Author Topic: Adaptive Asset Allocation- Thoughts?  (Read 5993 times)

NorCal

  • Handlebar Stache
  • *****
  • Posts: 1464
Adaptive Asset Allocation- Thoughts?
« on: September 03, 2014, 09:10:18 AM »
I recently read an interesting paper on "Adaptive Asset Allocation".  To simplify it somewhat, you have a basket of 10 asset classes, and invest in 5 best performing assets (US Stocks, European Stocks, Bonds, etc.), and weight each position using a minimum variance calculation.  A link to the white-paper is below.

On the surface, it seems to have a better risk/reward profile than simple indexing.  It is certainly better diversified.

Due to the frequent re-balancing, it would really only make sense in a tax-deferred portfolio.

I'm currently experimenting with a version of this in one of my retirement accounts, but I will be re-balancing only when the suggested portfolio changes dramatically.

Does anyone have any thoughts or experience with this type of portfolio?

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2328254

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Adaptive Asset Allocation- Thoughts?
« Reply #1 on: September 03, 2014, 06:38:07 PM »
Yes, I have a lot of experience with it and trade a system like this in both my retirement and taxable accounts. For me it's not about outright performing better than the market, but rather protecting the downside by not holding an asset that is continually going down in price, with no idea when or if it will return.

I'll share more after I put the boys to bed.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Adaptive Asset Allocation- Thoughts?
« Reply #2 on: September 03, 2014, 08:50:26 PM »
So my first question would be, how much experience do you have with mechanical trading systems? Statistics? Backtesting? Robustness testing? Market history? Trading psychology? Momentum? I'm not saying you have to be an expert in all those areas, but it sure doesn't hurt.

The problem you'll run into with this system, or any system really, is that it will go through a period where it doesn't do well. Or, it at least doesn't do well compared to the general market. You'll then be tempted to throw the system in the trash and put your money in index funds which have been doing well for some time. Then, shortly after that, just when you've written off that stupid AAA (adaptive asset allocation) strategy, the market tanks and AAA starts to perform well again. You realize you were dumb for ever doubting AAA, and you put your money back into the strat. And then the cycle starts all over again...

Strategies are easy (relatively). The hard part is sticking to them when they suck (which they will at some point). This specific AAA strategy has provided stable returns but has underperformed the S&P the past 3 years. Will you continue to trade it, month in and month out, for 3 straight years? While your friends are making more money than you being 100% invested in US stocks? While they mock you for all that hard work but less returns? Will you understand where the risks and returns come from in a system like this, and when they come?

What I'm getting at, is that yes, I trade a system very similar to this paper. I am indebted to the guys who wrote it, and all their papers and blog posts really. They helped me understand concepts (and the math behind them) that I could not find elsewhere at the time. However, I spent literally thousands of hours understanding all the ins and outs of diversified trend following and momentum trading before I started live trading it. I know what to expect, and I have reasonable expectations based on all the research that I've done.  It fits my personality and I know what I'm getting myself into. I would never tell someone else to trade my strategy unless they had the same conviction I had and understood how it works and what to expect.

Don't ever blindly trade someone else's system. Ever. You need to trade the system that fits you and your personality. You'll have to go on a soul-searching mission to really figure that out. Personally, the system as presented in this paper is too complex IMO. Simpler is almost always better. You can do just as good or better on the return side by intelligently deciding which assets to trade (this is the most important and takes the most research and thought time), then just buying the assets that have done well recently and holding them equal weight. No correlation, no risk-parity, no volatility targeting or mean-variance whatever. Simple. Works. The more complex, the easier it is to break. Simpler is more robust. But that's just me.

I recommend reading everything from Mebane Faber and Gary Antonnacci as they have papers, books and blog posts dealing with these types of diversified momentum systems. I also recommend reading Michael Covel's books, Following the Trend by Andreas Clenow, and The Way of the Turtle by Curtis Faith. You need to know the good, the bad and the ugly. Know WHY something works, not just how it works.

NorCal

  • Handlebar Stache
  • *****
  • Posts: 1464
Re: Adaptive Asset Allocation- Thoughts?
« Reply #3 on: September 03, 2014, 10:09:15 PM »
Thank you for the insight.  This is very helpful.

I do have a basic background in the statistics and modelling.  At least I did when I finished my MBA about 5 years ago.  I modeled out the Turtle system for class, and still have the "Turtle Trader" and "Trend Following" on my bookshelf.  I was a little rusty on my correlation charts and volatility measurements, but it came back pretty quick.

I have a different psychological approach to this than you do.  To me, the most valuable aspects of this particular approach is risk management and smoothing returns over time.  It's a strategy that's statistically likely to trail the market in the super-good years, but it will more likely outperform in the so-so years, and might even go up in years the market is down.  At least it should go down less than the overall market.  If I can manage a 10 year CAGR comparable to index funds without the ~50% drawdowns of the stock market, I will be happy.

I'm testing it out with a minimum variance portfolio until I get more comfortable.  Then I may decide to add more risk over time. 

The one aspect I need to learn better is backtesting.  Are there any particularly good backtesting products you've used?  I haven't found anything that I like yet.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Adaptive Asset Allocation- Thoughts?
« Reply #4 on: September 05, 2014, 12:40:30 PM »
ETFReplay.com is a good place to start backtesting ideas like this. Also there's quite a few tests back to the 70s and 80s in the research papers from Mebane Faber and Gary Antonacci.

Jags4186

  • Pencil Stache
  • ****
  • Posts: 587
Re: Adaptive Asset Allocation- Thoughts?
« Reply #5 on: September 05, 2014, 01:48:23 PM »
Sounds an awful lot like market timing to me.

Back testing complex investment strategies sounds like a recipe for disaster.


hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Adaptive Asset Allocation- Thoughts?
« Reply #6 on: September 05, 2014, 02:20:31 PM »
I understand most on this board look at anything beyond buy and hold index investing as market timing, and that's ok. I would never tell someone to venture outside of a diversified set of index funds unless they had done the research themselves.

The momentum effect has been researched time and time again throughout all the finance world and has probably more support than any other factor out there. The same finance professors that preach efficient markets and index investing also recognize the outperformance of momentum. It's persistent across all worldwide markets in stocks, bonds, commodities, currencies and real estate. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2435323

What we're talking about here is essentially a momentum system with a bunch of fancy weighting mechanisms to reduce volatility. As I said before, this particular system is fairly complex for my taste and works just fine keeping it as simple as possible. But no simpler.

Like I initially posted, I do not use this system to try to outperform the market. I use it for risk management to keep me in asset classes that are going up, and out of ones that are going down. It just so happens that systems like this can outperform the total returns of index funds simply by missing out on the downside. It's the black swan years that really do a number on returns.

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Adaptive Asset Allocation- Thoughts?
« Reply #7 on: September 05, 2014, 03:10:41 PM »
I understand most on this board look at anything beyond buy and hold index investing as market timing, and that's ok. I would never tell someone to venture outside of a diversified set of index funds unless they had done the research themselves.

The momentum effect has been researched time and time again throughout all the finance world and has probably more support than any other factor out there. The same finance professors that preach efficient markets and index investing also recognize the outperformance of momentum. It's persistent across all worldwide markets in stocks, bonds, commodities, currencies and real estate. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2435323

What we're talking about here is essentially a momentum system with a bunch of fancy weighting mechanisms to reduce volatility. As I said before, this particular system is fairly complex for my taste and works just fine keeping it as simple as possible. But no simpler.

Like I initially posted, I do not use this system to try to outperform the market. I use it for risk management to keep me in asset classes that are going up, and out of ones that are going down. It just so happens that systems like this can outperform the total returns of index funds simply by missing out on the downside. It's the black swan years that really do a number on returns.

+1

I'm a huge fan of momentum, and particularly love the research of Gary Antonacci.

I am currently thinking about using a monthly absolute momentum screen in conjunction with a more aggressive equity exposure in my 403B in order to improve upside and (more importantly) reduce drawdown risk.

I particularly enjoyed this paper.

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2244633

NorCal

  • Handlebar Stache
  • *****
  • Posts: 1464
Re: Adaptive Asset Allocation- Thoughts?
« Reply #8 on: September 06, 2014, 05:08:35 PM »
Having studied both EMH and some momentum trading, I think there is value in both.  Momentum strategies typically have good years and bad years at different times than the general market.  So even a diehard EMH adherent would agree that counting a momentum portfolio as a separate asset class in a larger portfolio could increase returns or reduce risk.

Most of the serious literature I've read on momentum trading generally claims a long term CAGR somewhere around 15%.  I don't think this is an accident. Since momentum generates mostly short term capital gains, and the marginal tax rates on short term is near 50%, the big money is still seeing an after tax CAGR of 7-8%.  So it's really a market strategy that can generate a higher pre tax return than indexing but a comparable post tax return.

Of course, this creates opportunities for those who can invest in a tax deferred account.

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Adaptive Asset Allocation- Thoughts?
« Reply #9 on: September 06, 2014, 05:31:14 PM »
Having studied both EMH and some momentum trading, I think there is value in both.  Momentum strategies typically have good years and bad years at different times than the general market.  So even a diehard EMH adherent would agree that counting a momentum portfolio as a separate asset class in a larger portfolio could increase returns or reduce risk.

Most of the serious literature I've read on momentum trading generally claims a long term CAGR somewhere around 15%.  I don't think this is an accident. Since momentum generates mostly short term capital gains, and the marginal tax rates on short term is near 50%, the big money is still seeing an after tax CAGR of 7-8%.  So it's really a market strategy that can generate a higher pre tax return than indexing but a comparable post tax return.

Of course, this creates opportunities for those who can invest in a tax deferred account.

I tend to agree with your concern of the tax inefficiency of momentum strategies. On the other hand it's proponents point out that you tend to take a lot of short-term losses along with long-term gains since by definition you hold winners and sell losers.

It's sort of a moot point for me since about 85% of my portfolio is in retirement/HSA/college savings accounts.

What I find so attractive about dual momentum strategies as well as absolute momentum strategies  are  their incredible downside protection. On some level you have the best of both worlds, incredible upside potential during bull markets, and diminished draw downs during market corrections and bear markets. 

I have been using a simple ETF momentum strategy in my HSA account this year and I've been very impressed. I started investing in February with $6550 and as of close on Friday the account currently stands at $7200. Or a better than 10% appreciation.

What was particularly impressive was making trades but I would never have dreamed of making like trading out of US stocks and into long-term treasuries a couple months ago.  (The account is currently 100% emerging markets)

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Adaptive Asset Allocation- Thoughts?
« Reply #10 on: September 07, 2014, 06:37:17 AM »
Agree on everything above. Great thoughts guys. Taxes are a huge consideration here.