Wednesday, March 22, 2017

Rethinking option correlation and diversification

Here's a portfolio I see in several option videos/discussions when it comes to selling premium:
Strangles/verticals in several underlyings, with the key risk management metric being your max risk in any position is 1-10% of the account. (smaller max risk for large accounts)

For example, in a $100k account, they might have SPY, QQQ, and DIA verticals with $1000 max draw down.  (there are many other positions in metals, bonds, etc, but those aren't the issue)

This sounds OK right? Our max draw down is $1k, and we are within our main risk management bounds.  But then at the same time, they will talk about underlying correlation!
With SPY, QQQ, and DIA having a .7 or higher correlation, (they move almost together), you are just making a giant Frankenstein hybrid SPY beta weighted position.

 Comparing the .7 movement correlation roughly to the drawdown, you are basically adding $700 to your max SPY drawdown from each correlated underlying, turning your risk profile from a 1% max drawdown in SPY to 2.8%. 

I see so many options posts just listing through their positions and adjustments, "here is my SPY strangle, my QQQ vertical, my DIA strangle, hum di dum..."
I just want to smack them out of it, "NO YOU GET TO PICK ONE!"  I'm visualizing it like a kid with 3 candy bars and their parent yelling "YOU ONLY GET ONE!"

Lets take a step back though- Obviously many people are successfully trading short premium with these over correlated positions, so I see a fork in the road:
Either
 1. Realize that you need to shrink position size because you are more correlated than you think
2. Care less about correlation/ diversification!

I'm leaning toward #2, not in the sense that I want equal SPY, QQQ, DIA positions, or have a bunch of metal positions, but just going back to the "PICK ONE" ! 
Additionally, if you are using the conservative 1% of account max drawdown per position, you are going to run into correlation issues to fill up that many positions!

If I was setting up a portfolio with US equities, metals, bonds, etc- Just pick one etf per area, based on volatility/ premium.
Go bigger on just SPY, GDXJ, TLT.  In my case I do advocate short vol as an SPY replacement (combined with a cash position), rather than smaller positions in SPY, QQQ, DIA, SLV, GLD, GDX, etc.
In practical terms for the small investor, we are just saving transaction costs, getting similar exposure for a third of the stock and option commissions. 
 

(Here are the 3 main indexes moving in lock-step during yesterday's 1% down move)

Furthermore, in a crash scenario, or at least 10%+ correction, correlations get even more heightened.  In short vix land specifically, that >10% correction is the main risk we are dealing with anyway, so a portfolio of short vix would perform similarly to a portfolio of "semi uncorrelated" indices and stocks. (Obviously short vix would spike down much harder, but the overall movement and recovery conceptually would be similar.)
In a large correction, the only real diversification is long vix, which is an overall portfolio drag.  So what real protection do you get from diversification?

Buffet: "Diversification is protection against ignorance. It makes little sense if you know what you are doing"
Cuban (idiot):  portfolio diversification “is for idiots.”

Here's a little real life diversification/correlation comparison to explore when correlations go to 1 and diversification fails:

Lets say I work at an IT desk and a user needs a special account setting changed to access some network files, etc.  Our boss plans there to be plenty of coverage because we have one person dedicated to account permissions, plus a few server admins who can access the function as backup.  We are diversified right? Lets say four people can do this function so someone will be there to help users in a timely manner.  Unfortunately in real life, they all go to lunch at the same time, or are all in the same admin meetings, so their correlation goes to one!  Instead of four people, they are one correlated lunch-eating entity!

  I'm only slightly ranting but it can be for a good purpose to show how really understanding your diversification and correlation risk can permeate every facet of our lives outside of investing.  I can guarantee none of these people in my lunch example have traded EVER because of that kind of correlated insanity would be squashed instantly.

Think about your intrinsic view on correlation, and if you are or actually can by defended from spiking correlation.  If there is no real defense from overnight spiking correlations in a correction, why not just be short vix in the first place? 

Yes I know this all seems massively risky but I think its an important discussion to add least address the shortfalls of the alternative. 
 

1 comment:

  1. Keep your eye on the shortvix guy and make some serious hard cash! Ignore this blog at your own peril.

    ReplyDelete