Antonacci Dual Momentum or M. Faber Ivy Portfolio... about the same?

[ochotona]

It seems to me that these two approaches are pretty much the same. Does anyone disagree? If you prefer one over the other, why?

The way I have been implementing M. Faber's ideas this year is:
1. Rank 10 securities* by summed 12 mo + 6 mo + 3 mo performance figures
2. Select the top 5 of 10, the other 2 slots go to cash
3. Furthermore, go to cash if 10 month moving average of adjusted closing prices
4. The 5 slot weights are all equal, 20%
5. Evaluate once a month, trade no more often than once a month

Antonacci sounds similar, except in his April 2014 SSRN paper he only looked at the 12 mo performance figure, which he used to rank, which is going to be quite similar to Faber's ranking metric, and he also used it as the risk kill-switch... buy if 12 mo performance > T-bill 12 mo performance, otherwise go to T-bills.

I looked at the timing signals for 10 mo moving average and the 12 mo performance > T-bill 12 mo performance and they are similar. The 10 mo MA has more head fakes, but also reacts faster to inflection changes at market tops and bottoms.

But otherwise, I can't see much difference between the two. Antonacci has a tiny bit less data gathering and computation, but not enough to make it materially easier.

Does anyone have any thoughts? Am I missing something?

* US large cap blend, US small cap blend, US REIT, ex-US REIT, ex-US developed blend, emerging market, Total Bond Market, Intermediate US Treasury, commodities, gold

[MachineGhost]

They both fire at different times and one can fire when the other does not.  Generally, I've found that the relative has inferior CAGR to the absolute.  Diversity is robustness.

Rotation strategies have done very poorly under QEternity.

[ochotona]

They both fire at different times and one can fire when the other does not.  Generally, I've found that the relative has inferior CAGR to the absolute.  Diversity is robustness.

Rotation strategies have done very poorly under QEternity.

But doesn't Antonacci discuss both relative and absolute momentum, just like one of the ways to implement Faber? There's the ranking aspect (relative) and the risk switch (absolute). It's in the April 2014 paper, I haven't gotten to the book yet.

I can see why QE short-circuits rotation. Just when things go sideways, they intervene, especially Super Mario who will do "whatever it takes", so you "porpoise" up-and-down around the moving average. Obviously, this can't go on forever, though it may go on for years and it's aggravating and a drag on a portfolio.

The alternative is sit there and take it, I'm not sure I like that either, because at some point all of this QE won't work any longer, and the asset bubble will finally pop.

[MachineGhost]

I can't remember, but relative in this case may be to T-Bills not a ranking of all funds, i.e. Robust Asset Allocation.  So you can do both ranking of the funds as well as dual triggers for risk control.

Have you seen this: http://decisionmoose.com/Moosistory.html

[Kbg]

Agree with MG...absolute generally helps. From Antonacci's book he compares returns using the S&P 500, 12 mo absolute mo and the 10 and 12 mo moving averages.  Abs mo is oh so slightly better than the two MAs but the difference could easily be noise. Abs mo had a higher sharpe (.69 vs. .68) but had a higher DD (29.58 vs. 23.26). So performance wise I'd say it was a wash. However, one big difference was trades per year. 1.2 per year for the 10 mo MA and .83 for Abs mo. My guess is the 12 mo trades per year is pretty close to the 10 mo, but I do find it interesting Mr. A doesn't provide the trades for the 12 mo MA explicitly in the book.

All in all I'd probably chose the one you can stick with better.

[ochotona]

I realized something was bothering me about both Faber's and Antonacci's works... they obviously have to backtest their methods on well-known publicly available sector indices. In their advice to individuals, they then cite well-known ETFs for home construction projects; VTI, VEU, VNQ, RWO, BND, IEF, DBC... their algorithms rotate these ETFs, turn them on and off.

But anyone who uses TDAmeritrade, Schwab, or Scottrade has access to scores or even hundreds of ETFs that trade for free. Why shouldn't the investor feel free to pick from the entire universe of tradeable ETFs, and not just the same circle of generic funds? Then if a given fund falls in the performance ranking as the months roll by, the investor would drop it and replace it with a fresh one from the current screening list.

Antonacci brings up a fault of Faber's but I'm not sure I like his solution, either. With Faber, you can end up with a pile of highly correlated ETFs at the same time. If they all have strong performance, and they're all above the 10 month MA, you'll get them. Antonacci breaks the portfolio into "modules", Equities, Credit, REIT, Stress. I'm just not sure if these buckets make sense to me. Equities and Credit I get. But like MG, I wonder why REIT stands apart from Equity; I'd rather have Equities and REIT funds fighting it out amongst themselves, competing for my risk asset Dollars, rather than reserve a cushy place for a REIT ETF to eat up my capital without a fight. And I don't understand why Mortgage ETF should compete against Equity ETF; I'm not sure I'd ever want to own a Mortgage ETF. Finally, I don't see why TLT and Gold have to do battle as the Stress pairing.

The problem I see with these little modules with pairs of things in each module is that if you choose the wrong pair, that's a problem, and it doesn't fix itself, it persists until you decide to change the pairing; but there is no methodology given (in the SSRN Dual Momentum paper... I am still waiting on the book). Two is an awfully small little universe.

Here are my thoughts:

Have three modules only. US Equities (40-60%), Intl Equities (20-40%), Bonds (20%). {Abs Momentum 80/20 will have no worse than passive 60/40 or passive 50/50 volatility; that's what I'm counting on}

Let all of the ETFs duke it out for my money in each of these three broad modules. Pick the top funds for each module, just make sure they're not in the same Morningstar categories. If there are zero choices meeting dual momentum criteria for a module, only then would that module go to cash. And I not consider weird categories (triple inverse small-cap Guatemalan...NO).

Mix REITs in with stocks. Delete the REIT category.

I think I'd just totally delete Commodities as a category at this point in time (Faber), every single commodity has been thrashed. That includes all metals. When they show some momentum in the future, re-instate. But why have it sitting there in cash for months / years, contributing to cash drag? It's not going to get better very quickly. I think I'd just keep any gold holdings off of the Momentum trading books, just keep track of them separately.

What I'm saying is, I don't care about backtesting. Backtesting can't predict my future, and when I'm retired, I'm not going to backtest to see how I did. I appreciate that Faber and Antonacci have backtested, to prove the general applicability of relative and absolute momentum. But I don't want to hamper myself by trading only generic, well-known ETFs just so I can minimize tracking error. Who cares about that? The only tracking I care about is the account balance. I think it's important to scan the entire universe of funds and use whatever is showing the strongest momentum for use in my momentum portfolio, rather than choose trackable but weaker funds.

[MachineGhost]

Generic, well-known ETF's have the highest liquidity, the lowest spreads and minimal tracking error with the NAV due to arbitrage.  Not everything in the same space has these virtures.  And generally the more negative virtures it has, the more volatile the price movements will be and the more fake momentum signals it will give.  You could probably deal with that by taking the momentum measurement and dividing it by the volatility for an efficiency ratio.

But I don't understand why you're not using the PP for the strategic portfolio weights.  THERE IS NOTHING BETTER!  Do you need more return or something?  Use a leveraged and/or cashless PP then.

Authors for some reason just don't know jack shit (is it that those who can't, teach, those who can, do?).  Either they keep their better ideas in reserve for their own use or the process of being an author makes them intellectually lazy.  They just don't dig deep enough or they would have known about clustering that disproves their stupid asset allocations matchups.  When you stop having your man crush on Faber, I believe the student will have surpassed the teacher.

[lordmetroid]

I choose to use all mutual funds and ETFs available to me, I simply sort them on the internet-broker homepage after the best 12 month %.

[ochotona]

But I don't understand why you're not using the PP for the strategic portfolio weights.  THERE IS NOTHING BETTER!  Do you need more return or something?  Use a leveraged and/or cashless PP then.

Authors for some reason just don't know jack shit (is it that those who can't, teach, those who can, do?).  Either they keep their better ideas in reserve for their own use or the process of being an author makes them intellectually lazy.  They just don't dig deep enough or they would have known about clustering that disproves their stupid asset allocations matchups.  When you stop having your man crush on Faber, I believe the student will have surpassed the teacher.

Regarding using the PP weights, I mentioned that I thought Commodities (including gold) are dead as momentum assets for a while, maybe another year? So removing the cash and the gold, we're down to a Momentum 50/50 Couch Potato portfolio; is that what you meant? Do you envision AbsMom on gold, LTT, and then DualMom on a universe of equities? Then, cash is just cash.

Given the reduction in drawdown resulting from the use of AbsMom, yes, I do want to swing for the fences until I reach my retirement target, then I will throttle back. I have 132 more months.

Oh, these guys never give away their complete recipe. My wife and I went to the famous Keo's Thai Restaurant in Waikiki (now sadly closed, after many years), then we bought the cookbook, and the recipes didn't work out, and my wife knows Thai cuisine. She concluded Keo left stuff out.

[ochotona]

I choose to use all mutual funds and ETFs available to me, I simply sort them on the internet-broker homepage after the best 12 month %.

It's so funny, there are so many warnings not to chase last year's winning fund. So ironic this is an actual strategy. It's as if you discovered cigarettes and liquor were good for you.

[MachineGhost]

Regarding using the PP weights, I mentioned that I thought Commodities (including gold) are dead as momentum assets for a while, maybe another year? So removing the cash and the gold, we're down to a Momentum 50/50 Couch Potato portfolio; is that what you meant? Do you envision AbsMom on gold, LTT, and then DualMom on a universe of equities? Then, cash is just cash.

Given the reduction in drawdown resulting from the use of AbsMom, yes, I do want to swing for the fences until I reach my retirement target, then I will throttle back. I have 132 more months.

I mean 25%x4 (or the Risk Parity PP weights which makes more sense with tactical) with the asset that is out of the market staying in cash (right now that is T-Bpnds and Gold according to my timing models), hopefully parked in the higher yielding MMSAs, breakable CD's, than in the gawd awful options available at brokers.  My MMSA pays 1% is far superior, so the money will stay there until its time.  Since this breaks my plan to autodeposit and autoinvest, it occurs to me that maybe we should extend the sequence of returns risk further out and not make tactical allocation a binary decision either.  Usually, hitting an extreme peak that causes a new signal precedes a shorter term pullback to work off the extreme sentiment anyway.  So I suggest dollar cost averaging in on a weekly basis over what the worst whipsaw chop period for the asset was.  How's that sound?

[MachineGhost]

Ah, well, I think only having one module lit up at a time isn't going to work very well as DecisionMoose's sore history shows, although I love the ideal but it just hasn't worked out in backesting without a lot of risk.  Momentum always increases risk, never reduces it.

The other problem is not having perpetual Treasury exposure leaves that spot from a peak to the exit signal completely unhedged.  That's a negative drag that the PP doesn't don't have.

Apparantely, I don't understand Antonacci's Dual Momentum at all if he's using two securities.  I thought it was about measuring momentum two different ways.  But I like the idea of whats called rotation among the best securities in any given asset class.  It just adds a layer of complexity to the PP and then there is the problem of deciding which of the various momentum ranking metrics works best historically.

[ochotona]

Ah, well, I think only having one module lit up at a time isn't going to work very well as DecisionMoose's sore history shows, although I love the ideal but it just hasn't worked out in backesting without a lot of risk.  Momentum always increases risk, never reduces it.

The other problem is not having perpetual Treasury exposure leaves that spot from a peak to the exit signal completely unhedged.  That's a negative drag that the PP doesn't don't have.

Apparantely, I don't understand Antonacci's Dual Momentum at all if he's using two securities.  I thought it was about measuring momentum two different ways.  But I like the idea of whats called rotation among the best securities in any given asset class.  It just adds a layer of complexity to the PP and then there is the problem of deciding which of the various momentum ranking metrics works best historically.

I've got about 20% of my assets in cash and US Treasuries at present. No plans to change, except buying physical gold on dips. So if I run dual momentum to screen and turn on/off the global equity opportunity set for the other 80%, I think I'll be OK. If you look at OptimalMomentum.com, Antonacci publishes the yearly performance for the GEM portfolio, a ballsy hop from US stocks to International Stocks to Bonds, only one thing at a time, and it's not any more volatile in standard deviation terms than a passive 80/20, and many fewer negative years. And the performance is huge, and he's just using three indices. What if you constantly look for best-in-class assets every month? There's a lot of potential, if you don't make stupid choices, like put it all in a small tiny illiquid markets, like a Cocoa ETF, a Carbon ETF (WTF!?!) or Japan Small Cap Healthcare ETF, which are all really hot now.

Yes, it's confusing, dual momentum does refer to two uses of momentum; to rank and to turn off and on. The fact that he had contests in modules pairing up "two" things against each other made me wonder at first if that's what he meant by "dual". But, no. That would be silly, actually. I think he just did it to make a point, and stop and not give away the really good recipe for Pad Panang or Tiger Cried.

I am posting 200 day MA in my screener, and I notice that many ETFs which are a "GO" from a 1-year performance basis are below the 200-day MA. We'd expect this because 200-day MA acts faster than 1-year performance, but it is sobering to see it on live data after the late August debacle. I'm not sure what to do about 200 day MA. It is extra protection against bear markets, or a bothersome PITA that keeps you from good oppurtunities? I have some time before I plan to go live with Faber or Antonacci, I will think about it more over the holidays after Santa brings the Dual Momentum book. My 401(k) can't rollover until February 2016, so not in a hurry.

[MachineGhost]

Yes, it's confusing, but dual momentum does refer to two uses of momentum; to rank and to turn off and on. The fact that he had contests in modules pairing up "two" things against each other made me wonder at first if that's what he meant by "dual". But, no. That would be silly, actually. I think he just did it to make a point, and stop and not give away the really good recipe for Pad Panang or Tiger Cried.

Oh okay, then I was thinking of triple momentum: relative strength, relative (vs T-Bills), absolute (vs MA).

GEM is the longest asset rotation backtest I've ever seen.  Are the rules for GEM available anywhere besides the pricey book?

Notice the consistent underperformance of GEM since 2009.  I get the feeling that despite the rally in U.S. stocks since then, some other asset subclass has been mostly overlooked in illustrative portfolios.

I am posting 200 day MA in my screener, and I notice that many ETFs which are a "GO" from a 1-year performance basis are below the 200-day MA. We'd expect this because 200-day MA acts faster than 1-year performance, but it is sobering to see it on live data after the late August debacle. I'm not sure what to do about 200 day MA. It is extra protection against bear markets, or a bothersome PITA that keeps you from good oppurtunities? I have some time before I plan to go live with Faber or Antonacci, I will think about it more over the holidays after Santa brings the Dual Momentum book. My 401(k) can't rollover until February 2016, so not in a hurry.

Absolute (vs MA) is better than Relative (vs T-Bills) in my backtesting, but I think robustness is preferable over greed.  It's also mentally easier to only put 50% in on a signal instead of 100%.

I actually have a variety of different momentum measurements and rankings for all stocks and some categories on my beta web site, but I can't be bothered to fix that right now.  Have other fires to put out first.

With what I use for general market timing nowadays (trend + volatility), only COMP and DIA have confirmed their uptrends (and last week).  SPX and R2K are still lagging as are most of the broad sectors.  Until I see broad participation, this is a dead cat bounce.  OTOH, we could easily do 1998 to 2000 again where MegaCap got all the attention to the detriment of everything else due to a stronger dollar and higher relative interest rates.  We're already in second place historically for valuation extremes, not yet surpassing 2000.  Bubblicious!

I couldn't invest in this crazy market without some kind of downside protection.  No ifs ands or buts.  Its stressful compared to a lazy portfolio but not as much as large MaxDD's that could have been prevented by not being lazy.

[MachineGhost]

To which I say... why not just use TMOM in all cases, except EAFE? Isn't that simpler than trying to engineer a kluge?

Robustness.  Nothing works all the time.  It's not that big of a deal to use two slightly different signals.  If there were a significant difference between Absolute and Trending instead of marginal, I'd say you have a case.

I swear Gray and company keep changing their labels.

So essentially their "Robust Asset Allocation" aka "Downside Protection" is a synthesis of Antonacci's absolute momentum (relative to T-Bills), Faber's absolute momentum (relative to MA), using Faber's dumb strategic asset classes and weightings, but without using Antonacci's relative strength rotation.  I trust I have it all clear by now.  Let's call all three the Triple Momentum Enchilada!

[ochotona]

By the way, pretty much nothing in the way of risk assets is above the 10 month moving average now. If things don't get better fast, this is going to be a

[align=center] CASHY NEW YEAR!  [/align]

[ochotona]

The problem with computing the trailing one year performance (TMOM) is that a spike up or down a year ago skews your results today, and it could be just market noise, like ISIS blows up a puppy farm or mows down a Quaker Meeting. Seems to me that a moving average smears away all of the noise.

Here's a simple way to get very close to a ten calendar month moving average of adjusted closes without having to download data and average it. Compute the 210 day moving average, which you can do on any chart. Subtract 1.3x the quarterly dvidend. For the SP500 recently, it's very close. Or, for government accuracy, just subtract 1x the quarterly dividend.

[ochotona]

Time-Series Momentum versus Moving Average Trading Rules


Ben R. Marshall
Massey University - School of Economics and Finance

Nhut H. Nguyen
Massey University

Nuttawat Visaltanachoti
Massey University - Department of Economics and Finance

December 22, 2014


Abstract:     
Time-series momentum (TSMOM) and moving average (MA) trading rules are closely related; however there are important differences. TSMOM signals occur at points that coincide with a MA direction change, whereas MA buy (sell) signals only require price to move above (below) a MA. Our empirical results show MA rules frequently give earlier signals leading to meaningful return gains. Both rules perform best outside of large stock series which may explain the puzzle of their popularity with investors yet lack of supportive evidence in academic studies.

Number of Pages in PDF File: 45

Keywords: G11, G12

JEL Classification: Technical analysis, time-series momentum, moving average, return predictability

[ochotona]

The Dual Momentum book is nice; nothing new from a content point of view, but good to have it in one place, in hardcopy.

Takeways:

The difference between 10 month moving average and 12 month total return as absolute momentum risk controllers are small. Not worth worrying about. Go with the easier, lazier measure... 12 month total return.

The Global Equities Momentum (GEM) portfolio is very cool - high return, high Sharpe (0.87), moderate MaxDD. Three components: VTI, VEU, BND (for me, SCHB,SCHF, SCHZ). You just rotate between the three.

You can do great with less risk if you have 30% in bonds and use Dual Momentum to rotate between 10-year Treasuries, High Yield, Corporate bonds, and Cash, this he calls Globally Balanced Momentum (GBM). The Sharpe is even higher at 0.98.

Finally, using single Absolute Momentum only to rotate in-and-out of an equally weighted US Stocks, Intl Stocks, Gold, 10-y Treasuries, Corporate bonds he gets returns which are similar to those when Meb Faber applied to 10 month moving average risk control to the Permanent Portfolio. This he calls the Parity Portfolio. Lowest MaxDD and highest Sharpe (1.06) of all three portfolios, I think this would be awesome for a retiree. They would end up being a very rich, comfortable retiree relative to the static 40/60 allocation.

Dual Momentum by Gary Antonacci. Get it, Read it. Place it in your library next to that other ground-breaking work, the Permanent Portfolio by Rowland and Lawson.

[ochotona]

The GEM portfolio end-of-month signal was to exit US equities today, and WHAM! Luckily, I was 83% cash and bonds already. Put in stop-limit sell orders for tomorrow for my remaining piece.

Faber points out that the 10 month moving average is also flashing bearish.

CASHY NEW YEAR!

[ochotona]

Thanks to Tyler for a Pixel chart of the GEM portfolio. This one totally kicks the PP to the curb as far as growth is concerned. You can quiet it down a lot more by adding gold, cash, and Treasuries.

[Tyler]

Lots of greeeeeen!

That said, one should be aware that once you account for trading fees and taxes for any momentum-based portfolio, the end result you see in your bank account may look a lot different than the one on the chart.  Be smart about it. 

[ochotona]

Yes, I trade commission-free ETFs in an IRA. Costs are low, taxes not an issue. On average, this portfolio reportedly trades less than once a year, but we just had a whipsaw, out on Sep 1, in on Nov 1, out again on Jan 1. I hope we're done whipsawing for now. This market has been trying to figure things out.

[lordmetroid]

I check my balance every day, usually more than once a day. I did this when I had the pemanent portfolio as well which is why such a passive investment scheme isn't suitable for me.

After the recent small downturn in mid December, I had my gains from the late November almost disappear because I didn't exit in time. So I decided to be even more active in my investments.
I now trade my fund on a daily basis. If the fund is worth below it's SMA( 8 ) I sell and cash out my gains. Then I wait until the trend reverses again. Considering I am trading using the 1 year % best fund, I am always betting with the trend.

So far so good, I ended up 2% positive for 2015 even though I had a Permanent portfolio  for most of the year which lost me 5% before I decided to switch strategy.
Taxes and fees are a non issue for me. I get taxed 0,42% of the total assets in my trading account for the year(the rate is decided once a year based on the Swedish state lending rates) and my broker do not have any fees for selling and buying mutual funds.


@Tyler thanks for the chart. Didn't see it being posted anywhere else but that is an awesome chart.

[ochotona]

He took the raw time series and processed it into a chart, that's just magic.