CTA的非相关性在哪里?_RCM Alternatives_HIT39

(bxyzz) #1

本专栏获得RCM Alternatives授权发表原文及其翻译。本文不构成投资建议,不代表个人观点,仅用于交流学习使用,不得用于商业用途。更多免责声明参见原文。


Where’s the Non Correlation? - RCM Alternatives​www.rcmalternatives.com图标

Those in the alternative investment space have long trumpeted the power of alternative investments (and particularly managed futures) to perform in a stock market crisis (see our evidence of that here and here). There’s no doubting that managed futures have performed during past crisis periods, but as the disclaimer says – past performance is not necessarily indicative of future results – and we are looking at a lot of red for managed futures this month. You have bellwethers such as AQR down about -4% from their highs in early October, and the ScoGen CTA index having put in a new five year low point (yikes), down nearly -7% on the year.

另类投资领域的投资者长期以来一直在鼓吹另类投资(尤其是管理期货)在股市危机中发挥的作用(点击此处查看我们的证据)。毫无疑问,在过去的危机时期,管理期货表现不俗,但正如免责声明所说,过去的表现并不一定预示着未来的结果,且本月我们见证了管理期货的大幅亏损。投资风向标诸如AQR指数较10月初的高位下跌了约4%,ScoGen CTA指数创下了新的五年低点(作惊讶状),较去年同期下跌了近7%。


So where is the crisis period performance? What gives?


Well, as always – the devil is in the details. You see, when we talk about crisis period performance for managed futures, we’re talking about periods longer than a single day, week, or even month. We’re talking about periods in which traditional investments see major shifts (2007 to 2008 is the classic example, when stocks and commodities fell over several months). You see, crisis periods have two parts. One is the crisis itself, which usually causes a reversal of the current market trend (in this case current trend was stocks, foreign currencies, and energies up; bonds and US Dollar down). The second part is the aftermath of the crisis in which new market conditions and trends emerge.

老生常谈,细节决定成败。要知道,当我们谈论管理期货的危机期表现时,我们谈论的是比一天、一周甚至一个月更长的周期,是传统投资发生重大变化的时期(2007年至2008年就是一个经典的例子,股票和大宗商品在几个月内持续下跌)。 可以发现,危机时期由两部分组成。一是危机本身,通常导致当前市场趋势逆转(当前趋势是股票、外汇和能源上涨;债券和美元下跌)。第二部分是新的市场条件和趋势出现的危机的后果。

Managed futures generally outperform during the second part of the crisis, and under perform during the first part. We need look no further than 2008 for evidence of how managed futures and macro react to shifts in market, where commodities saw a sharp reversal off of all-time highs, causing pain for managed futures from July through September, before the aftermath portion of the crisis kicked in through October, November, and December. [Past performance is not necessarily indicative of future results]

管理型期货通常在危机的第二阶段表现出色,而在第一阶段表现欠佳。我们不需要在2008年之前进一步寻找有管理的期货和宏观经济对市场变化的反应的证据,在这一市场中,大宗商品价格出现了历史最高点的大幅逆转,在当年10-12月危机余波到来前,于7月到9月期间重挫管理期货。 [过去的表现不一定代表未来的结果]


管理期货指数 2008年7月至12月

The problem for alternatives back in February was that there was no second part of the crisis. There was a vol spike and -10% correction in stocks, but it dissipated just as quickly as it appeared, eventually giving way to the regular scheduled programming of the slow crawl upwards. And the problem for alternatives now, is that we’re only in the first part of the crisis, yet to know whether there will be a second part. The problem for alternative investments like managed futures in both cases is that the existing trend most were involved in was UP, with recent multi-year highs in stocks as well as lows in bonds. Here’s what the side by side performance for managed futures has looked like in the initial stages of each sell off:




Why did they essentially mirror stock market performance when they are a non-correlated investment? Isn’t this what we have them sitting around for – to perform during a down move in US stocks?


A few ideas there. One, there was a clear up trend in stocks (and down trend in bonds) coming into both February and this October correction. Trend following models are designed to capture such trends across asset classes (including, for sure, stock indices and bonds). And there was a clear reversal of that trend. But, as we can see with a look at a simple 50 day/200 day moving average cross over trend following setup on the S&P 500, the signal of a down trend in the S&P was well after the bulk of the move had happened. And this is a rather quick trend following signal, using just 20 days and 50 days, whereas some managers may be out much further at 100 and 200 days. The result – these programs were slow to reverse to the down side, given the quickness of the move in relation to how slowly prices had risen before hand.



s&p up/down trend标普上/下行趋势; SPY 标普500ETF 20d MA 20日移动平均线 50d MA 50日移动平均线 T trend reversal signal T型逆转趋势信号

What’s more, some of this slowness to react may be intentional. We mentioned in our 2018 Managed Futures Outlook to beware the managed futures programs who may have ‘cheated’ a bit over the past few years by adding more short vol and long equity exposure, in a sort of ‘if you can’t beat ‘em, join ‘em’ mentality; and some of that could be going on here. The last thing you wanted to do in the stock market sell offs the past few years was get out of your long position. The sell offs were short lived and quickly reversed, teaching any manager (or machine using AI) that it is better to space out exits and not do a knee jerk reaction to down moves. It’s more than possible that the resulting performance profile has become capturing more stock downside than managed futures and macro programs may have in the past, as well as delaying signaling a down trend.


Finally, the fact that alternatives are losing money at the same time as stocks is a prime example of the confusion between non-correlation and negative correlation. We won’t blame you if correlation statistics are a little hard to wrap your head around. It’s easy to understand positive correlation (performance moves together) and negative correlation (performance moves in opposite directions), but non-correlation is a bit harder to grasp. Just what does a -0.06 correlation look like, anyway. How do we conceptualize a lack of correlation one way or the other? Most of us incorrectly conflate negative correlation with non-correlation, but the reality is that non-correlations really just means sometimes positively correlated, sometimes negatively correlated. As we showed in an in depth post on correlations, here’s what that sometimes positive, sometimes negative correlation looks like over time.



图表5: 管理期货与股市的滚动相关性

So we were up at one of those positive correlation peaks before this market sell off. That’s bad news. The good news, if markets keep going lower, we won’t remain there for long, with models reversing and taking on short positions as correlations move into negative territory. For now, this sure seems like a bit more of a real down move than previous times during the now 10yr old bull market run.



past 30 days performance=过去30天的表现