Working progress on model 3.

I expanded the model 3.3 up to 21 assets with results that continue to be solid without optimization. Beginning from Summer, I think that this model will be my new one with real money (now it's the classic model 3 that you can find in this blog).

It's also ready a variant of model 3 that I'll call model 4, with same rules but completely different time frames.
The correlation between the two models is positive and high (70%), but not too much.

I'll post them in May, when I'll have free time. I am also looking for other 1-2 Etfs candidate that are quite liquid (i just need tight bid/ask not volumes), enough track record for the benchmark and low correlation with most assets. I could also try to see how some short equity  Etfs work even if these models are not thought to go short.

Next week I'll post the new May allocation within the week end

MODEL 3.3 (Beta test) 3ETF 18 assets

I'm continuing developing a wider version of model 3.
At the moment asset correlations are increasing between stocks & bonds because of Central Banks liquidity. My target is to build a model with wider ETFs that, hopefully, will be able to resist and limit losses when the next crash will come within 1-2 years. On the other side I don't want to stay aside from markets hoping in better market metrics because often the better gains occur in the last part of the bubble (think about 1999).

This is a beta version and at the moment I'm not investing my money on this, because I always like to see it in real for at least 6 months after developed to see how it works and test if there's some formula bug in the excel file.

Basically it's the same "engine" of model 3, with these differences:
* It invests in 3 ETFs each month instead of 2
* ETF Universe is composed by 18 assets instead of 10.

All of them are listed in Italy but you can find them in other European Exchange such as Frankfurt. Just to clarify a doubt that one reader expressed to me: they are listed in Europe and are not hedged. That means that if I invest in US Equity, my performance is the sum of US Equity in local currency + $ performance vs €.

You can see the ETF universe (all Ishares) in the graph below and see how the model allocated money since 2001.

A problem I have at the moment is that many ETFs are recent and there's not enough track record to test them (or are illiquid).Therefore I find difficult to find other assets to add. For example I'd like to try some smart beta one, but is impossible at the moment.

Here is the performance of the model 3.3 on the hypothesis that I buy the benchmark index at the end of each month. Of course in practice the real performance will differ for slippage costs, bid/ask, capital gain taxes, difference between nav and price, commissions, but this is an interesting starting point in my view.

Allocations in 2015 were:
Jan Govt 15-30 / Treasury 7-10 / UK Gilts
Feb Govt 15-30 / Treasury 7-10 / UK Gilts
Mar S&P500€ / HY $ / Developed Markets Properties
Apr  Govt 15-30 / Gold € / UK Gilts

This is a work in progress because I'm going to add further assets.
The next step is to add many type of Govt (different maturities such as 1-3y, 5-7y, 7-10 and 15-30 and longer if available for euro and US) and corporate bond but I need to prepare the formulas to solve some problems with correlations. It will take time, but hopefully for the Summer I'll have a model that select among 30 assets.I'm also looking for other types of ETF to add other currencies exposure but it's not easy to find something in Italy outside of $, sterling and yen.

I remind that the target of model 3 is to take its risk, also exposing to losses around 7-8% in a month in exchange to obtain interesting risk-adjusted returns over the long term (an alternative to the pure equity investing for my portfolio). Sincerely I don't believe too much in the buy&hold unless you dont' buy in recession times...but it takes good guts!

April Allocation

March...what a great month!! It finished with a very good performance, even if I feared it because invested in overstretched assets. But this is the momentum power and the strength of trend follower methods. They are going to reward until.. the trend doesn't end. But who can say when it will finish?  We could be in a year like 2007 with strong fall coming in equity markets in the next quarters. On the other hand with all liquidity around, we could be in a market like 1998-99, where there were huge gains even if with bubbles all around. Personally I think that there's a bubble in Euro government bond, it's insane to have negative yields.  But I'll continue to surf market (and euro bonds), being conscious that at some point I'll have to take stop losses on the most recent trades.
Let's go and comment with the new allocations.

MODEL 1: it was allocated 100% on the US Equity. It closed with a gain higher than 3% thanks to currency effect. I remind you that all my strategies are elaborated by an European point of view with currency un-hedged. The new allocation for April is Euro Government 1-3years. Basically the model take profits and go conservative. If you can have a positive rate on a sight account this is the month for it. I have Conto Arancio at 1% ad I'll go on it.

MODEL 2: this model invested in US equity as well taking advantage of Dollar strength. The new allocation for April is Euro Government 15-30 years. This model is more volatile than Model 2 and finds extra performance in another stretched market.

MODEL 3: it invests in 2 ETFs each month. In March it was invested in Treasury 7-10 years and euro govies 15-30 that had both good performances in euro. In April the model confirms the euro government 15-30years, but switches from Treasuries to Gold. It's the first time since March 2014 that model invest in Gold and I must admit that it lost money in the previous two gold's allocations.

This is the table with monthly and ytd performance. It will be difficult that the 2015 will continue like in the first quarter.

Summary: all models are going to modify the asset allocation in April. 
Model 1 will invest in Euro-gov 1-3 years; 
Model 2 will invest in Euro-Govies 15-30 years; 
Model 3 will invest in Euro-Govies 15-30years and gold.

Model 1-2-3 - In practice with Markowitz

Since beginning of the year I described 3 models that I'm using at the moment.
Others are W.I.P (work in progress) therefore I don't want to mention them in this post.

Today I want to show how to combine my models to decide the % of portfolio for each one.

I'm going to use Markowitz optimization thanks to an excel written by Ravi Shukla.
http://myweb.whitman.syr.edu/rkshukla/Essentials/

Below I report the expected returns, standard deviation and correlations of 3 models
Constraints: no model can weigh more than 60%

 Then I ran an optimization looking basically for 3 theoretical portfolio:
- Constrained min. standard deviation
- Maximize expected return in a range 5-10%
- Max CAL slope assuming cash rate 0.3%.

You obtain 3 possible portfolios with different weights
For myself I prefer using the first on the left that gives
60% Model 1 - 16% Model 2 and 24% Model 3

I also keep some cash for discretionary strategies because I must optimize the fiscal side. In Italy Etfs are penalized because I can't compensate gains with losses, therefore I need other instruments to recover losses.

Finally someone could also use the portfolio that maximizes CAL (Capital Allocation Line) to move along the black line and rise expected returns or lower standard deviation reaching points not available with the efficient frontier. This is built with a +0.3% cash rate at 1 year.

March Allocation

Here we go with March allocation. 

MODEL 1: US Equity (from Euro Govt 1-3years)
MODEL 2: US Equity (from Treasury 7-10y)
MODEL 3: confirmed 50% Euro Govt 15-30 and 50% Treasury 7-10y.
MODEL BETA 3.2: 34% Euro Govt 15-30, 33% Treasury 7-10y and 33% US Equity (not real yet, just backtesting and studying)

Because these are un-hedged positions, the portfolio has an high $ exposure.
the portfolios shifts heavily into the stock markets, Model 1 allocated 100% on equity for the first time since 2005!

These are the results for February and Year to date.
Except Model 1, all other models had negative performance in February.
Because of very high January returns, portfolios are positive year to date.

Have a nice week

MODEL 3.2 Beta Version - More Etfs more assets

It's my pleasure to introduce an beta version of model 3 that I'm going to call MODEL 3.2
It's has a correlation about 80% with the "pure" model 3, because the logic of algorithms is the same
Therefore one can decide if use both or just one of them.
Basically there are 3 majors changes:
* 2 more assets (UK Gilts and Asia Pacific dividend shares)
* 1 more ETF each month -  model buys 3 ETFs rather than 2.
* Different time frame to calculate the market strength

Below the results (without including commission and slippage)

What I like is that:
1) again is not hyper-optimazed. I found another model much better with the same idea but I rejected because too optimized. I like that model 3.2 give similar results changing the variables and time frame over the long run, and this mean consistency for me;
2) even if the purchase of 3 ETFs a month means more commissions cost, you add diversification. It's not rare that in one month you see one ETF dropping a lot, compensated (all or almost) by the two.
3) the model invests on all 12 assets in the long term

Summary: this model is still in beta version, meaning I'm not investing real money with it. 
I want to investigate further in coming weeks and then beginning to use it
Must decide if/how combine it with the other three models to find the right weight. I think I'll publish a new post in coming weeks about how to use Markowitz to decide the model allocation.
I am also thinking about a version 3, with up to 5 ETFs each months and 20-30 asset to choose. But have to think how to filter correlations with a bigger number.
ALLOCATION: in January model selected  Treasury 7-10y, Euro govies 15-30 ad UK Gilts. January allocation was confirmed for February and this month lost some of January gain so far.

February Allocation

This 2015 began very well, with more aggressive models delivering solid gains
Basically one could also go flat for the rest of the year and be happy....
But it's not my style, therefore here we go.

MODEL 1 - Euro Government 1-3 years
The Most conservative model continue to be defensive and stay on the short end of the curve. If you are a retail, you can invest on 1month deposit. In Italy you can have 1% a year with one international Bank. At the moment with my saving I'm using it instead buying this Etf that yield basically zero.

MODEL 2 - Treasury 7-10 years
In December was invested in the long term euro curve with govies 15-30. In February the model switchs on the Treasury 7-10 years

MODEL 3 - Euro govies 15-30 and Treasury 7-10
It's the same allocation of January and had an high return.

Below the table with returns in 2015