Shorting the VIX

svxy-trade

Today I completed a trade using the ProShares Short VIX Short-Term Futures ETF, (ticker: SVXY). I entered at the close on 5/6/15 at $77.73 per share, And exited at the close today (05/14/15) at $83.64. That’s a profit of 7.6% before commissions, for six days of trading. Not bad!

It’s going to hurt my head to try and explain what the heck this thing is. But I’ll try anyway, and wince a little as I do. Someone correct me if I’ve screwed up.

“VIX” is the CBOE’s index of the implied short-term volatility of S&P500 futures. It’s an index that doesn’t track prices of stocks, but instead tracks the expectations of how crazy things are going to get. If the market thinks it’s in for a wild ride, volatility increases, and the VIX index goes up. Usually increased volatility is due to downward expectations in price, so the index is often called the ‘fear index’. You can read more about it here on the CBOE’s site.

Someone, or several someones, had the bright idea to create an ETF based around the VIX, so we can trade on volatility instead of price. The problem with VIX-based ETFs are two-fold though:

• They have a downward bias. I’m sure this has to do with futures expiration rollovers or contango or some other dance step…not my area of expertise. But over the long haul, they tend to drift downward. This makes any long plays harder because of the bias in the wrong direction.

• Mean-reversion systems often rely on a downward spike or significant event to act as a signal, and the return is made on the drift back to the mean. But the VIX tends to spike upward. Some bad news event triggers an increase of volatility, the VIX spikes upward, and then slowly drifts downward again. Volatility just doesn’t seem to spike downward in the same fashion.

Some other even more brilliant people decided it might be fun to have a short VIX-based ETF. This flips those two problems on their heads, turning them into features. A short-VIX ETF has both an upward bias, and spikes downward before reverting in an upward direction (usually).

I can trade that!

Below you can see two charts for the most recent time period. Note how the short-terms ups and downs are pretty similar between SVXY and SPY, but the longer term moves are different. For example the SVXY has been climbing pretty steadily over the past few months, while the SPY (S&P500 ETF) has been trading in a range. That means more opportunities for trades.

svxy-vs-spy

Unlike the VIX, which is direction-neutral in the long term, these ETFs have direction. So you really need to test with the ETF rather than the underlying VIX index. But the SVXY has only been trading since October 2011, which doesn’t provide quite the data I’d like for backtesting and developing systems.

Another thing to keep in mind: SVXY has a much greater daily percentage change than SPY. So adjust your positions accordingly, otherwise you might lose the farm. Today for example, SPY gained 1.04% from the previous day, while SVXY gained 2.04%. That’s almost twice the increase. On October 13th, 2014, SPY dropped -1.64%, while SVXY dropped -9.78%! Plan accordingly.

That said, I did come up with a nice little system that tests well. It tested so well in fact, that only one trade out of about 30 was a loser. Do I trust it? No, not completely. I don’t know how it’s going to behave during major market turmoil. But it looked sufficiently good that I was willing to put a little money on it. And the first trade was a great success.

I’m not prepared to discuss the system at the moment, but that’s not the point of this post. I love the proliferation of ETFs that has occurred over the past few years. There are many ways to make long-only trades that used to be impossible. Now if someone would come up with a triple-leveraged short corn futures ETF!

 

One thought on “Shorting the VIX”

  1. We all see how the etf Svxy
    Has a tendency to shoot to the moon..Is there an inherent upward bias just as the uvxy has a downward bias?

Leave a Reply

Your email address will not be published. Required fields are marked *