Larry Swedroe: Trend Following As Insurance

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Larry Swedroe: Trend Following As Insurance

Post by Random Walker »

https://www.etf.com/sections/index-inve ... nopaging=1

Trend following, or time series momentum, is the tendency for an asset’s recent price trend to continue in the near future. Trend is a changing asset allocation strategy since one is either buying or selling the asset class depending on recent performance. Trend strongly meets all 5 of Larry’s criteria for an investment worthy of consideration. Significantly, it is present across asset classes. Larry reviews a several papers on trend. Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities. Trend can really be of benefit in extended drawdowns, and Larry reviews data that goes a little further back to include 2008 to show the benefit of trend. Trend can have a big effect on lessening maximal drawdowns. TS Momentum can significantly protect against sequence of returns risk in retirement and improve withdrawal rates. Trend following requires one to be immune from tracking error regret and to appreciate that alternative histories can always play out. At the end Larry makes reference to the AQR managed futures fund. It’s also worthwhile to note that AQR incorporates TS Momentum into one of its multi style funds.

Dave
sixtyforty
Posts: 656
Joined: Tue Nov 25, 2014 11:22 am
Location: USA

Re: Larry Swedroe: Trend Following As Insurance

Post by sixtyforty »

Random Walker wrote: Mon May 13, 2019 8:13 am https://www.etf.com/sections/index-inve ... nopaging=1

Trend following, or time series momentum, is the tendency for an asset’s recent price trend to continue in the near future. Trend is a changing asset allocation strategy since one is either buying or selling the asset class depending on recent performance. Trend strongly meets all 5 of Larry’s criteria for an investment worthy of consideration. Significantly, it is present across asset classes. Larry reviews a several papers on trend. Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities. Trend can really be of benefit in extended drawdowns, and Larry reviews data that goes a little further back to include 2008 to show the benefit of trend. Trend can have a big effect on lessening maximal drawdowns. TS Momentum can significantly protect against sequence of returns risk in retirement and improve withdrawal rates. Trend following requires one to be immune from tracking error regret and to appreciate that alternative histories can always play out. At the end Larry makes reference to the AQR managed futures fund. It’s also worthwhile to note that AQR incorporates TS Momentum into one of its multi style funds.

Dave
I like the term 'insurance'. I think it's appropriate term for some of us following a trend strategy.
"Simplicity is the ultimate sophistication" - Leonardo Da Vinci
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by Random Walker »

sixtyforty wrote: Mon May 13, 2019 10:14 am
Random Walker wrote: Mon May 13, 2019 8:13 am https://www.etf.com/sections/index-inve ... nopaging=1

Trend following, or time series momentum, is the tendency for an asset’s recent price trend to continue in the near future. Trend is a changing asset allocation strategy since one is either buying or selling the asset class depending on recent performance. Trend strongly meets all 5 of Larry’s criteria for an investment worthy of consideration. Significantly, it is present across asset classes. Larry reviews a several papers on trend. Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities. Trend can really be of benefit in extended drawdowns, and Larry reviews data that goes a little further back to include 2008 to show the benefit of trend. Trend can have a big effect on lessening maximal drawdowns. TS Momentum can significantly protect against sequence of returns risk in retirement and improve withdrawal rates. Trend following requires one to be immune from tracking error regret and to appreciate that alternative histories can always play out. At the end Larry makes reference to the AQR managed futures fund. It’s also worthwhile to note that AQR incorporates TS Momentum into one of its multi style funds.

Dave
I like the term 'insurance'. I think it's appropriate term for some of us following a trend strategy.
Have you seen the AQR paper on 100 years trend following with the smile curve? That really emphasizes the insurance nature to me. I was a little surprised this paper didn’t come up in the article.
I’m a big believer in efficient markets, so I tend to have relatively more trouble swallowing the behavioral anomalies, but from Larry’s book the t-stats for TS Momentum are some of the biggest of all the factors.

https://www.aqr.com/-/media/files/paper ... esting.pdf

Dave
User avatar
arcticpineapplecorp.
Posts: 15080
Joined: Tue Mar 06, 2012 8:22 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by arcticpineapplecorp. »

Investors looking to protect themselves from severe drawdowns (such as retirees, for whom sequence risk should be a paramount concern) should consider including an allocation to a diversified trend-following strategy, usually in the form of a managed futures fund (such as AQR Managed Futures Strategy Fund), as it has historically provided insurance against tail risks.
So how nuch of an allocation? How much insurance?

With the 50/40/10 (50 stocks/40 bonds and 10 trend)
He found that the diversified managed futures strategy provided diversification benefits (adding a unique source of risk and return) that has historically supported a 0.8% higher safe withdrawal rate.
So instead of a 100%-95% success rate of 4% withdrawals over 30 years:
Miller examined the period 1926 through 2012. He found that, historically, a 4% initial withdrawal rate from a portfolio of 50% stocks and 50% bonds with a 30-year horizon had a failure rate ranging between 0% and 5%, depending on the bond index used.
I can increase my withdrawals rate from 4% to 4.8% thanks to trend?
It's hard to accept the truth when the lies were exactly what you wanted to hear. Investing is simple, but not easy. Buy, hold & rebalance low cost index funds & manage taxable events. Asking Portfolio Questions | Wiki
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

arcticpineapplecorp. wrote: Mon May 13, 2019 11:20 am I can increase my withdrawals rate from 4% to 4.8% thanks to trend?
I would look at it this way. If I can data-mine the few worst cases away using past data, the safe withdrawal rate is now 4.8% because I data-mined those worst cases away.

The insurance is the belief that data mining away a very small number of scenarios based on a few factors will produce similar results for a small number of future scenarios.

Statistically, I would say this is unlikely, but others obviously disagree.

Take your pick of beliefs.
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: Larry Swedroe: Trend Following As Insurance

Post by Beliavsky »

Swedroe writes a lot, and I wonder how to reconcile all that he writes.

Google "swedroe market timing" and you get many articles by Swedroe criticizing the practice. Trend following is a market timing strategy. I agree with Swedroe that it is worth considering, and I have backtested such strategies.

If you believe that markets are not completely efficient and have some predictability based on trend, why can't they also be predicted based on other indicators, such as CAPE for equities, carry and real interest rates for bonds, and carry and purchasing power parity for currencies? Isn't it as plausible to invest in asset classes based on valuation as it is to use trend? Swedroe cites academic research supporting trend-following. There is also lots of research on the approaches I listed.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

sixtyforty wrote: Mon May 13, 2019 10:14 amI like the term 'insurance'. I think it's appropriate term for some of us following a trend strategy.
I agree. You should expect to pay 'premiums' in the form of whipsaws on a fairly regular basis with most trend following strategies, and you make a 'claim' during bear markets. Just like with insurance, you shouldn't expect to outperform the market in terms of absolute returns over the long-term. Over the last 30 years, for instance, the 200 DMA strategy with cash as the out-of-market asset had remarkably similar returns to buy-and-hold, but the maximum drawdown was about two-thirds smaller (about -17% instead of -51%). But you would have underperformed buy-and-hold to the tune of about 5% annually for the 1990s and the 2010s. If you aren't confident that you could stick with such a strategy for a decade or longer of underperformance relative to buy-and-hold, then you probably shouldn't adopt it.
Random Walker wrote: Mon May 13, 2019 10:34 amI’m a big believer in efficient markets, so I tend to have relatively more trouble swallowing the behavioral anomalies, but from Larry’s book the t-stats for TS Momentum are some of the biggest of all the factors.
I'm not necessarily convinced that TS momentum is entirely incompatible with the concept of efficient markets. But frankly, I don't think that markets are as efficient as many believe them to be. For instance, publishing a factor's historic outperformance of the market shouldn't reduce that factor's premium going forward, but it consistently has.

Beyond that, if the data we are confronted with is incompatible with our hypothesis, we may need to revise our hypothesis. We might even need to take more drastic measures, as indicated below.

"When you discover that you are riding a dead horse, the best strategy is to dismount."
-Native American proverb
The Sensible Steward
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by Random Walker »

Beliavsky wrote: Mon May 13, 2019 5:07 pm Swedroe writes a lot, and I wonder how to reconcile all that he writes.

I think everything he writes is internally consistent. The underlying themes include nearly efficient markets, modern portfolio theory, and investments that improve the characteristics of the portfolio as a whole. The foundation for all of this is strong academic evidence.

Google "swedroe market timing" and you get many articles by Swedroe criticizing the practice. Trend following is a market timing strategy. I agree with Swedroe that it is worth considering, and I have backtested such strategies.

If you believe that markets are not completely efficient and have some predictability based on trend, why can't they also be predicted based on other indicators, such as CAPE for equities, carry and real interest rates for bonds, and carry and purchasing power parity for currencies? Isn't it as plausible to invest in asset classes based on valuation as it is to use trend? Swedroe cites academic research supporting trend-following. There is also lots of research on the approaches I listed.
Two of the AQR funds recommended by Larry incorporate much of what you discuss.
QSPIX invests in 4 styles across 4 asset classes. The styles are value, CS Momentum, defensive, carry. The asset classes are equities, bonds, commodities, currencies.
QRPRX (which is better for taxable accounts) invests in 6 styles across 3 asset classes. The styles are value, CS momentum, defensive, carry, trend, variance risk premium. The asset classes are equities, bonds, currencies.

https://www.aqr.com/~/media/files/paper ... esting.pdf

Dave
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: Larry Swedroe: Trend Following As Insurance

Post by stlutz »

willthrill81 wrote: Mon May 13, 2019 6:27 pm If you aren't confident that you could stick with such a strategy for a decade or longer of underperformance relative to buy-and-hold, then you probably shouldn't adopt it.
I don't think this is where the problem really arises. Folks who like trading systems tend to like to tinker with them. That tinkering isn't so much due to recent performance--system builders tend to believe in their systems whether they are working or not. But they do tend to believe in improving or optimizing their systems. So, one starts out with the 200 DMA, then adds a 50 DMA to the mix, then adds in CAPE, then housing starts, etc. The result is that there really isn't any consistent strategy over time.

Swedroe likes to tinker. The main thing I've observed is that the key is be able to put an academic veneer on top of an old idea. "Market timing" is bad; "Time series momentum" is good. "Picking undervalued stocks" is bad, but "factor investing" is good. And so on.

This isn't just an investing thing. Athletes who try new training methods tend to get better more than those who just go out and run or ride their bike. On the other hand, the later group are more likely to still be into their sport of choice 10 or 20 years from now. When it come to saving for retirement, better to imitate the later group than the former.

We've had a lot of debate on this forum over the years as to whether the new ideas/portfolio tinkering that Swedroe continually promotes adds value or not. From a numerical perspective, that's really hard to ascertain as it's hard to figure out what his advice would have been for any particular investor at a certain point in time. But I can demonstrate to myself that I've done better by trying to be mererly "good enough" as opposed to "optimal" in investing. Which is frustrating as I'm an instinctual optimizer.
DonIce
Posts: 1453
Joined: Thu Feb 21, 2019 5:44 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by DonIce »

Market timing by any other name is still market timing.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Larry Swedroe: Trend Following As Insurance

Post by Northern Flicker »

Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities.
If that is true, it pretty well refutes time series momentum as a statistically significant effect. It is by definition a short-term effect.
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

Northern Flicker wrote: Tue May 14, 2019 12:46 am
Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities.
If that is true, it pretty well refutes time series momentum as a statistically significant effect. It is by definition a short-term effect.
That's like saying - the case for holding bonds is refuted by stocks going up for ten years.
Amateur Self-Taught Senior Macro Strategist
User avatar
nisiprius
Advisory Board
Posts: 52212
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Larry Swedroe: Trend Following As Insurance

Post by nisiprius »

Although words have a range of meanings it would be best if people would not call investments "insurance" or insurance "investments." The core of the word "insurance" is sure, certain, secure. The usual meaning of "insurance" is a contract with an insurer to pay certain amounts of money if certain things happen, enforceable in a court of law, and as certain as it's possible for humans and the legal system to make it.

An alleged tendency of an investment strategy to behave in a certain way is not "insurance." During the 1987 crash, when the Dow dropped 22% in a single day, "portfolio insurance" didn't provide any protection at all, and someone whose "portfolio insurance" didn't insure their portfolio could do nothing about it but feel sad.

Swedroe's conclusion is
The evidence shows that while trend-following strategies will tend to lag the market over even long periods (creating the risk of tracking-error regret for impatient investors), they tend to perform best just when their benefits are needed most—in severe market downturn.
"Tend to" (or better, "have tended to") is not "insurance."
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Re: Larry Swedroe: Trend Following As Insurance

Post by Beliavsky »

nisiprius wrote: Tue May 14, 2019 7:58 am Although words have a range of meanings it would be best if people would not call investments "insurance" or insurance "investments." The core of the word "insurance" is sure, certain, secure. The usual meaning of "insurance" is a contract with an insurer to pay certain amounts of money if certain things happen, enforceable in a court of law, and as certain as it's possible for humans and the legal system to make it.

An alleged tendency of an investment strategy to behave in a certain way is not "insurance." During the 1987 crash, when the Dow dropped 22% in a single day, "portfolio insurance" didn't provide any protection at all, and someone whose "portfolio insurance" didn't insure their portfolio could do nothing about it but feel sad.
The stock market fell a good bit on the Friday before Black Monday, and on that Monday I believe S&P futures closed lower than where they opened. So someone following a dynamic portfolio insurance strategy that sells as the market falls would have lost less from the Thursday close before the crash to the close of the crash day than someone who was not "insured". It is true that dynamic portfolio insurance strategy does not provide the certainty that buying a put option does, but buying puts incurs the cost of the volatility risk premium.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by Random Walker »

nisiprius wrote: Tue May 14, 2019 7:58 am Although words have a range of meanings it would be best if people would not call investments "insurance" or insurance "investments." The core of the word "insurance" is sure, certain, secure. The usual meaning of "insurance" is a contract with an insurer to pay certain amounts of money if certain things happen, enforceable in a court of law, and as certain as it's possible for humans and the legal system to make it.

An alleged tendency of an investment strategy to behave in a certain way is not "insurance." During the 1987 crash, when the Dow dropped 22% in a single day, "portfolio insurance" didn't provide any protection at all, and someone whose "portfolio insurance" didn't insure their portfolio could do nothing about it but feel sad.

Swedroe's conclusion is
The evidence shows that while trend-following strategies will tend to lag the market over even long periods (creating the risk of tracking-error regret for impatient investors), they tend to perform best just when their benefits are needed most—in severe market downturn.
"Tend to" (or better, "have tended to") is not "insurance."
Hi Nisi,
Your point is well taken. But I do think it is worthwhile for investors to at least keep the concept of insurance in their mind. With each investment we purchase there is a transfer of risk and an expected premium for assuming the risk. I think looking at investments from that perspective can be valuable.

Dave
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

Forester wrote: Tue May 14, 2019 3:50 am
Northern Flicker wrote: Tue May 14, 2019 12:46 am
Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities.
If that is true, it pretty well refutes time series momentum as a statistically significant effect. It is by definition a short-term effect.
That's like saying - the case for holding bonds is refuted by stocks going up for ten years.
Precisely. Trend following is historically very likely to lag BAH during the good times. Analyses I've shown above demonstrate that.

Many of us are okay with lagging the market during the good times (e.g. 2010-2019) if we can still get good returns during the poor times (e.g. 2000-2009).
The Sensible Steward
User avatar
nedsaid
Posts: 19275
Joined: Fri Nov 23, 2012 11:33 am

Re: Larry Swedroe: Trend Following As Insurance

Post by nedsaid »

stlutz wrote: Mon May 13, 2019 11:41 pm
willthrill81 wrote: Mon May 13, 2019 6:27 pm If you aren't confident that you could stick with such a strategy for a decade or longer of underperformance relative to buy-and-hold, then you probably shouldn't adopt it.
I don't think this is where the problem really arises. Folks who like trading systems tend to like to tinker with them. That tinkering isn't so much due to recent performance--system builders tend to believe in their systems whether they are working or not. But they do tend to believe in improving or optimizing their systems. So, one starts out with the 200 DMA, then adds a 50 DMA to the mix, then adds in CAPE, then housing starts, etc. The result is that there really isn't any consistent strategy over time.

Swedroe likes to tinker. The main thing I've observed is that the key is be able to put an academic veneer on top of an old idea. "Market timing" is bad; "Time series momentum" is good. "Picking undervalued stocks" is bad, but "factor investing" is good. And so on.

This isn't just an investing thing. Athletes who try new training methods tend to get better more than those who just go out and run or ride their bike. On the other hand, the later group are more likely to still be into their sport of choice 10 or 20 years from now. When it come to saving for retirement, better to imitate the later group than the former.

We've had a lot of debate on this forum over the years as to whether the new ideas/portfolio tinkering that Swedroe continually promotes adds value or not. From a numerical perspective, that's really hard to ascertain as it's hard to figure out what his advice would have been for any particular investor at a certain point in time. But I can demonstrate to myself that I've done better by trying to be mererly "good enough" as opposed to "optimal" in investing. Which is frustrating as I'm an instinctual optimizer.
I like Larry a lot and own three of his books. I read his articles as they come out, not every one, but many of them. His writings and whatever interactions I have had with him on the forum have taught me a lot. I do believe in the academic research.

The problem is that there are lots of smart people on Wall Street. If Larry knows about it, chances are pretty good that a lot of those smart people do too. When you have smart folks at institutions that have much greater resources available to them than us ordinary investors, they are going to try to take advantage of whatever market inefficiencies are out there. It is an open question whether or not individual investors can benefit. Fortunately, there are low-cost products out there that individuals can use like the various factor ETFs. So at least we can give it the old college try with really good financial instruments. But then you realize that you need to be in the "right" instruments and presumably that is where Larry's firm comes in.

Markets and the economy are dynamic. Lots has changed during my 35 years of investing. There are better and more efficient ways to invest now. Also much more access to information. Fortunately, investors can use simpler methods of investing that might work better than all the fancy stuff. So we don't have to market geniuses.

So on the one hand, there are timeless principles that never change. On the other hand, there are other things that change constantly. So it is the tried and true vs. what is modern and new. In the old days, buying and holding a portfolio of blue chip stocks, reinvesting the dividends, and doing minimal trading was probably the best way to invest 50 years ago. But it did take know how and patience. Today, the principle of low-cost, low turnover investing is the same but the more modern way is to hold an S&P 500 Index Fund. So the principle stays the same but the method changes.

Good enough is good enough. Close enough is close enough. I can see why lots of folks here just do the simple index fund portfolios. The markets are pretty efficient.
A fool and his money are good for business.
User avatar
jeffyscott
Posts: 13484
Joined: Tue Feb 27, 2007 8:12 am

Re: Larry Swedroe: Trend Following As Insurance

Post by jeffyscott »

willthrill81 wrote: Mon May 13, 2019 6:27 pmI'm not necessarily convinced that TS momentum is entirely incompatible with the concept of efficient markets. But frankly, I don't think that markets are as efficient as many believe them to be. For instance, publishing a factor's historic outperformance of the market shouldn't reduce that factor's premium going forward, but it consistently has.

Beyond that, if the data we are confronted with is incompatible with our hypothesis, we may need to revise our hypothesis. We might even need to take more drastic measures, as indicated below.

"When you discover that you are riding a dead horse, the best strategy is to dismount."
-Native American proverb
I don't know if anyone has ever made it a proverb, but I would add: Too many riders kills the horse. Which is really just another way of stating Rekenthaler's rule.

IMO, this has likely been activated by the time Larry publishes and bogelheads discuss. Therefore, I think trend following is probably already dead as a beneficial strategy. (I did not read the article, though)
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

jeffyscott wrote: Tue May 14, 2019 10:39 am
willthrill81 wrote: Mon May 13, 2019 6:27 pmI'm not necessarily convinced that TS momentum is entirely incompatible with the concept of efficient markets. But frankly, I don't think that markets are as efficient as many believe them to be. For instance, publishing a factor's historic outperformance of the market shouldn't reduce that factor's premium going forward, but it consistently has.

Beyond that, if the data we are confronted with is incompatible with our hypothesis, we may need to revise our hypothesis. We might even need to take more drastic measures, as indicated below.

"When you discover that you are riding a dead horse, the best strategy is to dismount."
-Native American proverb
I don't know if anyone has ever made it a proverb, but I would add: Too many riders kills the horse. Which is really just another way of stating Rekenthaler's rule.

IMO, this has likely been activated by the time Larry publishes and bogelheads discuss. Therefore, I think trend following is probably already dead as a beneficial strategy. (I did not read the article, though)
Well, it certainly worked well from 2000-2009, when stocks had real losses but every trend following strategy I've seen did well. Again, the problem I see with trend following as a strategy for anyone, human or otherwise, is the ability to stick with it when it is underperforming the market significantly for lengthy periods of time.
The Sensible Steward
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by Random Walker »

jeffyscott wrote: Tue May 14, 2019 10:39 am
willthrill81 wrote: Mon May 13, 2019 6:27 pmI'm not necessarily convinced that TS momentum is entirely incompatible with the concept of efficient markets. But frankly, I don't think that markets are as efficient as many believe them to be. For instance, publishing a factor's historic outperformance of the market shouldn't reduce that factor's premium going forward, but it consistently has.

Beyond that, if the data we are confronted with is incompatible with our hypothesis, we may need to revise our hypothesis. We might even need to take more drastic measures, as indicated below.

"When you discover that you are riding a dead horse, the best strategy is to dismount."
-Native American proverb
I don't know if anyone has ever made it a proverb, but I would add: Too many riders kills the horse. Which is really just another way of stating Rekenthaler's rule.

IMO, this has likely been activated by the time Larry publishes and bogelheads discuss. Therefore, I think trend following is probably already dead as a beneficial strategy. (I did not read the article, though)
I certainly appreciate that “if the Bozos know it, it isn’t worth knowing”. But markets have been efficient for a very long time. Even I. The 1600’s information travelled rapidly from South America to Europe. So I think there is strong reason to expect behavioral anomalies from the past to persist despite advances in technology.

Dave
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

Random Walker wrote: Tue May 14, 2019 11:09 am
jeffyscott wrote: Tue May 14, 2019 10:39 am
willthrill81 wrote: Mon May 13, 2019 6:27 pmI'm not necessarily convinced that TS momentum is entirely incompatible with the concept of efficient markets. But frankly, I don't think that markets are as efficient as many believe them to be. For instance, publishing a factor's historic outperformance of the market shouldn't reduce that factor's premium going forward, but it consistently has.

Beyond that, if the data we are confronted with is incompatible with our hypothesis, we may need to revise our hypothesis. We might even need to take more drastic measures, as indicated below.

"When you discover that you are riding a dead horse, the best strategy is to dismount."
-Native American proverb
I don't know if anyone has ever made it a proverb, but I would add: Too many riders kills the horse. Which is really just another way of stating Rekenthaler's rule.

IMO, this has likely been activated by the time Larry publishes and bogelheads discuss. Therefore, I think trend following is probably already dead as a beneficial strategy. (I did not read the article, though)
I certainly appreciate that “if the Bozos know it, it isn’t worth knowing”. But markets have been efficient for a very long time. Even I. The 1600’s information travelled rapidly from South America to Europe. So I think there is strong reason to expect behavioral anomalies from the past to persist despite advances in technology.

Dave
I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The Sensible Steward
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

willthrill81 wrote: Wed May 15, 2019 12:17 am

I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The ideas that
1. computers can't form a strategy
2. computers are just an intermediary to give information for humans to make a decision

shows us very clearly where trend following might go wrong.

These ideas are already outdated, and are likely to be completely false in the very near future.


Some simple examples to highlight where we are now:

IBM already has used their computer not to inform humans but to hire workers.
Computers can already beat the best gamers in games when forming and executing strategies is how you win.
Computers currently assist drivers, but will be doing the driving within the next 10-20 years.

It takes very little imagination to see where we are headed - computer-centric decision making that will have to be better than what a human can do. No automated driving system will be implemented if it performs the same - causing the current amount of crashes people make. Automated systems often have to be much better than human systems at decisions to make them acceptable. Being better at decision making means different decision making.

You better hope that faster and better decision making doesn't change the period or timing of market changes, or how the market responds as a whole, or your data-mined past system may be garbage.

A good strategy thoroughly investigates the risks instead of pretending they don't exist.
Northern Flicker
Posts: 15363
Joined: Fri Apr 10, 2015 12:29 am

Re: Larry Swedroe: Trend Following As Insurance

Post by Northern Flicker »

Forester wrote: Tue May 14, 2019 3:50 am
Northern Flicker wrote: Tue May 14, 2019 12:46 am
Trend has performed poorly though over the last 10 years: buy and hold has beaten trend in every asset class except commodities.
If that is true, it pretty well refutes time series momentum as a statistically significant effect. It is by definition a short-term effect.
That's like saying - the case for holding bonds is refuted by stocks going up for ten years.
Actually it is very different. The case for holding stocks is that over a sufficiently long period they are likely to have a higher return than bonds.

Time series momentum is the hypothesis that over relatively short periods of time stocks that have been increasing in value will continue to do so and those that are decreasing in value will continue to do so. A 10 year period for tine series momentum would be a large number of out-of-sample tests. Momentum is a trading strategy, not the defining property of an asset class that can be expected to mean revert. Momentum trades today are independent from those years in the past.
skeptic42
Posts: 105
Joined: Mon Feb 11, 2019 4:27 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by skeptic42 »

Behavioral anomalies belong to humans, but why should they translate to easily exploitable market anomalies?
That seems to be a convenient narrative, but where is the proof?

Trend followers claim that the effectiveness of trend following is based on behavioral anomalies, right?
So, the burden of proof is on their side. I don't understand how this can work, I see contradictory explanations:
1. Humans follow the herd, so they buy the last winners and sell the last losers.
2. Humans have a strong loss aversion, so they sell their winners too early and hold (buy) their losers too long.
Both seem to explain why trends in the market could emerge, but by contradictory mechanisms.

Can someone proof that behavioral anomalies generate market price anomalies which can be easily exploited to systematically beat the market?
If enough money chases market anomalies, these anomalies should disappear, why is that not true for trends?
FireProof
Posts: 960
Joined: Thu May 05, 2011 12:15 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by FireProof »

Pure technical trading, now? Buying high, selling low? At least a value tilt has a fairly plausible explanation. I look forward to Larry's day trading seminar on death crosses in the crypto market.
User avatar
JoMoney
Posts: 16260
Joined: Tue Jul 23, 2013 5:31 am

Re: Larry Swedroe: Trend Following As Insurance

Post by JoMoney »

FireProof wrote: Wed May 15, 2019 3:48 am Pure technical trading, now? Buying high, selling low? At least a value tilt has a fairly plausible explanation. I look forward to Larry's day trading seminar on death crosses in the crypto market.
The way many people seem to implement a "value" strategy, it's not any better, they're just doing the opposite...buying more when price goes down and selling when price goes up. Putting something like 'book' value in the denominator under price is slightly more nuanced... but not much. Many have no idea, nor even an opinion on what the value of the company they're buying is, and are just arbitrarily following some metric (or choosing an index that is).
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

skeptic42 wrote: Wed May 15, 2019 3:40 am
Can someone proof that behavioral anomalies generate market price anomalies which can be easily exploited to systematically beat the market?
If enough money chases market anomalies, these anomalies should disappear, why is that not true for trends?
For whatever reason there are periods where markets move up and then move down. Stocks didn't move up in 2005 then down in 2006, and so on, in alternate years. Up performance & down performance are encapsulated by years.

A dumb trend system is guaranteed to avoid a prolonged bear market. The question is, are the whipsaws compensated for by selling in Jan '08 and buying back in, Sept '09. We may not have another bad GFC bear market until the 2030s. What if markets are choppy & sideways in the meantime, with subdued returns? That would be bad for trend following.

Another way of thinking about behaviour - arguably, stocks were fairly valued (given the environment) in September 2008, but oversold by March 2009. But no one knew at the time when the pain would end. Do you sell when stocks are 40% down, or wait until they're 60% down?

July 2008 Under-reaction - "the market is down 20% but I am staying the course"

February 2009 Over-reaction - "I just can't take this any more"
Amateur Self-Taught Senior Macro Strategist
skeptic42
Posts: 105
Joined: Mon Feb 11, 2019 4:27 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by skeptic42 »

Forester wrote: Wed May 15, 2019 7:35 am
skeptic42 wrote: Wed May 15, 2019 3:40 am
Can someone proof that behavioral anomalies generate market price anomalies which can be easily exploited to systematically beat the market?
If enough money chases market anomalies, these anomalies should disappear, why is that not true for trends?
For whatever reason there are periods where markets move up and then move down. Stocks didn't move up in 2005 then down in 2006, and so on, in alternate years. Up performance & down performance are encapsulated by years.

A dumb trend system is guaranteed to avoid a prolonged bear market. The question is, are the whipsaws compensated for by selling in Jan '08 and buying back in, Sept '09. We may not have another bad GFC bear market until the 2030s. What if markets are choppy & sideways in the meantime, with subdued returns? That would be bad for trend following.

Another way of thinking about behaviour - arguably, stocks were fairly valued (given the environment) in September 2008, but oversold by March 2009. But no one knew at the time when the pain would end. Do you sell when stocks are 40% down, or wait until they're 60% down?

July 2008 Under-reaction - "the market is down 20% but I am staying the course"

February 2009 Over-reaction - "I just can't take this any more"
Yes, a dumb trend system is guaranteed to avoid a prolonged bear market, but it could create its own bear market by a bad sequence of whipsaws. In contrast, a dumb buy and hold strategy is guaranteed to avoid whipsaws.

But I am still missing the proof that these persistent behavioral anomalies translate into a systematic profit (more return or less risk than the market). I don't see it, it looks like random noise. To profit from behavioral anomalies, the market participants have to change their behavior and one has to be positioned accordingly, but that contradicts the premise that behavioral anomalies are persistent.
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

skeptic42 wrote: Wed May 15, 2019 8:09 am
Yes, a dumb trend system is guaranteed to avoid a prolonged bear market, but it could create its own bear market by a bad sequence of whipsaws. In contrast, a dumb buy and hold strategy is guaranteed to avoid whipsaws.
This has happened since the GFC... AFAIK 60/40 has beaten trend following despite being less exposed to stocks.
But I am still missing the proof that these persistent behavioral anomalies translate into a systematic profit (more return or less risk than the market). I don't see it, it looks like random noise. To profit from behavioral anomalies, the market participants have to change their behavior and one has to be positioned accordingly, but that contradicts the premise that behavioral anomalies are persistent.
Gray/Vogel study 1976-2014, 'DIY Financial Advisor'

S&P buy&hold 11.74%, 50.21% drawdown

S&P 12mo momentum 11.78%, 29.58% drawdown

Consider this includes two perfect V-shape deep drawdowns in a decade, favourable to TF... it's very much up in the air. In the long run, would TF which is 70% of the time invested, beat a fixed 70-30 portfolio?

Yet this is only applying a model to the S&P 500. A popular variation is switching between US and ex-US stocks.

edit: I am uncertain if TF translates into a profit. It may depend on deep drawdowns occuring every 10-15 years, versus waiting over a quarter of a century.
Amateur Self-Taught Senior Macro Strategist
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

abc132 wrote: Wed May 15, 2019 1:33 am
willthrill81 wrote: Wed May 15, 2019 12:17 am

I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The ideas that
1. computers can't form a strategy
2. computers are just an intermediary to give information for humans to make a decision
I never said that computers aren't 'making decisions'. But if humans are not happy with the decisions that the computers are making, they will change how the computers make those decisions. Computers don't need money.
The Sensible Steward
skeptic42
Posts: 105
Joined: Mon Feb 11, 2019 4:27 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by skeptic42 »

Forester wrote: Wed May 15, 2019 9:12 am In the long run, would TF which is 70% of the time invested, beat a fixed 70-30 portfolio?
Obviously, if the TF buy and sell signals have a predictive power, then TF could beat the fixed portfolio (after costs).
If the TF buy and sell signals turn out to be completely random, then, in the long run, TF which is 70% of the time invested should have the same expected return but with more volatility than a fixed 70-30 portfolio.

Is the apparent predictive power of trends more than a data-mined artifact? The narrative of persistent behavioral anomalies is missing the link to how one can profit from them by more than luck. And if everyone knows about trends, why are they not arbitraged away? Or are they and trend followers will experience it some day? :twisted:
User avatar
nisiprius
Advisory Board
Posts: 52212
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Larry Swedroe: Trend Following As Insurance

Post by nisiprius »

Fred Schwed wrote this in 1940:
In 'Where are the Customer's Yachts' (1940), Fred Schwed wrote:All I was ever able to conclude from my informal studies was that chart reading is a complex way of arriving at a simple theorem, to wit: when [stocks] have gone up for a considerable time, they will continue to go up for a considerable time; and the same holds true for going down. This is simple, but it does not happen to be so. The easiest way of perceiving that it is not so is to go get a properly drawn chart and look at it.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

skeptic42 wrote: Wed May 15, 2019 10:29 am
Forester wrote: Wed May 15, 2019 9:12 am In the long run, would TF which is 70% of the time invested, beat a fixed 70-30 portfolio?
Obviously, if the TF buy and sell signals have a predictive power, then TF could beat the fixed portfolio (after costs).
If the TF buy and sell signals turn out to be completely random, then, in the long run, TF which is 70% of the time invested should have the same expected return but with more volatility than a fixed 70-30 portfolio.

Is the apparent predictive power of trends more than a data-mined artifact? The narrative of persistent behavioral anomalies is missing the link to how one can profit from them by more than luck. And if everyone knows about trends, why are they not arbitraged away? Or are they and trend followers will experience it some day? :twisted:
It's difficult to claim that a strategy that has worked quite well over the long-term across the total stock market, factor indices, industries, sectors, ex-U.S. equities, global currencies, fixed income, and commodities is nothing more than a sample-specific artifact.
The Sensible Steward
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

nisiprius wrote: Wed May 15, 2019 10:54 am Fred Schwed wrote this in 1940:
In 'Where are the Customer's Yachts' (1940), Fred Schwed wrote:All I was ever able to conclude from my informal studies was that chart reading is a complex way of arriving at a simple theorem, to wit: when [stocks] have gone up for a considerable time, they will continue to go up for a considerable time; and the same holds true for going down. This is simple, but it does not happen to be so. The easiest way of perceiving that it is not so is to go get a properly drawn chart and look at it.
So you deny the academic research which has found that TS momentum to be significant across time, asset classes, and geography? Equating that factor with chart reading doesn't seem appropriate.
The Sensible Steward
BJJ_GUY
Posts: 882
Joined: Wed Mar 13, 2019 7:45 am

Re: Larry Swedroe: Trend Following As Insurance

Post by BJJ_GUY »

skeptic42 wrote: Wed May 15, 2019 10:29 am
Is the apparent predictive power of trends more than a data-mined artifact? The narrative of persistent behavioral anomalies is missing the link to how one can profit from them by more than luck. And if everyone knows about trends, why are they not arbitraged away? Or are they and trend followers will experience it some day? :twisted:
Not specific to this post only, but this thread generally seems to be kind of blending various ideas, terminology, and other complex topics into the same thing. I'm inclined to believe a large majority of the debate is due more to folks referring to similar but different 'things which makes a conversation tough. Interesting topic nonetheless.

Related to the quoted comment, it might help if you think of trend-following as a 'reactive' not 'predictive' strategy. With that in mind, the next level question you ask about consistency of behavioral tendencies is more helpful to think about as explanatory, ex-post. Because trend-following in the most naive form does not attempt to predict anything, and because it's purely price based, there actually isn't an ex-ante logic needed.
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

willthrill81 wrote: Wed May 15, 2019 10:27 am
abc132 wrote: Wed May 15, 2019 1:33 am
willthrill81 wrote: Wed May 15, 2019 12:17 am

I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The ideas that
1. computers can't form a strategy
2. computers are just an intermediary to give information for humans to make a decision
I never said that computers aren't 'making decisions'. But if humans are not happy with the decisions that the computers are making, they will change how the computers make those decisions. Computers don't need money.
You said exactly those words, that computers could not form a strategy.

If you wish to state that you misspoke or were wrong, that would be helpful if you no longer feel that way.

I'm telling you why your strategy could fail - from the inability to understand if/how the market might be different in the future than the past.

There are no guarantees, but explaining the risks of a strategy is part of developing a good strategy. I would warn anyone that sees you completely ignoring this risk that the strategy has not been well thought out if it ignores risks rather than addressing them. HEDGEFUNDIE welcomed such questions in their thread, and I would encourage you to address the comments of others if you are interested in making a strong argument.
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

skeptic42 wrote: Wed May 15, 2019 10:29 am
Forester wrote: Wed May 15, 2019 9:12 am In the long run, would TF which is 70% of the time invested, beat a fixed 70-30 portfolio?
Obviously, if the TF buy and sell signals have a predictive power, then TF could beat the fixed portfolio (after costs).
If the TF buy and sell signals turn out to be completely random, then, in the long run, TF which is 70% of the time invested should have the same expected return but with more volatility than a fixed 70-30 portfolio.

Is the apparent predictive power of trends more than a data-mined artifact? The narrative of persistent behavioral anomalies is missing the link to how one can profit from them by more than luck. And if everyone knows about trends, why are they not arbitraged away? Or are they and trend followers will experience it some day? :twisted:
Alpha Architect found that trend following the S&P did beat fixed allocations... again with the caveat that the 2000s were *great* for models which deal with deep V bottoms & recoveries.

https://alphaarchitect.com/2018/10/16/w ... following/

1973-2017

76.11% stocks 23.89% tbills
9.35% cagr, 40.39% DD

Trend (same exposure to risk assets over the timeframe)
10.87% cagr, 23.58% DD
Amateur Self-Taught Senior Macro Strategist
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

BJJ_GUY wrote: Wed May 15, 2019 11:06 am Related to the quoted comment, it might help if you think of trend-following as a 'reactive' not 'predictive' strategy. With that in mind, the next level question you ask about consistency of behavioral tendencies is more helpful to think about as explanatory, ex-post. Because trend-following in the most naive form does not attempt to predict anything, and because it's purely price based, there actually isn't an ex-ante logic needed.
Spot on. 'Technical analysis', broadly speaking, is an attempt to predict the future. At its core, trend following is indeed merely reacting to current information on the basis of historical odds.

Many, perhaps even most, Bogleheads will readily admit that after stocks decline in value by 50%, the likelihood of another 50% decline is less than it was before; it's not impossible, but it's less likely. Historically, this has certainly been true. What is being subtly admitted here is that stocks' volatility is not static but is actually dynamic. Similarly, after a 50% decline, stocks' expected returns are also higher. Broadly speaking, trend following seeks to determine the conditions that are likely to impact volatility and expected returns and react accordingly.

Undoubtedly the most widely studied such condition is an asset's price over time. Historically, when U.S. stocks were trading below their 200 DMA, their volatility was markedly higher and the return for holding them was very low. Conversely, when stocks were trading above their 200 DMA, their volatility was lower and their return was significantly higher. There were certainly periods where the opposite was true, but over the long-term, this has been the case. As a group, trend followers simply choose to be invested in stocks only when the expected return is sufficiently high and/or expected volatility is sufficiently low.

Where this practice has gotten a bad rap is the fact that while humans seem to have a strong understanding of trends existing throughout the world, they have a difficult time taking advantage of them in securities like stocks because they are often basing their decisions in large part on their emotions, not objective data. And among those who rely exclusively on objective data, they are often constantly trying to refine which data they are using and how, letting 'perfect be the enemy of good'.
The Sensible Steward
User avatar
nisiprius
Advisory Board
Posts: 52212
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Larry Swedroe: Trend Following As Insurance

Post by nisiprius »

willthrill81 wrote: Wed May 15, 2019 10:59 am...So you deny the academic research which has found that TS momentum to be significant across time, asset classes, and geography? Equating that factor with chart reading doesn't seem appropriate...
My point is that there have been momentum advocates and momentum skeptics for at least eighty years. A 1937 research paper, for example, concluded that "inertia" was a real effect, but that it "could not be employed by speculators with any assurance of consistent or large profits."

And so it goes. There is no "the" academic research. The same Grey and Vogel also asked, in 2018, in AlphaArchitect, Are trend-following and time-series momentum research results robust? And they cited academic research: Huang, Dashan and Li, Jiangyuan and Wang, Liyao and Zhou, Guofu, January 10, 2019, Journal of Financial Economics (JFE), Time-Series Momentum: Is It There?
Time series momentum (TSM) refers to the predictability of the past 12-month return on the next one-month return and is the focus of several recent influential studies. This paper shows that asset-by-asset time series regressions reveal little evidence of TSM, both in- and out-of-sample. While the t-statistic in a pooled regression appears large, it is not statistically reliable as it is less than the critical values of parametric and nonparametric bootstraps. From an investment perspective, the TSM strategy is profitable, but its performance is virtually the same as that of a similar strategy that is based on historical sample mean and does not require predictability. Overall, the evidence on TSM is weak, particularly for the large cross section of assets.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Always passive
Posts: 1267
Joined: Fri Apr 14, 2017 4:25 am
Location: Israel

Re: Larry Swedroe: Trend Following As Insurance

Post by Always passive »

Larry Swedroe is a consultant of Alpha Architect. These people use Quant Investing strategies, so I guess that Larry is far from being a passive investor.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

abc132 wrote: Wed May 15, 2019 11:14 am
willthrill81 wrote: Wed May 15, 2019 10:27 am
abc132 wrote: Wed May 15, 2019 1:33 am
willthrill81 wrote: Wed May 15, 2019 12:17 am

I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The ideas that
1. computers can't form a strategy
2. computers are just an intermediary to give information for humans to make a decision
I never said that computers aren't 'making decisions'. But if humans are not happy with the decisions that the computers are making, they will change how the computers make those decisions. Computers don't need money.
You said exactly those words, that computers could not form a strategy.
No, I didn't. I said it was debatable as to whether computers can strategize. It comes down to the difference between tactics, which computers clearly excel at, and strategy. For instance, the arena of chess, world-class humans do not always lose every game to the best computer programs out there, even though the latter can evaluate millions or billions of positions per second, which no human can do. How is it possible for these programs to not win every single time? Many speculate that it's because humans can strategize, which computers cannot do, and that computers can only beat top human players because their tactical ability is so superior to that of humans. I'm not saying that this is correct, nor am I saying that computers cannot form what is generally referred to as a strategy in any arena, only that it is not a settled issue as of now.
The Sensible Steward
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Larry Swedroe: Trend Following As Insurance

Post by willthrill81 »

nisiprius wrote: Wed May 15, 2019 11:31 am
willthrill81 wrote: Wed May 15, 2019 10:59 am...So you deny the academic research which has found that TS momentum to be significant across time, asset classes, and geography? Equating that factor with chart reading doesn't seem appropriate...
My point is that there have been momentum advocates and momentum skeptics for at least eighty years. A 1937 research paper, for example, concluded that "inertia" was a real effect, but that it "could not be employed by speculators with any assurance of consistent or large profits."

And so it goes. There is no "the" academic research. The same Grey and Vogel also asked, in 2018, in AlphaArchitect, Are trend-following and time-series momentum research results robust? And they cited academic research: Huang, Dashan and Li, Jiangyuan and Wang, Liyao and Zhou, Guofu, January 10, 2019, Journal of Financial Economics (JFE), Time-Series Momentum: Is It There?
Time series momentum (TSM) refers to the predictability of the past 12-month return on the next one-month return and is the focus of several recent influential studies. This paper shows that asset-by-asset time series regressions reveal little evidence of TSM, both in- and out-of-sample. While the t-statistic in a pooled regression appears large, it is not statistically reliable as it is less than the critical values of parametric and nonparametric bootstraps. From an investment perspective, the TSM strategy is profitable, but its performance is virtually the same as that of a similar strategy that is based on historical sample mean and does not require predictability. Overall, the evidence on TSM is weak, particularly for the large cross section of assets.
The paper you quote very narrowly defines TS momentum. It would be akin to saying that the value premium doesn't exist because the one, very specific means we use to define it does not appear to function in just the way we state that it should. That doesn't jive with Larry's research in this area nor that of many other academics. The link I posted above demonstrates that trend following has indeed outperformed over many asset classes and factors across time and geography.
The Sensible Steward
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

willthrill81 wrote: Wed May 15, 2019 11:45 am
abc132 wrote: Wed May 15, 2019 11:14 am
willthrill81 wrote: Wed May 15, 2019 10:27 am
abc132 wrote: Wed May 15, 2019 1:33 am
willthrill81 wrote: Wed May 15, 2019 12:17 am

I'm inclined to agree that 'behavioral anomalies' are likely to persist. Many point to computers doing the majority of trading and say that such things are gone, but computers are only intermediaries doing what humans have programmed them to do. If the computers stop performing to the satisfaction of those running them, they will reprogram them accordingly. So I don't really see things as being meaningfully different in that regard today.
The ideas that
1. computers can't form a strategy
2. computers are just an intermediary to give information for humans to make a decision
I never said that computers aren't 'making decisions'. But if humans are not happy with the decisions that the computers are making, they will change how the computers make those decisions. Computers don't need money.
You said exactly those words, that computers could not form a strategy.
No, I didn't. I said it was debatable as to whether computers can strategize. It comes down to the difference between tactics, which computers clearly excel at, and strategy. For instance, the arena of chess, world-class humans do not always lose every game to the best computer programs out there, even though the latter can evaluate millions or billions of positions per second, which no human can do. How is it possible for these programs to not win every single time? Many speculate that it's because humans can strategize, which computers cannot do, and that computers can only beat top human players because their tactical ability is so superior to that of humans. I'm not saying that this is correct, nor am I saying that computers cannot form what is generally referred to as a strategy in any arena, only that it is not a settled issue as of now.
How is it that a human with better strategy doesn't win every chess event? Does that person not have strategy because they don't win every event?

There is a logical disconnect between applying a single case to the whole.

Your example failed when it tried to go from a single case to every case.


We are talking about how an investment strategy will work in the future, so the context of the discussion needs to be whether or not computers will be able to form investing strategies in the future, that are different from that of humans.

The answer is clear and obvious.


Are you willing to acknowledge that there is a real risk that the future will not look like the past in a way that causes your trend strategy to fail?

I will define failure as not preventing the big downturns while capturing most of the upsides.
garlandwhizzer
Posts: 3565
Joined: Fri Aug 06, 2010 3:42 pm

Re: Larry Swedroe: Trend Following As Insurance

Post by garlandwhizzer »

I have a practical application question for willthrill81 about his trend following strategy in action in the recent past. Please excuse me if you have answered this before, willthrill81, in some other post but I am curious. The market decline last fall climaxed on Christmas Eve with a 19+% drop in the S&P 500. I suspect but do not know for sure that such a decline initiated a sell of equity in your trend following strategy. I assume that your equity was in a tax-deferred account so taxes would not have been initiated by this move. If equity had been in a non-tax deferred account that move would have generated unanticipated taxes right at the end of the tax year, not a good thing. So I assume this trend following strategy is presumably for tax deferred accounts only for tax efficiency.

The other question I have is about subsequent market action. Having sold equity at or near the bottom of that long decline, presumably you were not on board when the market suddenly skyrocketed back up, recovered all those loses, and even went on to new highs. Did your strategic rules for trend following allow you to take advantage of that rather quick and rather steep rebound? I realize that the market could have kept on dropping and turned into a severe bear market. Trend following protected you from that outcome but it seems that protection does not come free of charge. There was an opportunity cost relative to a buy and hold strategy in this case. That doesn't suggest that trend following is a bad strategy, just that like other strategies that protect from downside market volatility, it costs for that protection. Picking the right point to get back into the market after a decline can be tricky.

Market timing is difficult because it involves making two decisions well--when to sell out and when to buy back in. For most of us, it's hard enough to get one of these right, let alone both. Just curious, how well did your trend following parameters navigate this recent episode?

Garland Whizzer
BJJ_GUY
Posts: 882
Joined: Wed Mar 13, 2019 7:45 am

Re: Larry Swedroe: Trend Following As Insurance

Post by BJJ_GUY »

abc132 wrote: Wed May 15, 2019 11:59 am
Are you willing to acknowledge that there is a real risk that the future will not look like the past in a way that causes your trend strategy to fail?

I will define failure as not preventing the big downturns while capturing most of the upsides.
With that definition, I'd say there is low probability of not 'failing'

I'm not a fan of trend-following strategies, but I do understand the logic supporting the investment for certain types of investors.

I am skeptical most of these (naive trend-following) strategies will have attractive returns across a full cycle. In fact, I don't believe most investors are anticipating the strategies will capture most of the upside during a vertical market. However, because sell-offs tend to be more dramatic, these strategies are most certainly being counted on to do well in a bear market as long as there is some reasonable level of persistence (gaping down, and then recovery isn't so good).

If one were to believe in trend-following as an additive allocation within the broader total portfolio context, it's likely because they intend to rebalance from trend and into cheaper equities when a sell-off occurs. Adding to equity at cheaper valuations will reduce the amount of time for the portfolio to recover post-recession. Also, by trimming a well-performing trend strategy (and rebalancing elsewhere), the investor is reducing the inevitable whipsaw losses.

Said another way, it's hard to see much value being added from certain strategies - and especially those that act like they are long volatility - using static assumptions. Of course this introduces execution risk for the person overseeing the total portfolio. Specifically, when should rebalancing efforts take place and for how much? It's easy to be too early, or too late, but that doesn't mean the exposure can't be beneficial, only that there is some complexity to that answer.
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

This is the unemployment filter. It's hit rate in terms of filtering the S&P 500 moving average is just about faultless, post WWII. The time spent invested jumps from around 70% to 88% - that's a ton of avoided whipsaws. Trend is only active in the grey areas. Otherwise, the strategy is invested 100% whatever is happening in the market.

Almost seems too good to be true!

Image
Amateur Self-Taught Senior Macro Strategist
BJJ_GUY
Posts: 882
Joined: Wed Mar 13, 2019 7:45 am

Re: Larry Swedroe: Trend Following As Insurance

Post by BJJ_GUY »

Forester wrote: Wed May 15, 2019 12:44 pm This is the unemployment filter. It's hit rate in terms of filtering the S&P 500 moving average is just about faultless, post WWII. The time spent invested jumps from around 70% to 88% - that's a ton of avoided whipsaws. Trend is only active in the grey areas. Otherwise, the strategy is invested 100% whatever is happening in the market.

Almost seems too good to be true!

Image
What trend are you using?
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

BJJ_GUY wrote: Wed May 15, 2019 12:39 pm
abc132 wrote: Wed May 15, 2019 11:59 am
Are you willing to acknowledge that there is a real risk that the future will not look like the past in a way that causes your trend strategy to fail?

I will define failure as not preventing the big downturns while capturing most of the upsides.
With that definition, I'd say there is low probability of not 'failing'

I'm not a fan of trend-following strategies, but I do understand the logic supporting the investment for certain types of investors.

I am skeptical most of these (naive trend-following) strategies will have attractive returns across a full cycle. In fact, I don't believe most investors are anticipating the strategies will capture most of the upside during a vertical market. However, because sell-offs tend to be more dramatic, these strategies are most certainly being counted on to do well in a bear market as long as there is some reasonable level of persistence (gaping down, and then recovery isn't so good).

If one were to believe in trend-following as an additive allocation within the broader total portfolio context, it's likely because they intend to rebalance from trend and into cheaper equities when a sell-off occurs. Adding to equity at cheaper valuations will reduce the amount of time for the portfolio to recover post-recession. Also, by trimming a well-performing trend strategy (and rebalancing elsewhere), the investor is reducing the inevitable whipsaw losses.

Said another way, it's hard to see much value being added from certain strategies - and especially those that act like they are long volatility - using static assumptions. Of course this introduces execution risk for the person overseeing the total portfolio. Specifically, when should rebalancing efforts take place and for how much? It's easy to be too early, or too late, but that doesn't mean the exposure can't be beneficial, only that there is some complexity to that answer.
I'm trying to cross the gap to acknowledging there are additional risks to trend following. If this happens, the discussion can proceed to how big those risks are, and why.

If risks to an investing model can not be acknowledged, I would say the strategy is not a good one. Hopefully it is a lucky one.
abc132
Posts: 2435
Joined: Thu Oct 18, 2018 1:11 am

Re: Larry Swedroe: Trend Following As Insurance

Post by abc132 »

Forester wrote: Wed May 15, 2019 12:44 pm This is the unemployment filter. It's hit rate in terms of filtering the S&P 500 moving average is just about faultless, post WWII. The time spent invested jumps from around 70% to 88% - that's a ton of avoided whipsaws. Trend is only active in the grey areas. Otherwise, the strategy is invested 100% whatever is happening in the market.

Almost seems too good to be true!

Image
This of course implies that a countries stock performance and economic state relies on human capital/labor (employment).

What are the chances this is less true in the future? Are they non-zero?

Could people actually get paid to not work if machines are doing the work?


I think 10% of the thought put into past trend following should be put into what the future might look like.

Human employment was very important in the past, but may or may not be in the future.
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: Larry Swedroe: Trend Following As Insurance

Post by Forester »

BJJ_GUY wrote: Wed May 15, 2019 12:49 pm What trend are you using?
That article paired 10mo MA with 12mo unemployment trend - but other lookbacks would perform similarly. The UER filter is doing all the heavy lifting.

abc132 wrote: Wed May 15, 2019 12:57 pm This of course implies that a countries stock performance and economic state relies on human capital/labor (employment).

What are the chances this is less true in the future? Are they non-zero?

Could people actually get paid to not work if machines are doing the work?
I'd like to understand better why unemployment usually creeps up before the market turns south. Maybe in the future it will be different?

This market timing method seems too good to be true, but it is what it is. The temptation to make heavy use of it is there.
Amateur Self-Taught Senior Macro Strategist
Post Reply