Paul Allen

Long only, medium-term horizon, long-term horizon
Paul Allen
Long only, medium-term horizon, long-term horizon
Contributor since: 2012
Company: WallStreetCourier
It does not really matter which kind of moving average combinations you use. They all provide somehow a draw-down protection but you will never outperform a buy and hold strategy in the long run.
Just check out this study, where all MA combinations have been analyzed:
Golden Cross or other Simple Moving Average Crossover Strategies? A Critical Quantitative Review
http://bit.ly/1tnWxcx
Hi, now it is possible to create and back-test all modern portfolio construction techniques (Maximum Diversification, Risk Parity, Momentum etc) on your own online. Just check out: http://bit.ly/1KtimCO .. Have fun to implement your trading ideas...
Hi, now it is possible to create and back-test all modern portfolio construction techniques (Maximum Diversification, Risk Parity, Momentum etc) on your own online. Just have a look at: http://bit.ly/1KtimCO .. Have fun to implement your trading ideas...
Here is a great tool for every DIY Investor, as all modern portfolio construction techniques (risk parity, momentum, minimum variance ...) can be tested online for free...
http://bit.ly/1KtimCO
Well, the degree of diversification can be measured with the so called diversification ratio! By maximizing that ratio, the portfolio achieves maximum diversification with the given investment universe. So by applying that ratio to all common asset classes, the investor creates the most diversified portfolio.
Thanks for the interesting article. Basically, within all our published articles about multi-asset portfolios, we came to the same conclusion! This also holds for risk-parity portfolios.
Hi, thanks for the interesting article. I totally agree with our outcome. Furthermore, the argument that a large decline in a bond market would automatically lead to significant losses in a risk parity portfolio is a myth. Well, risk parity is not risk parity. Nowadays risk parity means equal risk contribution. That means each asset class should contribute the same amount of risk/volatility to the total portfolio. The risk parity approach you described above is called inverse volatility weighting scheme. As the overall portfolio volatility is not an additive function of volatilities, this concept ignores the correlation between asset classes. The weightings between those two concepts would be only the same if you use two asset classes, as in such a case you could ignore any diversification effect. I have used a Monte Carlo simulation, to generate 300 years of daily data to compare 10 popular portfolio construction techniques (http://seekingalpha.co...) and the outcome is quite surprising.
Well somehow it would make sense, to use a momentum scheme and weight the correspondening assets according to a risk parity/maximum diversification/minimum variance etc approach afterwards, instead of using a Sharpe Parity Portfolio.
FYI: The portfolio gained 8.8% in 2014 with a Sharpe Ratio above 1!
Hi MJS,
Well, as professional investor you get the data prior 1957. This is mainly due to the fact that Bloomberg Professional provides a generic index before that time. According to Bloomberg, the data prior 1957 is confirmed by S&P. However, might be possible that as non-professional investor do not get the data prior 1957.
Anyhow, the outcome of the research paper is pretty clear. Simple moving average crossovers will never outperform a simple buy and hold!
All data is coming from Bloomberg directly. They provide data for the S&P 500 prior 1957. According to their data-team, the pricing is provided by S&P directly and therefore they use a generic index prior 1957! But i guess the data is not available prior 1957 for non-professional investors.
Here is an comprehensive study about all possible combinations of Moving Averages, which more or less covers all questions. The outcome is not suprising at all but a quite interesting read.
http://bit.ly/1tnWxcx
Moving average crossovers are worth looking at but they should not be the sole source of information. I conducted a study of all possible crossovers versus a buy and hold. The outcome is not a big suprise at all.
http://bit.ly/1tnWxcx
FYI: The WSC All Weather Model Portfolio reached a new All-Time-High!
http://bit.ly/St2aVV
Well, if you consider the specific maximum losses of each underlying asset class, yes. Moreover, 2008 was a rare-event, where apart from long-term bonds and gold every major asset class suffered strong declines. As already mentioned, this portfolio was designed for an inflationary environment and 2008 inflation was of course no topic at all. Anyhow, we are also offering the "WSC All Weather Portfolio" which was designed to perform reasonable well during all predominant market conditions.
http://bit.ly/St2aVV if investors are searching for absolute return portfolio.
Thanks for your comments. The main purpose of the portfolio is to generate stable returns although its investment universe mainly consists of non/less-interest-rate sensitivity asset classes. Therefore this portfolio is a great solution for investors who are searching a protection in an rising interest rate/ inflationary environment.
Leif, you have made a good point. Assuming normal distributed returns have been always a criticism in finance, even with the assumptions of Markowitz. The main purpose of the article was to compare different portfolio construction techniques with the same underlying assumptions relative to each other, rather to focus on the absolute outcome of the individual financial ratios. You are definitely correct; there are enough parameters to work on when it comes to MC simulation. Whether to use kendall (which was used in the article), spearman or person, or how to estimate the dependences between the random ratios when it comes to copulas. So basically, the article was focused on different portfolio construction techniques, rather than modeling returns.
Fred, please have a look at our other articles, where we have used recent data for the risk parity and the maximum diversification approach.
Andrew,suviror ship bias is to some extend always an issue when it comes to single equity analysis. Anyhow, the main point is the relationship between risk and return as well as how different portfolio construction techniques will affect the total outcome.
Good point mike, regarding the scale. Anyhow, it is more about the direction of the regression line.
Nick, thanks for the post. Well I would say timing such environment might be a difficult task. However, there is a growing number of critism of the capitalization weighted indexing and in our opinion, there are more efficient ways how to construct portfolio. Anyhow, the inverse volatility technique from the SPLV is defenitely a better way and maybe some index providers will launch a similar ETF for the whole S&P 500, if the demand for such a solution is big enough.
Wuwei,
Well, we have solved the MDP with the constraints of no short-selling, since it does not make sense to short an asset class that has a positive risk premium. To find the right combination, you have to implement a non-linear optimization algo, where you can add the specific constraints. Furthermore, we have used a rolling variance-/covariance matrix, with an exponential weighting factor, in order to react on different correlation regimes. This is one of our major risk management tools within the WSC All Weather Portfolio. If correlations tend to rise, the weighting factor is putting more weight on the last couple of events, able to react on the regime switch. So therefore we cannot exaclty determine the lookback period, as it tends to change over time.
Good point. Well, if we would have used more asset classes, the outcome would have been basically the same in terms of the overall ranking for each portfolio. Of course, the performance and drawdown ratios would have been different, but the overall outcome would have been pretty much the same.
David,
I would be happy to read more about it :-).
Larry, you are correct. It would make more sense to substitute the equity part within a diversified portfolio with dividend growth stocks.
Thanks Charlie and Daniel for the post. As already mentioned in the article, the results should be seen as an additional source of information rather than an all explaining framework.
However, the main purpose of this article was an evaluation of different kind of portfolio construction techniques rather than forecasting the price action of any underlying asset class. Moreover, we wanted to highlight the impact of certain tail events on these portfolios, which we have not seen over the last couple of decades. In addition, the risk/volatility of asset classes tend to be pretty stable of time and therefore it was quite simple to make a return assumption, if we consider Sharpe Ratios of 0.2 to 0.3.
Transaction costs are having definitely an impact. The portfolio turnover ratio for the MDP is ranging from 6 to 13 percent a year and therefore transaction costs can be ignored.
Thanks for the comment! Our company is highly dedicated to market timing as well and therefore we also publish a Weekly Technical Market Comment on our website on a weekly basis. One of our main principals is to diversify among different kind of uncorrelated investment strategies (market timing and different kind of modern portfolio construction techniques) since it is the only free lunch an investor has. However, timing a market timing strategy is kind of tricky and if we would know what is happening tomorrow or what the market favors right now, diversification would make no sense at all.
I think the term risk parity weightings are quite misleading, since you are only calculating the inverse volatility weights and not the risk parity ones. Within a risk parity portfolio every underlying contributes the same amount risk to the total portfolio. Since the total risk of a portfolio is not an additive function of volatility, the inverse weighting method is not a risk parity one. Therefore the single weightings are not contributing the same amount of risk to the overall portfolio.
I think the term risk parity weighting is quite misleading, since you are only calculating the inverse volatility weights and not the risk parity ones. Within a risk parity portfolio every underlying contributes the same amount of risk to the overall portfolio volatility. Since the overall portfolio volatility is not an additive function of the underlying volatilities, the inverse volatility weightings are not contributing the same amount risk to the total portfolio.
Cullen,
Good points! I totally agree with your article, that an active manager who follows an index does not provide any advantages for investors (http://seekingalpha.co...) and that new portfolio construction techniques are worth the management fees (http://seekingalpha.co...).
Hi Bernie,
we provide the montly allocations of this portfolio in our members area of our website: http://bit.ly/TVnfWn
... I agree with the fact that there are a lot of trading opportunities around, especially when it comes to portfolio construction and different commoditiy/fx .. markets. However, right now the correlations within the S&P 500 as well as among the Russell 2000 are indicating that investors should not expecting too much outperformance from pure stock picking.
Good point, Back then, when we have designed the portfolio, we where thinking about that as well. Since the Vanguard Total World Stock Index ETF also includes Europe and Emerging Markets, we thought it might be a better approach to split those countries, as the correlation between those tend to change quite significantly over time.