Forums  > Basics  > are trend following CTAs long vol ?  
     
Page 1 of 1
Display using:  

ComteZero


Total Posts: 523
Joined: Jun 2004
 
Posted: 2016-10-13 17:58
seems that people are unclear with that...

poll !

/* Trust is good, no trust is better. */

eeng


Total Posts: 15
Joined: Dec 2014
 
Posted: 2016-10-19 06:21
So after a while of voting, should we debate now?

My (naive) take is that, considering a Gaussian random walk process, a CTA will prefer a large mu with small sigma so whipsawing does not take it out from positions. In other words, clean trends without much volatility. Also, from a portfolio point of view, different asset return will be less correlated when volatility is low, which will also smooth out overal returns.

Would anyone care to prove me wrong?

Martinghoul


Total Posts: 859
Joined: Oct 2008
 
Posted: 2016-10-19 12:00
Sorry to be a pedant...

Is this question referring to CTAs as businesses or purely the strategy? IMHO, given what I know about investor preferences, CTAs businesses are definitely long vol, in a grand sort of sense.

Insofar as I may be heard by anything, which may or may not care what I say, I ask, if it matters, that you be forgiven for anything you may have done or failed to do which requires forgiveness...

goldorak


Total Posts: 1000
Joined: Nov 2004
 
Posted: 2016-10-19 12:30
Tail protection for long investors: Trend convexity at work

> The performance of trend following strategies can be ascribed to the difference
between long-term and short-term realized variance.

Case closed.

If you are not living on the edge you are taking up too much space.

dgn2


Total Posts: 2054
Joined: May 2004
 
Posted: 2016-10-19 18:47
You can also derive similar conclusions and extend them to multiple instruments or to other, more complex trend-following strategies using Monte Carlo.

The simple exercise of running trend-following on geometric Brownian motion and seeing that you prefer smooth paths to volatile paths is easy and a great first step in understanding the strategy. How volatility scales (i.e., not with square root time) can change the exact results but not really the basic conclusion.

If you make the stochastic process more realistic you can results closer and closer to reality.

...WARNING: I am an optimal f'er

ComteZero


Total Posts: 523
Joined: Jun 2004
 
Posted: 2016-10-31 14:51
well, first 4 answers poll showed 100% "yes": CTAs are long vol.
now, after 41 votes, it's 51% yes, 49% no.

there is wide taken-for-granted belief among HF investors that CTAs are long vol.
what stroke me was a CTA hedge fund manager explaining that he wants to insert VIX futures in his model to short it since "CTAs are by construction long vol". I know his model, it's a basic trend following setup based on moving averages.

i feel @goldorak summarized: usually they are long long-term variance and short short-term variance.

a trend-following CTA will have a good performance if you have a trend without too much short term realized volatility, in this case, winning position will grow as the trend is confirmed, as in any gamma-long position.
but if we reach the same point with very high realized vol, the model will spend its time taking "saloon-doors" until AUM is fully spent.

so, CTAs are highly path-dependents, and in most case, we cannot consider them "long realized vol" as basis case.
some investors even seem to consider CTAs equal to a long straddle (as my interlocutor), in this case, they forget that a vanilla option is a contract without path-dependency. something CTAs are not.

if it's true that without high short-term variance (and this trigger level is related to the model, and how noise is filtered to extract trend), CTAs are long gamma (long term), but they are not a straddle, and never long _implied_ vol (so shorting short-term implied vol such as VIX is just insane)


p.s.
problem was my interlocutor has 15y experience in CTA industry.

/* Trust is good, no trust is better. */

FDAXHunter
Founding Member

Total Posts: 8353
Joined: Mar 2004
 
Posted: 2016-11-12 11:29
The question was: "Are trend-following CTAs long volatility?"

The question was not: "Are CTAs long volatility?" (this question is almost akin to asking "Are futures traders long volatility?")

The answer to the former is clearly yes. Regress the daily returns of any trend-following CTA against volatility (realized or implied) and you'll see that they are long volatility.

The Figs Protocol.

ComteZero


Total Posts: 523
Joined: Jun 2004
 
Posted: 2016-11-18 19:14
even a good regression doesn't mean that volatility is reason of positive performance on TF CTAs, you cannot conclude on causality. i feel "being long" something means that you have direct benefit from its increase in value (and vice versa), so there's a causality in it.

second, a futures on S&P (or DAX would you prefer...) will regress quite well with, let's say against VIX. but you can't say that "a short futures position is long volatility".

as Balthazar put in a short exchange, people tend to focus on volatility and forget autocorrelation measure. He underlined that a TF CTA will perform with positive autocorrelation and may have poor returns with bad autocorrelation no matter where vol is.

again, you cannot take for granted that TF CTAs are long volatility : volatility is not trend, and even with trend and high volatility, a high negative autocorrelation will result in negative returns ("saloon doors" scenario). All depends in their model and how they filter/handle short term vs medium/long term movements

/* Trust is good, no trust is better. */

FDAXHunter
Founding Member

Total Posts: 8353
Joined: Mar 2004
 
Posted: 2016-11-19 12:02
Yes, you are right. I did not mean to strongly imply causality. Bad verbiage on my part. I should have said: "trend-following returns tend to be correlated with changes in general volatility measures".

Having said that, the term is not entirely incorrect: The basic component employed in many trend-following strategies tends to look suspiciously like a synthetic long gamma strategy. And this is what drives the volatility dependence of the typical trend-following CTA.

Similarly, equity long-short books tend, at a basic level, to represent synthetic short gamma strategies. As a result, their volatility sensitivity tends to resemble that of a short volatility position.

The Figs Protocol.

rickyvic


Total Posts: 117
Joined: Jul 2013
 
Posted: 2016-11-22 22:22
Basically right now TF CTAs are some sort of alternative beta, the "pure" ones are less common, because they dont perform for some reason.

The typical TF uses fx carry and short vol passive plays (this not as often) to increase the sharpe that is normally pretty low, because of the drawdowns, since they are directional mainly. Again mainly because they tend towards diversifying as much as possible with short vol strategies.

Issue is that diversification with these helps increasing the sharpe yes, but the result is that the product might have some serious drops when you need them to perform, i.e. periods of stress so high vol.

Forget past data for a sec, old samples do not represent the current portfolio really and ask yourself the question, how much are TF diversifying with these pretty dangerous (so to speak) short vol strategies? You will always have some long vol bias but I expect that to be less strong over the next few years and especially less stable correlation with a vol index.

Last thing is diversification by asset class and how much this really works in a time of high and unstable correlations, one more reason why I think TF have troubles. I am mentioning this because this creates even more dangerous situations for carry or short vol strategies.

I hope this helps even though I admit it is pretty generic.

"amicus Plato sed magis amica Veritas"
Previous Thread :: Next Thread 
Page 1 of 1