Meb Faber Timing Model

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
Topic Author
tinscale
Posts: 396
Joined: Thu Dec 17, 2009 11:16 pm
Location: North Carolina

Meb Faber Timing Model

Post by tinscale » Wed Mar 25, 2020 9:24 am

Of course, most (if not all) wish they got out right before all this mess.

I thought of timing models that rely on moving averages, and remembered Mebane Faber’s 2006 Tactical Asset Allocation Model. His system is:

- set up a portfolio of 5 ETFs — S&P 500, EAFE, a commodity index, a REIT index, and 10-year treasury index

- buy every ETF if its price is above the 10-month simple moving average

- sell the ETF if it drops below the 10-month simple moving average (and stay in cash until you get the signal to buy again)

Here’s an article/analysis dated March 21, 2020 that concludes the following:

Had you invested in this strategy from Jan 1973 to Mar 2020, your $100,000.00 would now be worth $7,357,812.39. That’s about a 9.53% compound annual return with a maximum drawdown of (12.87%). The MAR ratio is 0.74. The longest time you had to wait for it to recover from its previous equity peak is 27.1 months.

In contrast, had you invested in S&P 500, your $100,000.00 would now be $7,675,060.49, or 9.63% compounded, but you would have to suffer through a maximum drawdown of (55.25%) to get that. This yields a MAR ratio of 0.17. The longest time you would have to wait for it to make a new equity high is 73.5 months.

Our tests show that the Faber TAA model works, and we have high confidence it will continue to work in the future. The rules of the model are simple, and, being a monthly strategy, fairly low maintenance. As Mebane said in his paper, the objective of this model is to produce equity-like returns with bond-like drawdowns, and we agree that this strategy delivers on that vision.

I don't know that I would follow the strategy, but I wish I had been watching the 200-day SMA. I liquidated my only international ETF (SPEM) on Jan 30 when concerns about the virus in China started coming out, but didn't make the connection to the US impacts, panic, etc. and therefore did nothing with my US portfolio. :oops:

rascott
Posts: 1717
Joined: Wed Apr 15, 2015 10:53 am

Re: Meb Faber Timing Model

Post by rascott » Wed Mar 25, 2020 9:31 am

Several of us have discussed the MA timing model quite a bit. Nearly 100% equity returns for decades and counting..... with drawdowns similar to a 40/60 AA.

I think going forward even more near-retirees / retirees should consider it over a bond heavy portfolio offering negative real expected returns.

But any level of market timing.... even systemic models
... are generally frowned heavily upon here.

Lee_WSP
Posts: 2120
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: Meb Faber Timing Model

Post by Lee_WSP » Wed Mar 25, 2020 9:59 am

During bull markets it performs better than a 60/40 with the added cost of having to watch the stock market like a hawk; or you can use the tactical asset allocation ETF (can't recall the ticker).

However, during bull markets it trails 100/0 fairly significantly with additional costs.

The only two times it has absolutely positively shown it's mettle is 2000 and 2008. This third time is still up in the air, but for now, it's looking like either a wash or a win. Certainly not a loss.

User avatar
nisiprius
Advisory Board
Posts: 40248
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: Meb Faber Timing Model

Post by nisiprius » Wed Mar 25, 2020 10:27 am

How well did it work in a real-world test using real money: the GTAA (Global Tactical Asset Allocation) ETF, which launched in late 2010, and in which Faber said he had put over 90% of his net worth?

Vanguard LifeStrategy Conservative fund (40/60, blue)
GTAA, orange

Image

The time period represents the period during which Faber was personally one of the fund managers. He left, the ETF continued for a while with similar performance, and was eventually shuttered.
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
vineviz
Posts: 5599
Joined: Tue May 15, 2018 1:55 pm

Re: Meb Faber Timing Model

Post by vineviz » Wed Mar 25, 2020 10:35 am

rascott wrote:
Wed Mar 25, 2020 9:31 am
Several of us have discussed the MA timing model quite a bit. Nearly 100% equity returns for decades and counting..... with drawdowns similar to a 40/60 AA.

I think going forward even more near-retirees / retirees should consider it over a bond heavy portfolio offering negative real expected returns.

But any level of market timing.... even systemic models
... are generally frowned heavily upon here.
I'll add that anyone considering a timing model like this should make sure they understand just how sensitive such models are to model specification risk and rebalance timing luck.

See this analysis by Corey Hoffstein: https://blog.thinknewfound.com/2020/03/what-the-trend/
One of the hardest problems that allocators face is disentangling luck from skill. Investment performance is often used as a proxy, but herein we hope to have demonstrated that thousands of basis points of relative performance difference can be due to numerous lucky events, namely:

• The luck of which model you use.
• The luck of when you rebalance (e.g. monthly; weekly; continuously).
• The luck of the execution achieved.

Of course, at Newfound we would argue that you can largely avoid these sources of luck through diversification. Don’t use only one model: use several. Don’t rebalance in discrete time steps (e.g. monthly): rebalance partially but more continuously. And if you can avoid it, try not to execute all at once during high volatility days: execute throughout the day.

All of these activities can help curb the impact of bad luck. But they will also curb the impact of good luck. It is entirely possible that you could pick the right model, rebalance at just the right time, and get the best execution of the day. We’d argue, though, that achieving consistency of outcomes requires working to reduce the impact of randomness.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

rascott
Posts: 1717
Joined: Wed Apr 15, 2015 10:53 am

Re: Meb Faber Timing Model

Post by rascott » Wed Mar 25, 2020 10:43 am

vineviz wrote:
Wed Mar 25, 2020 10:35 am
rascott wrote:
Wed Mar 25, 2020 9:31 am
Several of us have discussed the MA timing model quite a bit. Nearly 100% equity returns for decades and counting..... with drawdowns similar to a 40/60 AA.

I think going forward even more near-retirees / retirees should consider it over a bond heavy portfolio offering negative real expected returns.

But any level of market timing.... even systemic models
... are generally frowned heavily upon here.
I'll add that anyone considering a timing model like this should make sure they understand just how sensitive such models are to model specification risk and rebalance timing luck.

See this analysis by Corey Hoffstein: https://blog.thinknewfound.com/2020/03/what-the-trend/
One of the hardest problems that allocators face is disentangling luck from skill. Investment performance is often used as a proxy, but herein we hope to have demonstrated that thousands of basis points of relative performance difference can be due to numerous lucky events, namely:

• The luck of which model you use.
• The luck of when you rebalance (e.g. monthly; weekly; continuously).
• The luck of the execution achieved.

Of course, at Newfound we would argue that you can largely avoid these sources of luck through diversification. Don’t use only one model: use several. Don’t rebalance in discrete time steps (e.g. monthly): rebalance partially but more continuously. And if you can avoid it, try not to execute all at once during high volatility days: execute throughout the day.

All of these activities can help curb the impact of bad luck. But they will also curb the impact of good luck. It is entirely possible that you could pick the right model, rebalance at just the right time, and get the best execution of the day. We’d argue, though, that achieving consistency of outcomes requires working to reduce the impact of randomness.

Good analysis. Diversification of systems used may be the best.

aristotelian
Posts: 6976
Joined: Wed Jan 11, 2017 8:05 pm

Re: Meb Faber Timing Model

Post by aristotelian » Wed Mar 25, 2020 10:55 am

Intuitively it doesn't make sense to me to sell low and buy high. I wonder what the performance would be like if you reversed the decision rules and sold when above the 10MMA.

User avatar
Topic Author
tinscale
Posts: 396
Joined: Thu Dec 17, 2009 11:16 pm
Location: North Carolina

Re: Meb Faber Timing Model

Post by tinscale » Wed Mar 25, 2020 11:03 am

aristotelian wrote:
Wed Mar 25, 2020 10:55 am
Intuitively it doesn't make sense to me to sell low and buy high. I wonder what the performance would be like if you reversed the decision rules and sold when above the 10MMA.
They analyze that at the link I provided. "This strategy turned an initial investment of $100,000.00 to $137,604.17 from Jan 1973 to Mar 2020 – a 0.68% compounded annual return. In terms of risk, the biggest drawdown so far is (45.26%), and the longest drawdown took 234.5 months from start to finish. The risk-reward ratio, as measured by MAR, is 0.01."

User avatar
Forester
Posts: 897
Joined: Sat Jan 19, 2019 2:50 pm
Location: UK

Re: Meb Faber Timing Model

Post by Forester » Wed Mar 25, 2020 1:11 pm

tinscale wrote:
Wed Mar 25, 2020 9:24 am
Of course, most (if not all) wish they got out right before all this mess.

I thought of timing models that rely on moving averages, and remembered Mebane Faber’s 2006 Tactical Asset Allocation Model. His system is:

- set up a portfolio of 5 ETFs — S&P 500, EAFE, a commodity index, a REIT index, and 10-year treasury index

- buy every ETF if its price is above the 10-month simple moving average

- sell the ETF if it drops below the 10-month simple moving average (and stay in cash until you get the signal to buy again)

Here’s an article/analysis dated March 21, 2020 that concludes the following:

Had you invested in this strategy from Jan 1973 to Mar 2020, your $100,000.00 would now be worth $7,357,812.39. That’s about a 9.53% compound annual return with a maximum drawdown of (12.87%). The MAR ratio is 0.74. The longest time you had to wait for it to recover from its previous equity peak is 27.1 months.

In contrast, had you invested in S&P 500, your $100,000.00 would now be $7,675,060.49, or 9.63% compounded, but you would have to suffer through a maximum drawdown of (55.25%) to get that. This yields a MAR ratio of 0.17. The longest time you would have to wait for it to make a new equity high is 73.5 months.

Our tests show that the Faber TAA model works, and we have high confidence it will continue to work in the future. The rules of the model are simple, and, being a monthly strategy, fairly low maintenance. As Mebane said in his paper, the objective of this model is to produce equity-like returns with bond-like drawdowns, and we agree that this strategy delivers on that vision.

I don't know that I would follow the strategy, but I wish I had been watching the 200-day SMA. I liquidated my only international ETF (SPEM) on Jan 30 when concerns about the virus in China started coming out, but didn't make the connection to the US impacts, panic, etc. and therefore did nothing with my US portfolio. :oops:
Can't see the point of treating US & ex-US differently. And I would use an established CTA for commodity & bond trend. There's probably a synergy from mixing market timing with buy & hold. One third each stocks, bonds & timing would work quite well. I do think the recent selloff has made month-end strategies obsolete however. If it's possible to lose 40%+ inside a month then the timing strategy needs to be at least divided by 4 and check weekly rather than monthly signals.

In the end it's better to use public funds for all this even if they're flawed.

klrjaa
Posts: 53
Joined: Sat Mar 23, 2019 10:55 am

Re: Meb Faber Timing Model

Post by klrjaa » Fri Mar 27, 2020 9:46 am

Forester wrote:
Wed Mar 25, 2020 1:11 pm
tinscale wrote:
Wed Mar 25, 2020 9:24 am
Of course, most (if not all) wish they got out right before all this mess.

I thought of timing models that rely on moving averages, and remembered Mebane Faber’s 2006 Tactical Asset Allocation Model. His system is:

- set up a portfolio of 5 ETFs — S&P 500, EAFE, a commodity index, a REIT index, and 10-year treasury index

- buy every ETF if its price is above the 10-month simple moving average

- sell the ETF if it drops below the 10-month simple moving average (and stay in cash until you get the signal to buy again)

Here’s an article/analysis dated March 21, 2020 that concludes the following:

Had you invested in this strategy from Jan 1973 to Mar 2020, your $100,000.00 would now be worth $7,357,812.39. That’s about a 9.53% compound annual return with a maximum drawdown of (12.87%). The MAR ratio is 0.74. The longest time you had to wait for it to recover from its previous equity peak is 27.1 months.

In contrast, had you invested in S&P 500, your $100,000.00 would now be $7,675,060.49, or 9.63% compounded, but you would have to suffer through a maximum drawdown of (55.25%) to get that. This yields a MAR ratio of 0.17. The longest time you would have to wait for it to make a new equity high is 73.5 months.

Our tests show that the Faber TAA model works, and we have high confidence it will continue to work in the future. The rules of the model are simple, and, being a monthly strategy, fairly low maintenance. As Mebane said in his paper, the objective of this model is to produce equity-like returns with bond-like drawdowns, and we agree that this strategy delivers on that vision.

I don't know that I would follow the strategy, but I wish I had been watching the 200-day SMA. I liquidated my only international ETF (SPEM) on Jan 30 when concerns about the virus in China started coming out, but didn't make the connection to the US impacts, panic, etc. and therefore did nothing with my US portfolio. :oops:
Can't see the point of treating US & ex-US differently. And I would use an established CTA for commodity & bond trend. There's probably a synergy from mixing market timing with buy & hold. One third each stocks, bonds & timing would work quite well. I do think the recent selloff has made month-end strategies obsolete however. If it's possible to lose 40%+ inside a month then the timing strategy needs to be at least divided by 4 and check weekly rather than monthly signals.

In the end it's better to use public funds for all this even if they're flawed.
Month end strategies were not made obsolete. There are many strategies that use signals different from moving averages, or use different moving average lengths, and therefore have been in risk off mode for months. See https://allocatesmartly.com/where-tacti ... day-03-23/

Regarding public funds, the facts don't bear it out when compared to DIY TAA strategies
https://allocatesmartly.com/this-crisis ... ical-etfs/

User avatar
Forester
Posts: 897
Joined: Sat Jan 19, 2019 2:50 pm
Location: UK

Re: Meb Faber Timing Model

Post by Forester » Fri Mar 27, 2020 10:09 am

klrjaa wrote:
Fri Mar 27, 2020 9:46 am
Forester wrote:
Wed Mar 25, 2020 1:11 pm
tinscale wrote:
Wed Mar 25, 2020 9:24 am
Of course, most (if not all) wish they got out right before all this mess.

I thought of timing models that rely on moving averages, and remembered Mebane Faber’s 2006 Tactical Asset Allocation Model. His system is:

- set up a portfolio of 5 ETFs — S&P 500, EAFE, a commodity index, a REIT index, and 10-year treasury index

- buy every ETF if its price is above the 10-month simple moving average

- sell the ETF if it drops below the 10-month simple moving average (and stay in cash until you get the signal to buy again)

Here’s an article/analysis dated March 21, 2020 that concludes the following:

Had you invested in this strategy from Jan 1973 to Mar 2020, your $100,000.00 would now be worth $7,357,812.39. That’s about a 9.53% compound annual return with a maximum drawdown of (12.87%). The MAR ratio is 0.74. The longest time you had to wait for it to recover from its previous equity peak is 27.1 months.

In contrast, had you invested in S&P 500, your $100,000.00 would now be $7,675,060.49, or 9.63% compounded, but you would have to suffer through a maximum drawdown of (55.25%) to get that. This yields a MAR ratio of 0.17. The longest time you would have to wait for it to make a new equity high is 73.5 months.

Our tests show that the Faber TAA model works, and we have high confidence it will continue to work in the future. The rules of the model are simple, and, being a monthly strategy, fairly low maintenance. As Mebane said in his paper, the objective of this model is to produce equity-like returns with bond-like drawdowns, and we agree that this strategy delivers on that vision.

I don't know that I would follow the strategy, but I wish I had been watching the 200-day SMA. I liquidated my only international ETF (SPEM) on Jan 30 when concerns about the virus in China started coming out, but didn't make the connection to the US impacts, panic, etc. and therefore did nothing with my US portfolio. :oops:
Can't see the point of treating US & ex-US differently. And I would use an established CTA for commodity & bond trend. There's probably a synergy from mixing market timing with buy & hold. One third each stocks, bonds & timing would work quite well. I do think the recent selloff has made month-end strategies obsolete however. If it's possible to lose 40%+ inside a month then the timing strategy needs to be at least divided by 4 and check weekly rather than monthly signals.

In the end it's better to use public funds for all this even if they're flawed.
Month end strategies were not made obsolete. There are many strategies that use signals different from moving averages, or use different moving average lengths, and therefore have been in risk off mode for months. See https://allocatesmartly.com/where-tacti ... day-03-23/

Regarding public funds, the facts don't bear it out when compared to DIY TAA strategies
https://allocatesmartly.com/this-crisis ... ical-etfs/
I think a DIY month-end model will be caught equally flat-footed as these disappointing ETF options, in a rapid sell off. To be fair most bear markets in the future should occur due to deteriorating financial markets, not external shocks.

klrjaa
Posts: 53
Joined: Sat Mar 23, 2019 10:55 am

Re: Meb Faber Timing Model

Post by klrjaa » Fri Mar 27, 2020 12:22 pm

Forester wrote:
Fri Mar 27, 2020 10:09 am
klrjaa wrote:
Fri Mar 27, 2020 9:46 am
Forester wrote:
Wed Mar 25, 2020 1:11 pm
tinscale wrote:
Wed Mar 25, 2020 9:24 am
Of course, most (if not all) wish they got out right before all this mess.

I thought of timing models that rely on moving averages, and remembered Mebane Faber’s 2006 Tactical Asset Allocation Model. His system is:

- set up a portfolio of 5 ETFs — S&P 500, EAFE, a commodity index, a REIT index, and 10-year treasury index

- buy every ETF if its price is above the 10-month simple moving average

- sell the ETF if it drops below the 10-month simple moving average (and stay in cash until you get the signal to buy again)

Here’s an article/analysis dated March 21, 2020 that concludes the following:

Had you invested in this strategy from Jan 1973 to Mar 2020, your $100,000.00 would now be worth $7,357,812.39. That’s about a 9.53% compound annual return with a maximum drawdown of (12.87%). The MAR ratio is 0.74. The longest time you had to wait for it to recover from its previous equity peak is 27.1 months.

In contrast, had you invested in S&P 500, your $100,000.00 would now be $7,675,060.49, or 9.63% compounded, but you would have to suffer through a maximum drawdown of (55.25%) to get that. This yields a MAR ratio of 0.17. The longest time you would have to wait for it to make a new equity high is 73.5 months.

Our tests show that the Faber TAA model works, and we have high confidence it will continue to work in the future. The rules of the model are simple, and, being a monthly strategy, fairly low maintenance. As Mebane said in his paper, the objective of this model is to produce equity-like returns with bond-like drawdowns, and we agree that this strategy delivers on that vision.

I don't know that I would follow the strategy, but I wish I had been watching the 200-day SMA. I liquidated my only international ETF (SPEM) on Jan 30 when concerns about the virus in China started coming out, but didn't make the connection to the US impacts, panic, etc. and therefore did nothing with my US portfolio. :oops:
Can't see the point of treating US & ex-US differently. And I would use an established CTA for commodity & bond trend. There's probably a synergy from mixing market timing with buy & hold. One third each stocks, bonds & timing would work quite well. I do think the recent selloff has made month-end strategies obsolete however. If it's possible to lose 40%+ inside a month then the timing strategy needs to be at least divided by 4 and check weekly rather than monthly signals.

In the end it's better to use public funds for all this even if they're flawed.
Month end strategies were not made obsolete. There are many strategies that use signals different from moving averages, or use different moving average lengths, and therefore have been in risk off mode for months. See https://allocatesmartly.com/where-tacti ... day-03-23/

Regarding public funds, the facts don't bear it out when compared to DIY TAA strategies
https://allocatesmartly.com/this-crisis ... ical-etfs/
I think a DIY month-end model will be caught equally flat-footed as these disappointing ETF options, in a rapid sell off. To be fair most bear markets in the future should occur due to deteriorating financial markets, not external shocks.
Hi Forester,

Thanks for the response. Some will but some won't. There are some very fast twitch DIY models that are specifically built to respond quickly, albeit with more whipsaws along the way. And even some of the slower twitch models use volatility and/or correlation and/or total weight of evidence in the asset selection criteria which will stay clear of heavy equity allocation months in advance of a fast moving market action, for better or worse.
Some folks tend to think the only means of providing protection is an N month moving average which is such a simplistic view of how TAA works. All good TAA models have a means to turn off risk via allocation to alternate assets, including cash. The salient point IMO is nobody knows, so spreading bets across models addresses the what, how, and when of an overall approach.

Not dancing on the grave, but I pointed out multiple times to Will and others any portfolio rooted in a singular strategy was nonsense. My concern is folks here will take Will's bad model as a proxy for the overall state of TAA effectiveness, and throw out the baby with the bathwater.

DonIce
Posts: 836
Joined: Thu Feb 21, 2019 6:44 pm

Re: Meb Faber Timing Model

Post by DonIce » Fri Mar 27, 2020 3:00 pm

Forester wrote:
Fri Mar 27, 2020 10:09 am
To be fair most bear markets in the future should occur due to deteriorating financial markets, not external shocks.
Why is that?

Post Reply