth and Trend: A Simple, Powerful Technique for Timing the Stock Market

I haven't run any comparisons or backtests, other than observing when the indicators trigger or not. My spreadsheet is doing an OR on the indicators with an AND for the S&P 500. Of course, I can alter this to see how the strategy would have reacted to just triggering on S&P 500 AND Unemployment, but that wasn't my purpose.

When I have time later, I will look at how the strategy would have behaved during the previous recessions if using just the Employment trigger.

-B

Hydromod
Posts: 407
Joined: Tue Mar 26, 2019 10:21 pm

Re: My trend following strategy and experience

Post by Hydromod » Sun Dec 01, 2019 3:03 pm

Y'all might want to take a look at some related backtesting, if you haven't already.

Allocate smartly: backtest of UE only

Investing for a living: backtest with UE and additional economic indicators

I find these two sites to have interesting reads.

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sun Dec 01, 2019 3:52 pm

Hydromod wrote:
Sun Dec 01, 2019 3:03 pm
Y'all might want to take a look at some related backtesting, if you haven't already.

Allocate smartly: backtest of UE only

Investing for a living: backtest with UE and additional economic indicators

I find these two sites to have interesting reads.
Thanks for sharing those.

While this doesn't relate specifically to your posts, the empirical argument for trend following is very compelling to me. As such, I find it very interesting that many believe that publicly available information shouldn't be useful for trend following systems, even though it clearly has. (And that's not merely with 'the advantage of hindsight' as investors were using indicators like the 200 DMA going back at least to the 1950s and probably well before then; trend following in the world of investing is far older than stocks.) This ignores the fact that there are literally infinite ways to combine and/or utilize that information, meaning that it's literally impossible for every potential way to be used at all, much less for every potential way to meaningfully impact the market. Add to that the fact that all market participants have different goals, investment horizons, biases, perceptions, etc., and the idea that information simply being publicly available destroys its usefulness flies out the window at a very high rate of speed, IMHO.

Note that I am not saying that this means that it's easy to 'beat' (i.e. earn outsized absolute returns) the market, which is obviously false. The seeming randomness of the world alone, combined with many other factors, makes it very challenging to sort between what's actionable and what isn't. In this context, no trend following system will ever have 'perfect' performance going forward. But, as the saying goes, perfect is the enemy of good.
“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

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sun Dec 01, 2019 4:48 pm

willthrill81 wrote:
Sun Dec 01, 2019 3:52 pm
Note that I am not saying that this means that it's easy to 'beat' (i.e. earn outsized absolute returns) the market, which is obviously false. The seeming randomness of the world alone, combined with many other factors, makes it very challenging to sort between what's actionable and what isn't. In this context, no trend following system will ever have 'perfect' performance going forward. But, as the saying goes, perfect is the enemy of good.
Reviewing the few recent recessions in my earlier post, it was only the 1973 recession where the S&P 500 indicator outpaced the other indicators. In all the following recessions, almost all of the indicators were triggered before the S&P 500 itself fell below its 200 DMA (or 9MMA).

Sure, this is only a small subset of recessions, and this critique is susceptible to claims of recency bias. Still, if the choice is having no predictors to use at all, or observing a majority of predictors triggering and then waiting for the S&P 500 to show downward movement, I'd prefer the latter.

I'm not inclined to jump just because the S&P 500 fell in a vacuum. If the S&P 500 dips below its 9MMA without a majority of primary indicators not already having triggered, I'd likely ignore it until more lights turn red. If it's a real recession, this approach would still exit fairly near the peak instead of closer to the trough, in my opinion.

I wouldn't necessarily follow this thinking in the reverse, however. Once the S&P 500 shows renewed upward movement and re-crosses its 9MMA, I'd want to see the indicators also reversing (if not cleared). I wouldn't wait for the triggers to totally clear if the 9MMA was re-crossed, but I would want to see those indicator measure showing a softening trend to believe that the bottom was reached and not just a temporary plateau.

-B

User avatar
BlueEars
Posts: 3845
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: My trend following strategy and experience

Post by BlueEars » Sun Dec 01, 2019 5:02 pm

Regarding using multiple economic triggers with SP500 moving averages, I think it is best to have preset rules for a sell and for a buy back. At least that is how I do things. I would think emotions might factor too strongly into events if one waits and makes decisions based on economic data flow without setting conditions ahead of time. But yes, conditions can change and models can be improved.

Also thanks Barsoom for the links and data summary. FWIW, I tried adding the retail sales and also the breadwinner unemployment data sets to my model but these did not seem to give better results then what I already use (as mentioned above).

User avatar
Topic Author
willthrill81
Posts: 19532
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: My trend following strategy and experience

Post by willthrill81 » Sun Dec 01, 2019 5:16 pm

BlueEars wrote:
Sun Dec 01, 2019 5:02 pm
Regarding using multiple economic triggers with SP500 moving averages, I think it is best to have preset rules for a sell and for a buy back. At least that is how I do things. I would think emotions might factor too strongly into events if one waits and makes decisions based on economic data flow without setting conditions ahead of time.
I share your sentiment. The only way I would ever employ a trend following system was if it had hard-and-fast rules for what to do and when. The moment that subjectivity enters the equation, I was afraid that my biases, emotions, etc. would get the better of me.
“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

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Sun Dec 01, 2019 7:07 pm

BlueEars wrote:
Sun Dec 01, 2019 5:02 pm
Regarding using multiple economic triggers with SP500 moving averages, I think it is best to have preset rules for a sell and for a buy back. At least that is how I do things. I would think emotions might factor too strongly into events if one waits and makes decisions based on economic data flow without setting conditions ahead of time. But yes, conditions can change and models can be improved.
Fortunately, since 1980 most of the economic indicators already triggered when the S&P 500 indicator finally triggered.

As I said, I'd be watching for a steady economic deterioration to be in progress, confirmed by the downward movement in the S&P 500. It's when the S&P 500 then crosses the 9MMA that I would act. A sudden drop in the S&P 500 that is not supported by multiple surrounding economic factors would be suspicious to me.
BlueEars wrote:
Sun Dec 01, 2019 5:02 pm
Also thanks Barsoom for the links and data summary. FWIW, I tried adding the retail sales and also the breadwinner unemployment data sets to my model but these did not seem to give better results then what I already use (as mentioned above).
You're welcome.

I posted earlier that those extra indicators are more of a cascade effect of longer-term leading indicators and not primary triggers (except Retail Sales, which is a primary trigger in the OP article). The intent was to provide additional economical context to confirm a recession as I described above.

For example, a reduction in Breadwinner Employment Growth would be expected to follow a reduction in Temporary Hire Employment Growth (temps are let go first). Both of these should lag Small Business Hiring Sentiment. So the cascade might look like this:
  1. Drop in Small Business Hiring Sentiment (leads unemployment impacts by about four months)
  2. Drop in Temporary Hire Growth
  3. Drop in Breadwinner Employment Growth
  4. Increase in New Jobless Claims (leading indicator if above 300,000)
  5. Increase in New Unemployment Claims from Lowest Point (leading indicator if above 69,000)
  6. Increase in Unemployment Rate (primary trigger)
Another cascade would be manufacturing based. Follow the semiconductor supply chain from front to back:
  1. China (#2 economy) Purchasing Manager Index (PMI) sentiment (below 50 triggers business pessimism)
  2. Japan (#3 economy) PMI (follows China)
  3. China supply chain satellites
    1. Taiwan PMI (leads USA PMI by about six months)
    2. South Korea PMI
    3. Singapore PMI (leads USA PMI by about four months)
  4. Eurozone (#2 economy, tied) PMI (below 50 shows Asia-Pac supply chain pessimism reaches Europe)
    1. Germany (#4 economy) dominates Europe manufacturing
    2. France PMI
    3. Italy PMI
  5. USA PMI (#1 economy)
  6. Reduction in Industrial Production Growth (primary indicator)
Rather than be surprised by a whip-saw or sudden drop in the S&P 500, my intent is to use the secondary leading indicators to raise awareness and provide an economic backdrop that I expect to eventually show up in the primary economic indicators used in the OP strategy. When those primary indicators trigger and the S&P 500 crosses below its 9MMA, I would have a higher degree of confidence that a true recession is in progress.

-B

Barsoom
Posts: 323
Joined: Thu Dec 06, 2018 9:40 am

Re: My trend following strategy and experience

Post by Barsoom » Mon Dec 02, 2019 2:22 pm

I finally added the last of the OP article indicators, Corporate Earnings Growth (Total Return EPS). The OP article gives it a 0.8% contribution to strategy return. The indicator triggers when TR EPS annualized growth rate is negative.

Currently, this metric is 4.2% and rapidly falling, based on Shiller data as of Sep 2019. Data for the past 12 months:
  • Oct-2018: 22.62%
  • Nov-2018: 22.61%
  • Dec-2018: 22.50%
  • Jan-2019: 21.50%
  • Feb-2019: 20.16%
  • Mar-2019: 18.41%
  • Apr-2019: 16.15%
  • May-2019: 14.35%
  • Jun-2019: 12.55%
  • Jul-2019: 9.50%
  • Aug-2019: 6.81%
  • Sep-2019: 4.20%
Straight-line forecasting the next three months from the prior nine months:
  • Oct-2019: 2.78%
  • Nov-2019: 0.38%
  • Dec-2019: -2.8% (triggered)
Industrial Production Growth is the only indicator that is currently triggered. The S&P 500 is about 6% above its 9MMA, so the strategy is not close to triggering at this time.

Prior recession behavior...

The 2000 recession was triggered in Oct 2000 by the Job Growth indicator, which triggered in Feb 2000. The TR EPS Growth indicator triggered at the same time as the Unemployment Growth indicator in Jan 2001. Industrial Production Growth and Retail Sales Growth both triggered in Feb-2001.

The indicator cleared in Sep 2002, four months after the Industrial Production Growth indicator. The strategy reentered the market in Apr 2003.

The 1981-1982 recession was triggered in May 1981 by the Unemployment indicator, which triggered in Aug 1979. The TR EPS Growth indicator triggered in May 1980. The Industrial Growth Rate indicator triggered in Mar 1980. All the other indicators were already triggered.

The indicator was the last to clear in Oct 1983. The Unemployment indicator cleared in May 1983, the Industrial Production indicator cleared one month earlier. The Retail Sales Growth indicator cleared in Nov 1982. The strategy reentered the market in Sep 1982.

For the 1980 recession, the S&P 500 only crossed its 9MMA for the two months of Mar-Apr 1980. The TR EPS Growth indicator didn't trigger until May 1980, but remained triggered throughout the 1981-1982 recession.

The 1973 recession was triggered in Nov 1973 by the New Housing Start indicator, which triggered in Aug 1973. The TR EPS Growth indicator didn't trigger until Dec 1974. The strategy reentered the market in Feb 1975. The TR EPS Growth indicator essentially missed the 1973 recession.

-B

Locked