A 2-3% Correction Could Wipe Out Most VIX Short Sellers

[sle]

Being short vol has been the best performer since '08 and continues to be even at these levels.

I have always maintained that being short various forms for risk premium is a perfect retail trade.
- Their horizon is in decades and not months, so they can survive a year-long drawdowns.
- They do not have a risk manager to tap them on the shoulder at the worst possible time.
- If the size is right, it might hurt sometimes but it will not destroy them.
However, investor memory is very short and temptation of leverage is very high. That’s what scares me.

 

[sle]

depends if you are selling ATM or near ATM or way OTM vol

Actually, selling OTM index puts has been a great trade, on average performing better than the index itself (risk adjusted).

My boss used to say that “selling risk premium is like sleeping with strangers”. I think it’s a good analogy - you want to do it when it’s attractive and you don’t want to do too much of it.

 

[rallymode]

depends if you are selling ATM or near ATM or way OTM vol

Perhaps I don't follow what you are saying but being short vol is being short vol(-$vega) at any given snapshot in time. Let's forget for a moment that the article is talking about selling vix futures so that maybe you can elaborate how strike dependency makes vol selling a lunacy compared to a long spy position.

 

[sle]

Perhaps I don't follow what you are saying but being short vol is being short vol(-$vega) at any given snapshot in time.

Selling weekly 25 delta calls at 4.5 vols qualifies as lunacy

 

[rallymode]

Selling weekly 25 delta calls at 4.5 vols qualifies as lunacy

I stand corrected

 

[jtrader33]

It’s trading the same number of contracts. The index is replicating a “30 day constant maturity futures contract”. To keep that maturity for N contracts, every day they have to sell N/30 of the fist and buy N/30 of the second.

Gotcha, thanks. I've seen the opposite argued a few places (here's one https://www.schaeffersresearch.com/content/news/2015/02/09/how-time-decay-affects-the-value-of-vxx) but don't trade it so never bothered to go through the prospectus. Figured you would know definitively since it's in your realm.

 

[sle]

Gotcha, thanks. I've seen the opposite argued a few places (here's one https://www.schaeffersresearch.com/content/news/2015/02/09/how-time-decay-affects-the-value-of-vxx)

I am surprised that someone who does research for a living would make such a blatant error.

 

[i960]

Gotcha, thanks. I've seen the opposite argued a few places (here's one https://www.schaeffersresearch.com/content/news/2015/02/09/how-time-decay-affects-the-value-of-vxx) but don't trade it so never bothered to go through the prospectus. Figured you would know definitively since it's in your realm.

This is because the typical "mainstream" assumption about time decay when it comes to ETFs is that it has to do with volatility drag with leveraged ETFs. "Time decay" can also apply to ETFs long an underlying in contango or short an underlying in backwardation. Absolutely nothing different about this from the roll yield being positive or negative with an actual futures position and a given curve shape. This isn't a revolutionary thing but outside of the futures context equity investors are continuously surprised by such things because they don't typically think about curves of anything.

 

[srinir]

It’s trading the same number of contracts. The index is replicating a “30 day constant maturity futures contract”. To keep that maturity for N contracts, every day they have to sell N/30 of the fist and buy N/30 of the second.

"30 day constant maturity futures contract" is correct, but number of N contracts may not be correct (or I might be interpreting wrong)

Here is the prospectus>
http://www.ipathetn.com/US/16/en/documentation.app?instrumentId=259118&documentId=6091544

on PS-24 page,

"equal notional amount" implies that it is dollar weighted. For eg: if first month contract VX1 value of 10, N contracts was sold for the amount of $1,000, then number of next month contract VX2 value of 11 bought would be be less than N.

 

[i960]

Yep agree. I don't think the contract count is constant. They're simply used as the standard unit of X% vs Y% of the total funds value.