GTAA - The path to the optimal model (Part 2)

1Yr

So first of all, thank you for this post. I've been waiting for this for the past few days. I definitely have to thank you again, we've written so much in the past few days. Every time I've written a new comment, even the app has crashed every time. I do not have so much idea in general, but you could give me a lot of input, so thank you again. In this sense, I also believe that I do not want to take over the complete system. I have decided for me first of all the focus on GTAA with asset differentiation. therein I have already looked at a few me.

•

1Yr

@willemkly Gladly. I think it's good how you're slowly working your way up to GTAA. The fact that you first choose the simpler variant should be a wise decision. It is a combination of the classic world portfolio with the security of the GD200 strategy. Whether you do Top1 or simply apply the strategy to the positions, you can consider. It is probably also a tax question and one of the portfolio size. The smaller, the more likely Top1, but consider the psychological factor in the implementation. You have to be able to cope with a drawdown of almost 30% or a return of 0 for 5 years.

•

1Yr

@Epi @Epi as an example now. Vanguard All World (does not belong now to a specific etf the I observe but times as a theoretical example) SMA200: 95.12€ Current price: 95.33€ so seen broken through at the end of the month Momentum: 1 month: 1.70% 3 months: -0.08% 6 months: 3.59% YTD: 4.07% So how would you assess this situation now?

••

1Yr

@willemkly added up, the momentum would be 9.28% in total and my call money gives me 4% so it would be to invest according to the approach or have I done something wrong?

••

1Yr

@willemkly What are your rules?

••

1Yr

@Epi well I have first so take over if just the 200 day line is broken at the end of the month and just the total momentum is above that of my call money account or am I forgetting something?

••

1Yr

@Epi ?

••

1Yr

@willemkly Depends on which model you have taken. For GD200 model, GD200 is enough. For Dual Momentum you need the total momentum, but then it is not enough as a reason to buy.

••

1Yr

@Epi i mean the overall momentum

••

Breadley 🥪@Stullen-Portfolio

1Yr

@Epi but the momentum would be above that of the cash, as it says with you there

••

1Yr

@willemkly
1/3/6 months + YTD seems to me personally to make little sense...rather 1/3/6 + 12 months

••

View all 23 further answers

Eric@Weisswurstsepp

10Mon

Question for my understanding: What does rule 2 mean in step 2? How can you be above the momentum of cash?

And can you generally determine the order of the rules yourself or does it follow a fixed order?

Generally on the subject, as this has really got me hooked 😌:
If I'm interpreting this correctly, can I choose any ETF I want for each asset when applying this strategy? Or are there criteria other than that it should be cheap and not correlated? Especially with bonds, there are many different maturities. Whereby I see this (hopefully correctly) as further "diversification" after regional diversification...

And what should the ideal asset weighting be? I would have the following in mind for "practicing":

Equities (All World) 45%
Bonds (Aggregate Bond) 30%
Commodities (Gold) 10%
Real estate (Reit) 10%
Crypto (Bitcoin) 5%

Cash (interest-bearing settlement account/money market ETF)

Questions about questions, but those who don't ask remain stupid 🤷🏼‍♂️😅

•

10Mon

@Weisswurstsepp If you don't ask, you stay stupid, how true! So:

1. an asset is above the momentum of cash if it has risen more in the last 1,3,6,12 months than cash in 1,3,6,12 months. So e.g. in Germany currently money market Xeon yields 3.7%pa in the last 12 months. If a bond ETF has only yielded 3.5%, it is not bought, but held in cash. Cash is basically just a bond with a term of 1 day.

2. you can set the rules in the order you want. But find out in each case what the content does and what brings more return. Most of the time it doesn't matter.

3. you are completely free to choose which ETFs you take. Because Portfoliovisualizer only allows backtests with US stocks, I use this as a guide and try to find ETFs that can be mapped and tested there. The aim is to achieve a return comparable to that there. Bonds, which maturity, which region, hedged etc? The game is interesting but quite time-consuming. Putting together random ETFs doesn't work. I had to experience that with my own wallet. My first attempt at GTAA pretty much backfired.

4. the system works best if you use completely uncorrelated asset classes. Gold and bonds improve almost every model.

5. the question of weighting depends on the model. If you use dual momentum, you don't have any, because you always have the TopX in your portfolio. The weighting is then determined over time at most, e.g. 10% of the time commodities, 50% equities. The somewhat gentler model to start with is a GAA with a 200d strategy. This means that you only hold the assets that are currently above this weighting. But that's just simple momentum. A perfectly legitimate strategy, but not the one I follow.

6. your allocation is therefore only suitable for the 200d strategy. The optimal allocation is not a question of taste, but of efficiency. Which one gives the highest return with the lowest risk in the last 10/20 years. You can and must try this out for yourself in the Portfolio Visualizer. Feel free to post the link here and I'll take a look.

7. for the start: Allworld is unfavorable because it aggregates too much. Splitting into less correlated factors or regions or sectors is more efficient. Aggregate bonds are unfavorable because the corporate bond component is too strongly correlated with equities. Concentrating on long-dated government bonds is more effective here, as they are classically good performers in bear markets. I would distinguish commodities (oil, etc.) from gold, both of which have a fairly low correlation. Reits often worsen the overall performance, you have to test them. I do not. Bitcoin can certainly be weighted higher, as you already have protection with the 200d line.

8. it is best to familiarize yourself with Portfolio visualizer and simply test different ideas. Read up on asset correlations, search the web for example strategies and try to understand them.

9 I have three models myself, a conservative one with a pension insurance, a dynamic one incl. BTC with TR (both currently approx. 40k each) and a 3x leveraged one (currently approx. 8k). The first two are doing what they are supposed to do and are yielding constant returns with little fluctuation (10-20%pa). The third one is going like crazy and is fun (+5% per month since the beginning). But it took a lot of work to get there. You have to do it first.

Good luck! 👍

•

Eric@Weisswurstsepp

10Mon

@Epi Thank you very much for your detailed answer and tips. You've actually answered more questions than I had 😌!

Regarding 5.: The thought that I didn't need the weighting for dual momentum occurred to me just as I pressed "Post". I just put it down to the late hour 😉

•

1Yr

Notes/Questions: If you stretch the period to 1995-2022, it looks quite different compared to the performance of the S&P500, since your comparison in the picture starts in a downturn phase of stocks. Why did you sell your BTC? What does the spread look like with the EM small caps? Is it worth it to take the higher spread e.g. compared to EM IMI? You have to keep in mind that you have higher spread and trading costs the more GTAA goes back and forth between assets. This is not included in the backtests. Say: as simple / few assets as possible. Additionally, the question of a buffer position arises for me here, in case an asset just held slips just one position below the holding limit. Much of what I just mentioned certainly doesn't apply to your bond shell, does cost something p.a. and you are somewhat limited, hence these questions especially regarding your offensive variant.

••

1Yr

@theflyingsquirrel Yes, of course you're right that the comparison looks different from 1995 onward. But not completely different. I chose 2001 for two reasons. 1. I see the current situation quite similar to 2001: Long bull market ahead, US large tech massively valued, interest rate hikes. So I want to know how the model copes with this phase. The comparison with a boom phase is analogous to 2015-2021, where you can see that GTAA is lagging a bit, but not much. What I find decisive about the GTAA is the outperformance in difficult stock market years. That was also the reason why I started to deal with GTAA. 2. very pragmatic: the tool gives me Nasdaq100 only until 2001.

••

1Yr

@theflyingsquirrel Thanks for pointing out the spread of the EM SC! Besides the tradability and TER, that is of course an important point that I would have to include more. In mine it is about 70 basis points, ie 0.7%. Relevant in the assessment would probably be the average holding period and performance. If one loses 0.7% when buying EM SC and the average gain is 1%, then that does nothing. But if the profit on EM SC is 10% with 5 months holding period and on EM IMI 5%, then it could be worthwhile. I'll have to do some research.

••

1Yr

@theflyingsquirrel The trading frequency is of course a point. But it does not depend 1:1 on the amount of assets. Rather it is the SMA and the selected momentum periods. With about 4 signals per year, this is also within limits, I think. Of course you can reduce the amount of signals, e.g. every 2 months one, but this is at the expense of the performance. If you drive GTAA with more assets, you also have a higher probability to catch one that outperforms all others. If you have only a few (3), then the signals could constantly switch back and forth between two mediocre ones. The idea with the buffer position is of course not new. I had also already considered, especially if the asset universe is somewhat larger. Unfortunately, I haven't been able to backtest it and I'm afraid that it doesn't really improve the model. But in principle, I agree with you. If you have a choice between two similarly good models, then you should take the one with fewer signals. Regarding BTC: I did not completely liquidate the position, but reduced it to about 20%. I found that I feel more comfortable with this size at this vola. Ie as long as the buy signal is active, I stock BTC every month.

••

1Yr

@Epi Interesting with the comparison of the current situation to 2000. What would be important to me to say: The more assets you have in the momentum selection, the higher the probability that two assets are somewhat more correlated, these then run relatively the same and then the signals if it goes very badly between them every month swings back and forth. I would just be careful with the selection: For me important parameters in the selection: Top1-3 select depending on portfolio volume; assets with little correlation, there EMU SC is well positioned in my opinion and you have the SC factor in it; select assets with clear trends; select currencies for the individual assets, mostly in €; as simple as possible and as complex as necessary because of spread costs etc.; always consider the spreads on the assets and weigh, I find Xetra Gold e.g. also better than Euwax 2; I'm more pro buffer position, because it makes the whole simpler and brings some inertia into the system, thereby suffers of course the momentum but could also reduce false signals in return, the lower costs by the less switching can be offset against a lower performance, Röhl has in his semi-well-researched video also mentioned that the buffer position hardly costs performance and it is therefore worthwhile, where he should have the knowledge no idea, but I just wanted to let you know.

••

1Yr

@Epi I'm curious what you find out in relation to the EM SC, I'm also looking at the spread of the new EM SC ESG ETF, maybe the spread is lower. By the way, I would be happy if you could explain your 2 strategies completely, because I notice that I tend in the same direction, then we can debate about it and both adjust our individual screws.

••

1Yr

@Epi not easy the whole topic, but it comes more and more light into the dark. Regarding BTC: just wondered because in your gq profile BTC is not displayed.

•

1Yr

@theflyingsquirrel BTC is not displayed? I have two accounts running there, TR and mylife. TR should display BTC. For me it does

••

1Yr

@theflyingsquirrel I agree, the more assets, the higher the correlation probability. That's why more is not necessarily better. When I test the assets, I also expand and concentrate the assets several times. With the TR-GTAA I am now at 6 assets plus BTC. 3x stocks, 2x bonds, 1x commodities. All low correlated.I have also thought about how to avoid the oscillation every month. Since this mostly happens in trendless phases, one could possibly add a trend indicator. On the other hand, you can also just sit out these phases and trade stubbornly. They rarely last longer than 3-6 months. The buffer position is possibly worthwhile from Top3 and with a large asset basket. With the mini-variant US Tech, EU Small, US Treasuries Top 1 this does not work well.

••

1Yr

@theflyingsquirrel You can look at the spreads via extraetf. They are actually highest for the EM SC, around 0.6%. Perhaps one can say that then with all momentum figures this spread must be included? Anyway, I compared this once with the EM IMI and the EM SC is clearly better in the system even if you subtract 0.6% for each signal. If money flows into the EM SCs, then neatly, the trends last and are stable. In this respect, one can justify the costs. In the overall portfolio, I come to an average spread of about 0.15%. Makes with 5 signals per year max 0.75% performance loss. I find justifiable, if the model can bring 15%. Gladly we can discuss our strategies in detail, gladly also with @randomdude. But not here in the public space. Ideas?

••

View all 40 further answers

1Yr

@Epi actually with me no, it also says on your profile page that you are holding 2 positions right now.

•