Tactical Asset Allocation + HBPP an intriguing combo

A place to talk about speculative investing ideas for the optional Variable Portfolio

Moderator: Global Moderator

User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

not so fast

Post by ochotona » Sat Jan 19, 2019 5:55 pm

I paid my $35 for Economic Pulse, and found from studying past issues that the models are twitchier. They trade more often. For example, their Developed Markets DM-COMP model traded five times in 2018, and got whipsawed four of those five times in the course of a month... GEM traded once. They did about the same returns for 2018.

You have to understand your temperament as an investor. I hate to trade, buyers or sellers remorse builds up in my soul after a trade, inevitably. In a few months, I forget about it. But it deteriorates my life for a while.

Someday I will be 100% in a lazy portfolio, maybe with a trend-following ETF like PTLC, where I don't have to trade. Why I don't do it now... it's not an internationally diversified security. It's costly, for me hundreds per month. GEM is free. Economic Pulse is $35 per month. I'm waiting for iShares to offer the same thing for less money. Then I'm gone.

2019 is key for me. I will keep subscribed to Economic Pulse, but I will watch how it reacts to what could be working out to be a gigantic bull trap... the sucker rally or rallies that pull in the retail investor before the drain is pulled out of the bathtub. We are going to have a recession someday, that is 100% certain. I'm confident both models would react properly to a true recessionary bear market, I just don't want to go through extra torture to get there.

If Economic Pulse gets whipped by a sucker rally, and if it whips a great deal in 2019, I won't use it.

Plus I am comparing the economic data with what I can harvest from elsewhere, namely Ned Davis Research Market Digest (free to me because I'm at Schwab). If they line up throughout 2019... I'll go with the free source of information.

So I'll keep posting GEM signals here, as long as I keep using them.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Dual Momentum GEM + HBPP a great combo, easy to test

Post by ochotona » Sun Jan 27, 2019 6:15 am

My hair is standing on end. Look at Corey Hoffstein's Threader tweetstorm... scroll down to where the text says "See US returns in the late 1930s".

The 12-mo lookback, the heart of GEM, did not hold up that well for the USA during the Great Depression. Very disturbing, FTSE it was much worse than the market in the 1900-1920 period. Germany, terrible - well, they lost WW2, their own damn fault. But Australia in the late 1980s? Sheesh.

I'm going back to bed.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Dual Momentum GEM + HBPP a great combo, easy to test

Post by ochotona » Sun Jan 27, 2019 9:01 am

I've had PAA on my mind for a while.

From AllocateSmartly.com: "The Protective Asset Allocation (PAA) strategy from Wouter Keller and JW Keuning’s paper: PAA: A Simple Momentum-based Alternative to Term Deposits. The strategy is notable for its aggressive use of a “crash protection” asset that has resulted in extremely low drawdowns relative to return."

Drawdowns about 35% as bad as the GEM, for example. 66% as bad as HBPP. But with 60/40-ish CAGR. That sounds nice. I can totally live with that!

It watches 12 ETFs, if 6 of them go down ("down" means below the respective ETF's 13 month SMA), then the whole thing crashes out. It's totally crashed out now. Another nice feature of the crash score is that it brings back the risk assets in steps; 1/6, 2/6, 3/6,...., 6/6.

When it starts coming back, you start allocating to the up to Top 6 ETFs by momentum score. That's kind of a pain, there about 4 trades per month, but the important thing for me is the % allocation to risk assets.

It's not hard to calculate, it's totally not worth playing AllocateSmartly.com for this. I bought a membership to AllocateSmartly in 2017 so I could spy on what they were doing, I liked what I saw with PAA, then I quit. I have a Google Sheet for Keller & Keuning Protective Asset Allocation

Just to have a look at the crash score, I just put in +1 and -1 for the above and below 13 mo SMA. I didn't want to pick numbers off of charts, hand-enter them, etc. Maybe someone can tell me how to automate it.

I doubt PAA is going back to risk assets on 1/31/2019, not unless the heavens open up.

I like that PAA looks for suckage throughout the world, across various asset classes. GEM and Economic Pulse both have this built in: "If things are OK in the USA, then it's risk-on baby!" I don't know if that works going forward. The US is 25% of equity market cap (the index used by VT filters out a lot of stuff). The US is about 40% of world GDP.

I changed to title of this entire thread... because we're veering into other TAA methods, not only GEM, and I'm fine by that. We should be a learning community.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Sun Jan 27, 2019 8:54 pm

Ocho,

The point is method diversification first and foremost. Switching to another single method has all the same problems as just sticking with GEM. One small anecdote...I use GEM and have it in several retirement accounts and update one per every six weeks. Just last year I had one account up 5.5% and another down 7% with the rest somewhere in between using the exact same system.

I've looked at this a lot and very simply, pick a time period that is relatively short and another that is relatively long. Another thing to do is to tranche which is I what I described above. I do both (though the tranching is primarily driven by account trading/holding restrictions).

You mentioned you had AS for a while, remember those start date variability witch's broom charts. Diversification of process and timing aims to be in the middle of what is available/possible.

Side note: PAA has moved on to VAA and DAA and DAA on steroids. Google trendxplorer. All are variations of the same basic idea. If you like GEM check out VAA-G4. If you move to a G12 variant of any of them realize you will never have a blowout good year and great patience is required because if your benchmark is the US stock market you are in for yearly frustration. A more appropriate (basic) benchmark would be perhaps the PP itself. In baseball terms, you are going to get lots of singles and some bunt outs, no grand slams and no shut outs. And if you don't like trading, then definitely not a good system as the 13612W is like a very quick light switch. Full on, to full off, to full on in three month's time is absolutely a normal and frequent occurrence.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Mon Jan 28, 2019 1:15 pm

Will reply later when I have time, but brief look at portfoliovisualizer.com reveals... you have to go to a 4 month lookback to avoid the nastiness of December 2018. 5 months won't do it. 4 month 75% weight, 12 month 25% weights works well. Better performance than standard GEM. Doesn't trade too much for me. I will try 4, 6, 8, 12 later. Weights will be critical.

DAA canary universe... VWO, BND? Interesting results, but the authors themselves say they can't explain why it works. That gives me pause, but it's an interesting observation.
Last edited by ochotona on Mon Jan 28, 2019 2:09 pm, edited 1 time in total.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Mon Jan 28, 2019 1:47 pm

BND is serving the same function as return on cash in DM. VWO, they pretty much say it is an artifact of data mining. I'm gonna guess two underlying fundamentals for emerging. 1) Emerging has always been on the high end of equity risk so it seems plausible it's one of the first things to get reeled in when risk off is coming. 2) Emerging has also historically been tied to commodity prices which of course nose over when manufacturing/the economy begins to turn south.

Looks like you read the paper...I think the "proof" that "a" canary universe works was about as established as anything can be in this field.

I think far more important than emerging equities is 13612W and the assets chosen (eg G4 or G12)
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Mon Jan 28, 2019 2:10 pm

Kbg wrote:
Mon Jan 28, 2019 1:47 pm
BND is serving the same function as return on cash in DM. VWO, they pretty much say it is an artifact of data mining. I'm gonna guess two underlying fundamentals for emerging. 1) Emerging has always been on the high end of equity risk so it seems plausible it's one of the first things to get reeled in when risk off is coming. 2) Emerging has also historically been tied to commodity prices which of course nose over when manufacturing/the economy begins to turn south.

Looks like you read the paper...I think the "proof" that "a" canary universe works was about as established as anything can be in this field.

I think far more important than emerging equities is 13612W and the assets chosen (eg G4 or G12)
Trading G12 would drive me crazy. I'm trying to stay with a simpler investible universe.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Mon Jan 28, 2019 3:48 pm

OMG that 1 3 6 12 "normalized" (annualized) lookback is a thing of beauty! Four trades a year, on average. Wow, but it works great! Click here to see.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Mon Jan 28, 2019 7:49 pm

The Weighting is more like 54,27,13,6...
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Mon Jan 28, 2019 7:57 pm

ochotona wrote:
Mon Jan 28, 2019 3:48 pm
OMG that 1 3 6 12 "normalized" (annualized) lookback is a thing of beauty! Four trades a year, on average. Wow, but it works great! Click here to see.

The only thing better than 1 3 6 12 is 2 3 6 12. What a strange demonstration of timing luck, fragility, whatever you want to call it.

WOW.. the 2 3 6 12 gets you out before the December 2018 debacle. You need TWO shorter period look-backs to gather enough performance statistics to exit end of November 2018. 1 3 isn't short enough, 1 2 is.

And 2 3 6 12 is better even if you exclude 2018. Trades 3x annually, not 4x. I like it.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Tue Jan 29, 2019 9:35 am

Ref 2 3 vs 1 3...you want distinct time differentiation. That’s kinda the point if you are trying to diversify.

You could go 2 4 8 12 16 or whatever, but if you are seeking to mitigate the unknown then be distinct. Something like 3 4 or 11 12 isn’t doing much diversification.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Tue Jan 29, 2019 3:50 pm

Kbg wrote:
Tue Jan 29, 2019 9:35 am
Ref 2 3 vs 1 3...you want distinct time differentiation. That’s kinda the point if you are trying to diversify.

You could go 2 4 8 12 16 or whatever, but if you are seeking to mitigate the unknown then be distinct. Something like 3 4 or 11 12 isn’t doing much diversification.
OK, but interesting thing, if you do 3 lookbacks, 11 12 13 months, you end up with CAGR / Drawdown / Sharpe results like a 12 month lookback, not surprising, but 20% fewer trades. You suppress the short-term noisiness of the signal by using the 1 month +/- dither. Less noise = less whipsaws.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Tue Jan 29, 2019 5:50 pm

ochotona wrote:
Tue Jan 29, 2019 3:50 pm
Kbg wrote:
Tue Jan 29, 2019 9:35 am
Ref 2 3 vs 1 3...you want distinct time differentiation. That’s kinda the point if you are trying to diversify.

You could go 2 4 8 12 16 or whatever, but if you are seeking to mitigate the unknown then be distinct. Something like 3 4 or 11 12 isn’t doing much diversification.
OK, but interesting thing, if you do 3 lookbacks, 11 12 13 months, you end up with results like a 12 month lookback, not surprising, but many fewer trades. You clobber the noisiness of the signal by the 1 month +/- dither.
Valid...in this case you are doing something very similar to a 3 period moving average so you are getting some smoothing and reducing the odds for one off daily/monthly return spikes that throw a signal. The cost is a bit of lag.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Tue Jan 29, 2019 6:50 pm

Not sure I agree any longer with Antonacci's assertion that you don't need stop-losses with GEM. The way I see it, an -7% stop-loss would have stemmed the bleeding somewhat in December, and would on average only fire off every five years on average. Any the investor doesn't have to do anything other than put the order. Hindsight is truly 20-20 !
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Tue Jan 29, 2019 7:58 pm

Stop losses always hurt system performance...and never use them on a LT system without equally clear reentry rules.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Wed Jan 30, 2019 8:30 pm

After lots of fussing, I find I like GEM 3 6 9 12. I'm going to use it. The link is here. Fortunately, I'm 100% compliant with it right now... nothing to do.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Thu Jan 31, 2019 8:09 pm

The regular GEM and 3-6-9-12 month (annualized performance scores) GEM are still both in bonds for February 2019.
HappyMan
Senior Member
Senior Member
Posts: 132
Joined: Thu Nov 09, 2017 7:01 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by HappyMan » Thu Jan 31, 2019 10:18 pm

ochotona wrote:
Thu Jan 31, 2019 8:09 pm
The regular GEM and 3-6-9-12 month (annualized performance scores) GEM are still both in bonds for February 2019.
Thanks for letting us know!

Trying to figure out how you conclude from the link to the Visualised Portfolio whether GEM 3-6-9-12 month needs to change?
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Fri Feb 01, 2019 5:09 am

HappyMan wrote:
Thu Jan 31, 2019 10:18 pm
ochotona wrote:
Thu Jan 31, 2019 8:09 pm
The regular GEM and 3-6-9-12 month (annualized performance scores) GEM are still both in bonds for February 2019.
Thanks for letting us know!

Trying to figure out how you conclude from the link to the Visualised Portfolio whether GEM 3-6-9-12 month needs to change?
You go to the "timing periods" tab, but it posts the next day. I also checked it by hand last night using PerfChart on Stock charts.com. I multiply the 3 mo score by 4, 6 mo by 2, and 9 mo by 1.3333 then divide the sum by 4.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Sun Feb 03, 2019 8:56 am

By the way the DAA risk filter was

100% cash October November
50% cash December January
100% risk-on February

Weird.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Sun Feb 03, 2019 7:13 pm

ochotona wrote:
Sun Feb 03, 2019 8:56 am
By the way the DAA risk filter was

100% cash October November
50% cash December January
100% risk-on February

Weird.
Ignore what I wrote. I think the DAA risk filter is complete BS. The time series makes zero sense.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Wed Feb 06, 2019 7:01 am

ochotona wrote:
Wed Jan 30, 2019 8:30 pm
After lots of fussing, I find I like GEM 3 6 9 12. I'm going to use it. Fortunately, I'm 100% compliant with it right now... nothing to do.
More fussing, and going back to Occam's Razor. The GEM with 4 6 8 10 12 month lookbacks, "normalized" (annualized) weights works great, but didn't catch December 2018. But to catch December 2018 you have to heavily-weight 4 months, which to me seems like over-tuning the model to solve just one problem, which might break something else.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Wed Feb 06, 2019 8:34 am

ochotona wrote:
Wed Feb 06, 2019 7:01 am
More fussing, and going back to Occam's Razor. The GEM with 4 6 8 10 12 month lookbacks, "normalized" (annualized) weights works great, but didn't catch December 2018. But to catch December 2018 you have to heavily-weight 4 months, which to me seems like over-tuning the model to solve just one problem, which might break something else.
You are going down a dark alley there my friend. I recommend you turn around as you have accurately diagnosed the problem. I think it best to simply understand the differences in lookback length and what that means for the system. Academically, anything between 3-12 months "works." And exactly none of us know what is going to work best in the future. It's a futile exercise. But here's what we do know:

1. The longer the MA or lookback, the longer the lag and the more draw down, when it happens, you are going to get
2. Longer terms yield fewer trades (corollary...and fewer taxable events)
3. Longer terms better exploit major market moves and are less prone to whipsaw (though, under the right conditions they can be just as bad as ST lookbacks. For example, times when the market action is right around the 200 day MA or one year lookback)

Now take the inverse of the above and that is what shorter terms provide.

One can take two approaches in implementation. Averaging various look backs and weighting the look back values, which at the end of the day puts one in the mid range of whatever you averaged with some smoothing and/or tilts the single computation to take on the characteristics of LT or ST lookbacks.

My personal approach is to take two distinct look backs (one ST and one LT) and trade 50% of the portfolio based on them separately. This way I know I'm going to get one that provides good diversification from the other with different characteristics. In a nice bull run, the LT is going to crush the ST. After the market has tubed, the ST is going to crush the LT. Averaging/weighting/segmenting are all valid approaches...and exactly none of us know which is going to to best going forward.

Now with regard to December...no momentum system is going to prevent that unless it is very short term in nature. Just plan on eating hard short term spikes down (or up). The hope with a LT system is that things like that are in the noise and you stay long through the recovery. Assuming the recovery continues the ST system if it exited will get back in much faster and enjoy some of the snap back. Point being, there is no perfection, only trade offs.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by ochotona » Wed Feb 06, 2019 1:18 pm

Oh I see. Multiple look backs traded separately so you get STEPS in allocation, not just binary. Got it. But it must be closely related mathematically to multiple look back scores, scores compiled, then one trade decision. Thanks for the thought.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: Tactical Asset Allocation + HBPP an intriguing combo

Post by Kbg » Wed Feb 06, 2019 2:20 pm

ochotona wrote:
Wed Feb 06, 2019 1:18 pm
Oh I see. Multiple look backs traded separately so you get STEPS in allocation, not just binary. Got it. But it must be closely related mathematically to multiple look back scores, scores compiled, then one trade decision. Thanks for the thought.
Maybe it yields the same end portfolio results, I doubt it, but I've never done that study either. However, at the signal level no way. A 3 month signal and a 12 month signal traded separately do not yield the same trades as a single month stand alone signal. The market path from 3 to 7.5mo and 7.5-12mo could be completely different.

If you want a quick demo...go to PV, run VFINX at 2, 7, 12 mo periods and look at the results. Also look at annual returns for the years 1987, 2003 and 2009. Then pick some years where there were nice bull runs. You will see the performance profile differences.

If you want to do that study of if it all averages out, feel free to. It would be interesting.
Post Reply