Monday, July 21, 2008

Time for Reader Input: Does the VIX Have Market Timing Value?

I am going to try something different today: I’ll frame an issue in the context of an article from a highly regarded pundit, skip my own commentary, and ask for readers to weigh in with their thoughts. We’ll see how this works, if it works at all…

First, a quick summary of the pundit and the issue. Mark Hulbert, who I happen to think highly of, is out today with an interesting point of view in Putting on the VIX over at MarketWatch. The subtitle, “Commendary: Not necessarily bullish that VIX briefly rose above 30 last week” summarizes Hulbert’s position nicely, but I want to highlight several quotes from the article.

First, let me start with Hulbert’s conclusion:

“Even in those situations in which the VIX does appear to have statistically significant ability to forecast market movements, it turns out that those abilities largely derive from no special insights on the part of the VIX itself, but instead because of the stock market's tendency to rebound after steep corrections.

I owe this latter insight to Samuel Eisenstadt, research director at Value Line, Inc. Eisenstadt several years ago designed an econometric test to see if high VIX readings contain any unusual market-timing information after market declines, above and beyond the already-apparent fact that the market has just fallen. He came away empty-handed. High VIX readings tell us nothing more than we would already know from casually reading the news headlines.

The bottom line? The stock market may indeed continue to rally over the next several weeks. But if it does so, it won't be because the VIX momentarily rose above 30 last week.”

Early on in the article, Hulbert makes the case for the attractiveness of the VIX as follows:

“…consider first how the stock market performed over the subsequent month following a VIX close above 30: On average over the past 18 years, it gained 3.8% (as measured by the Dow Jones Wilshire 5000 index). In contrast, the stock market gained just 0.7% over the subsequent month following VIX closes below 30… Furthermore, similar contrasts existed at the quarter, six-month and 12-month horizons.”

Hulbert then goes on to refute the data for VIX closes above 30:

“…there's no way of knowing how high the VIX will rise when it first crosses the 30 threshold. Sometimes, like last week, the stock market immediately rises when that ceiling is broken, leading in most instances to the VIX falling back below 30. On other occasions, however, the stock market continues declining despite the VIX rising above 30, causing the VIX in most cases to continue rising. The all-time high for the VIX is 45.74.

So even though, other things being equal, higher VIX readings are more bullish than lower ones, there's nothing particularly magical about the 30 level.

To correct for this sleight of hand, I focused on just those occasions in which the VIX initially broke the 30 ceiling, after having been below 30 for at least three months previously. The data now painted an entirely different picture: Following those occasions, the stock market on average performed no better than it did the rest of the time.”

Readers, what do you think? Do you find Hulbert’s arguments convincing? Why or why not? What other important perspective and/or data do you think Hulbert is overlooking here? I would be interested in hearing some reader opinions in the comments below.

22 comments:

Peter Birchler said...

Hi Bill

I did not read the whole article on MarketWatch but from the excerpts it seems that some areas were overlooked.

Focusing on any "magic number" like 30 seems irrelevant. I think any discussion of the VIX has to be put in context of statistical analysis or de-trend the data in some way.

Anonymous said...

VIX is just one brick in the wall. VIX over 30 in the context of an extended selling stampede and blatant govt interference in the market makes a compelling case for a bounce.

VIX spikes are useful precisely because they are spikes on the charts. VIX is not the final answer but it's easy to see that something is afoot that requires further investigation.

Anonymous said...

Indicators which worked well in the past, may not work as well in the future.

And indicators which seem to work well now, may turn out not to be reliable indicators after all, their "powers" more attributable to circumstance than anything else.

There has always been a steady stream of enthousiasm for the "indicator du jour", only to wilt when it proved not as reliable as expected.

VIX looks like a good candidate...

Jens said...

I agree with the above. There is no magic number such as 30, but VIX can give good insight in situations such as capitulation sell offs. Each situation is different and there is no magic number. I read some study a while back that VIX would have spiked to 120 in 1987 crash. it is also insightful to combine VIX futures (that are actually traded) together with the VIX spot. Many times you can see the VIX futures are trading at discount or premium to VIX and this will give you another clue to the current situation.

Mark Wolfinger said...

VIX was well over 150 in October 1987.

When he drills down to VIX being >30 after 3+ months of <30 and finds nothing predictive about the market that follows, I can readily accept that.

To me that's the over-riding factor here and I will maintain my (almost) market-neutral trading (currently with a bearish bias).

Jim said...

Emotions, both fear and greed, move the market over the short term. VIX measures fear in the market, therefore vix should be able to give some indication of investor sentiment and trend although it will not pick exact market tops and bottoms.

Anonymous said...

VIX appears, with the benefit of hindsight, as more of a measure of event risk--the kind of shocks that seemingly come from no where and shake everyone up. In other words, a measure of how badly markets got caught looking the other way and not properly hedging. August most likely shocked risk mangers to where they got very serious about buying longer term put protection. January was likely the final straw that brought Johnny-come-lately risk managers to the table. Could be helpful to track the change in deep OTM/long dated put interest v/s pre-Feb 2007 levels. If long dated OTM protection has indeed been accumulated then we could have an orderly slide of 5-10% without the customary 30+VIX spike to mark a capitulation bottom. Naturally, inverse funds muddy things a bit but there should be a discernable trend in put activity for anyone with the ability to investigate.

-dowoper8tr--

DollarBill said...

To me, a VIX of 30 or any specific number, is meaningless. I realize that I'm dealing with noise, and I set up my positions with that in mind.

Unexpected events always trump the consensus.

kaiser soze said...

Hulbert is wrong.

When options trading is as prevalent as it is today, vix spikes to 150 (like in 1987) are practically impossible. And VX spikes above 30 are meaningful indicators and here's why :

Anytime VIX gets above 30, ATM and OTM options get sufficiently rich that market makers have an over-riding incentive to sell volatility and make a quick buck. Of course, they would need a liquidity cushion to do this but the Fed has already assured that to the primary dealers with TAF, discount window, etc.

Longer term buyers also get interested because in addition to the fairly low prices they are obtaining when VIX spikes above 30, they also have the opportunity to sell premium rich covered call and further lower their downside risk.

Anonymous said...

I know that all indicators perform flawlessly, for a particular period of time, and then defy their god, and fall flat on their face. The VIX is no different. The Achilles heel of all technical indicators is this, if you follow them, are you are about to enter heaven or enter hell.

Eric said...

I think you have to include some other indicators. I think your VIX:VXV and VIX:TNX are excellent concepts. Alone, these may just get traders into trouble, but with other indicators as well as good company research, I think this tool is excellent at "confirming" some form of turnaround. Take the recent financial turnaround. Using your VIX:TNX, I was able to wait for my moment to dive into financials. But I did not just dive into any old companies, I did my research and found several good (not facing huge writedowns) companies in the financial sector that were too oversold. Goldman Sachs (GS) is one. I played this via options and made a modest killing, and believe GS still has room to run.
Another example is using Money Flow. I think by combining Money Flow and your VIX:TNX, you can pinpoint extremes. I love trading on extremes. I just wish I had more data on the Oil Volatility, based on Money Flow, I think oil is going to break even deeper.

Bill Luby said...

Great comments, all. Thank you. I really don't have much to add to what has already been said.

Eric, regarding OVX, I assume you know that the CBOE has calculated historical values going back to May 2007 that are available at CBOE: OVX historical values from 2007

Cheers,

-Bill

Brian said...

The lows of Nov '07, Jan '08, Mar '08, & last week all were accompanied by brief VIX breaks above 30, all of which signified the market's short-term low. We just saw its magic again last week, yet many market watchers are trying to play down the importance. Why?

The market doesn't offer many high probability signals that often, the VIX >30 is a HUGE one. Buying the S&P 500 when the VIX was over 30 on the aforementioned dates yielded 60-100+ points in the ensuing days/weeks. What's the purpose in just brushing this off as a "coincidence"?

As for the defense of "the VIX could just keep rising", that is true. But I'm not saying one should just blatantly buy when it prints >30, or try to impossibly top-tick it. A much better strategy is to buy-in when it breaks 30 to the *downside* (after piercing it to the upside). You don't get the low, but you get the trend/momentum on your side.

For the purpose of full disclosure: it's made me a gold mine lately and therefore I choose to defend it more than most. If anyone has a better tool that gave a buy signal last Tuesday, I'm all ears! I don't think there was a single better signal than the VIX.

StockRake said...

I think there are several points to where using the vix to market time is very useful, but its very seasonal. Moreover, its one tool and you need to know when to use it. Spikes are always relative to the recent trading range.

My personal stash of custom built indicators using VIX or VXN over the years produced amazing results in 2002, and terrible results in 2003. I didn't use them after 03 so I don't know how they performed since then.

I remember when volatility would rise and the market would rise with it. Different markets make the volatility indexes act different at different times.

brian, there were many other items suggesting a bounce was a high probability trade before the lulling move over VIX 30 happened.

Michael Stokes said...

If I could throw in my 2 cents. The vast majority of the VIX is predictable based on information we already know, but as we talked about way-back-when (if you recall), I spent a lot of time trying to decompose the VIX into it's constituent parts and I was left with the conclusion that some part of the VIX is trader-driven, and if we could isolate just that part, we might gleam good trading input. I just never quite got to the point of figuring out how to do that.

In any case, see link for a quick writeup on this concept of decomposing the VIX into what we already know: The VIX isn't Magical. Enjoy.

Michael S

Kurt Osis said...

The Vix is just a reflection of the volatility smile. It is not mean reverting. Just as the smile did not exist before 1987, and has not "reverted" back to 1986 levels since that time. This phenomenon is an indicative of nothing more than the volatility price assumption in trader's models, which is recursive, and could change permanently at any time. It so happens, that puts have a higher implied volatility than calls. So as underlying prices fall the increase in the value of calls is more than offset by the rise in the value of the puts. But when the underling prices rise the rise in value of calls does not offset the decrease in the value of puts. This appears to be "mean reversion"; it is not. With mean reversion the farther away you get from the mean the more likely you are to fall back. With the VIX the further you get from the "mean" the more likely the traders are to change their model. (This is what happened in 1987, when puts and calls used to have the same volatility. Arbitragers have been waiting 21 years for them to revert to the mean!) Currently the SEC "Ban" on naked shorting is one such exogenous factor that could permanently alter the put/call disparity.

In other words... pennies in front of a steamroller my comrades.


The Big Board
. . . It was about an Earthling man and woman who were kidnapped by extra-terrestrials. They were put on display in a zoo on a planet called Zircon-212.

These fictitious people in the zoo had a big board supposedly showing stock market quotations and commodity prices along one wall of their habitat, and a news ticker, and a telephone that was supposedly connected to a brokerage on Earth. The creatures on Zircon-212 told their captives that they had invested a million dollars for them back on Earth, and that it was up to the captives to manage it so that they would be fabulously wealthy when they were returned to Earth.

The telephone and the big board and the ticker were all fakes, of course. They were simply stimulants to make the Earthlings 7perform vividly for the crowds at the zoo--to make them jump up and down and cheer, or gloat, or sulk, or tear their hair, to be scared shitless or to feel as contented as babies in their mothers' arms.

The Earthlings did very well on paper. That was part of the rigging, of course. And religion got mixed up in it, too. The news ticker reminded them that the President of the United States had declared National Prayer Week, and that everybody should pray. The Earthlings had had a bad week on the market before that. They had lost a small fortune in olive oil futures. So they gave praying a whirl.

It worked. Olive oil went up.

Michael S said...

Have to take issue with comment above from Mr. Osis. Over broad time frames, yes, the VIX is not necessarily mean-reverting. However, over shorter time frames, the VIX is incredibly mean-reverting.

I used to have a report on this on our website (which was since taken down during a site redesign), but I'll repub that at some point on my blog.

Over shorter timeframes, the movement of the VIX is much more predictable than the underlying market.

Just my OP.

ms

Kurt Osis said...

I read that CSFB report; I believe that is to what you're referring. The problem with such analysis is that it’s impossible to distinguish between back fitting on to data and causal inference in such cases. If I take any set of data and alter the scale over overlapping time periods, I can find some scale at which the data will appear predictive. Essentially he's just maximizing his R-squared by throwing out all the samples he doesn't like.

Garbage in, garbage out.


The point isn't that there mightn’t be a put/call implied volatility ratio that is untenable, the point is that ratio can be rationalized by writing more calls, buying more puts, hedging in the underlying, or changing the implied volatility assumptions. Mean reversion implies that the term structure of implied validity is constant when clearly it is not.

With the result that, if you come across as Vix of 50 or 60 in a mean reverting market you would bet everything you can that index was going to fall. In our case we do not know the actual mean (if it exists). The observation of Vix 60 more likely and indicator that your hypothesis is wrong than it is an indicator that the market will rally. So again, the further Vix moves from its "mean" the more likely, that wasn't the mean after all.

http://finmath.stanford.edu/seminars/documents/Stanford.Smile.Derman.pdf

I don't like the way Derman approaches this (though he's smarter than, I, he's trying to patch up holes in Fisher Black's work that Fisher Black realized were there but he never thought the discrepancies would matter because no one would actually use his model to price an option), but at least you can see the complexity of the underlying structure.

Michael S said...

Not at all. Referring to a report that I use to carry at my site - I'm in the process of moving these now. Will link at some point in the future. More to follow.

All the best,
ms

steve said...

I agree with the several comments here that you can not used fixed number targets on vix when it is known to change ranges in a dramatic way over the years. This does not mean there is no way to use it, rather you need some method you find useful, to detect 'where in recent range' the vix is. I have done well using 200 day lookbacks, though like everything else they are not perfect.

I think a subject of far more importance is what Bill and other blogs/websites have brought up. That being that new ways of hedging may be skewing the vix. One example is if you are bearish, you could be a buyer of calls on a 2X short etf. Under its current formula, vix would not detect this as more fear because you bought a call.

Vix has been reformulated in the past and I hope will be again in the future.

The concept that all indicators fail eventually is , I feel, a half-truth. I think a more constructive observation would be that everything can change over time, but close study and following trend changes like calls on short funds can alert us to perhaps change our approach to how we use a given indicator. This has alreaday happened and been done in the past with both the put call ratio and the vix.

Also, I think if enough people email CBOE, we could instigate a reformulation sooner rather than later.

MarketSci.com said...

Kurt Osis (sweet handle): I realized that I never posted back to these comments with the promised report. This is a link to my take on demonstrating that the VIX is very predictable in the short-term:
http://marketsci.wordpress.com/2008/07/28/the-vix-is-very-predictable/

Look forward to your thoughts,
ms

James Slack-Smith said...

Hi All,

Earlier today I posted a reply to Michael's post on the predictability of the VIX. In the reply, I describe a couple of Dakota systems that I built to investigate the predictive power of the VIX.

The VIX is useful for predicting the short term daily trends of the SP contract. However, I would suggest that it is no more useful than using the SP contract history itself. Meaning the same 'model' works if you are using either the VIX or SP (reverse adjusted closing prices) to produce the market timing signal.

If anybody would like a copy of these systems or more information then feel free to email me.

BTW, Michael if you ever do write a book then consider my order in!

Regards,

James

jamess@adaptivetradingsystems.com

DISCLAIMER: "VIX®" is a trademark of Chicago Board Options Exchange, Incorporated. Chicago Board Options Exchange, Incorporated is not affiliated with this website or this website's owner's or operators. CBOE assumes no responsibility for the accuracy or completeness or any other aspect of any content posted on this website by its operator or any third party. All content on this site is provided for informational and entertainment purposes only and is not intended as advice to buy or sell any securities. Stocks are difficult to trade; options are even harder. When it comes to VIX derivatives, don't fall into the trap of thinking that just because you can ride a horse, you can ride an alligator. Please do your own homework and accept full responsibility for any investment decisions you make. No content on this site can be used for commercial purposes without the prior written permission of the author. Copyright © 2007-2013 Bill Luby. All rights reserved.
 
Web Analytics