Why Short Sector ETFs Aren't So Smart

So if I look at a broad index, such as the S&P 500, and then look at the returns of the two-times levered long and two-times levered short ETFs, the returns are more or less mirror images, with the two-times short fund only slightly underperforming. This is because the volatility of the S&P 500 on a daily basis is not extreme.

Another way of saying it is that these two-times long and short funds are small fish in a much bigger pond -- the water is so deep, these barely cause a ripple in the much larger market (not to mention that the intraday hedging can be done in a liquid futures market). The activity in these funds does not influence the broader market; the tail does not wag the dog.

But these smaller sub-index funds are much bigger fish in a much smaller pond. The tail does wag the dog, and there is not a deep futures market with which to hedge. And here is where you begin to see significant underperformance in these levered short-sector ETFs, likely because these funds are having an inordinate effect on their sectors -- and the volatility they help create leads to their own demise.

I took two recent trading days looking at the SKF (the ProShares Ultrashort Financial, SKF, the two-times levered short financial sector ETF), and just at a high level looked at the dollar volume in the ETF traded that day, and compared it with the dollar volumes traded in some of the underliers. Note, that isn't to say that every dollar traded in the ETF translated directly into dollars traded in the underliers, but the results were pretty staggering.

The SKF closed Wednesday, Nov 19, at $222 and change. Daily volume has averaged 31.5 million shares (volume was actually slightly lower than that on the 19th). Now, this is not scientific (or indeed even accurate), but it just gives you a sense. At $222 and average volume of 31.5 million, that means (if every share sold at the close, which it didn't, but again this is just to illustrate a point) that the day's dollar volume in this short ETF was close to $7 billion. Since this is double levered, that is really close to $14 billion in volume in the sector. I understand that each trade represents a buyer and a seller of the risk, but bear with me here.

The same day, Goldman Sachs(GS Quote) closed at $55, with roughly 30 million shares changing hands, representing 1.65 billion of dollar volume. Citigroup(C Quote) closed at $6.40 with a (then) whopping 340 million shares changing hands, representing 2.2 billion in dollar volume traded. JPMorgan Chase(JPM Quote) traded 90 million shares and closed at $28 and change, so roughly $2.5 billion to $2.6 billion in dollar volume. Merrill Lynch(MER Quote) had about $1 billion in dollar volume. The volume created by the SKF swamps all of these.

On Dec 4, assuming average price of $135, the SKF traded 29,248,827 shares, representing just shy of $3.95 billion in dollar volume traded. Since this is a double-levered product, that represents just under $7.9 billion of volume in the underliers. Goldman Sachs traded 23,838,644 shares at an average price of around $68, giving us roughly $1.6 billion in volume. Goldman accounts for 2.59% of the index associated with SKF. That means that basically, $204,610,000 of the $7.9 billion in SKF was associated with Goldman Sachs, or roughly one-eighth of the day's volume in Goldman.