根据经验，我（或者人们）通常认为，使用多策略（趋势跟踪为主+其他策略）会比纯趋势跟踪效果要好。有没有什么能够证明这个观点呢？似乎并没有特别有力的支持证据。

证据1：每种策略的相关性很少为1，两个相关性不为1的策略之间通过配置，可以提高收益风险比，增加分散的效果。（马科维茨的均值方差投资组合理论）

证据2：一个纯趋势跟踪的指数与一个CTA基金指数的差别，说明在过去，忽略存活偏差、样本偏差的基础上，CTA基金（多策略）略好于纯趋势跟踪策略。

然而，今天读到一篇论文，给出了一个结论，说是纯趋势策略的分散效果要比多策略要好。

最后的四段话非常有意思，值得细细品味：

Yet, CTAs are not all equal when it comes to providing these
diversification and risk mitigation benefits.

Multistrategy CTAs offer more of an absolute return type
of profile with undoubtedly a much smoother ride than
pure trend strategies. For that reason, they are easier to
defend but they do not provide similar diversification and
risk mitigation properties. The type of strategies that were
added alongside trend have effectively deteriorated their
diversification benefits. One of the explanations could be
that what has been added to improve the stand-alone
Sharpe ratios is convergent and experiences drawdowns
synchronously with equity markets.

Trendfollowing returns are overall positive but more
erratic and therefore, it is harder to hold on to the
strategy. Nevertheless, they are by construction positively
skewed and convex, which means that they can provide
strong positive returns when the traditional portfolio (be
it equity/bonds or hedge funds) is experiencing painful
drawdowns. They offer true diversification. In short, no
pain, no gain.

The choice between multistrategy and pure trend CTAs is
not an easy one. It depends on your investment objectives
and what you have in your portfolio. But for those who
have already engaged in adding alternative strategies to
their portfolio, the question is “are they really diversifying
or just diversified?”

虽然，我现在没有数据去否定这篇论文的观点，但是，总觉得这篇论文的论证过程怪怪的。这篇论文中，有些观点，值得学习：

1、volatility is not necessarily the most dangerous risk.What most investors actually want to avoid is a loss, and more so, a permanent loss.
2、Correlation is tricky and also unstable.
3、We can conclude that if one of your goals is to protect your portfolio against permanent losses, looking at correlation for diversification will not necessarily provide the full picture. In that context, it is important to evaluate the actual drawdown mitigation properties of the strategies under review.
4、A strategy is said to be convex when its return profile is non-linear, having both positive returns when the underlying asset exhibits extreme losses and
gains, whereas the skewness characterizes the standalone return distribution. In other words, convexity reveals the full potential of skewness, when properly ‘located’. An example of a strategy with positively-skewed returns and
without convexity is the “Short Bias”.