Journalists Have a Responsibility Not To Get It Wrong

My gratitude to Don for finding a 'news' story from Bloomberg and beginning the discussion.  To me it's more like a short (fiction) story than a news article. 

It never ceases to amaze me how some financial reporters find people to interview.  So often they quote people who either don't understand what they are discussing, or don't know how to speak in terms the layman can understand.  Once the interviews are complete, the journalist(s) have the impossible task of writing a story  that makes sense and provides useful advice and/or information.  And they must do that based on what they have been told by the 'experts.'

A case in point: 
Jeff Kearns and Michael Tsang posted this piece on earlier today: "VIX Fails to Forecast S&P 500 Drop, Loses Followers"

Be certain to begin with Don's take on this topic.  He takes the time to kindly mention that the writer's conclusions are not completely erroneous.

Here's the first line in the Bloomberg story:

"Investors are starting to abandon
volatility as a forecasting tool for stocks after one of the
most-used measures of price swings failed to anticipate the
biggest monthly decline in U.S. equities in 21 years.

No evidence is offered showing that anyone is abandoning VIX (the CBOE volatility Index) as a tool, although one expert was quoted as saying that
“It used to be that an extreme one-week move in the VIX
either up or down would give you predictive power, and now it’s
just completely broken down.”

What does that mean?  In the past, when VIX was thought to have reached a very high level (as the result of a severe market decline), the 'predictive power' concluded that the market would next move higher.  That was often the case.  To suggest that it's 'just completely broken down' because VIX continued to rise (and the market continued to fall) is absurd.  And the guy who said this manages volatility strategies for pension funds.  Is it any wonder that pensions (and other funds managed by professionals) have not been properly hedged and have incurred large losses this year?

b) The VIX is not intended to 'forecast' anything specific.  As Don clearly describes, when the VIX was near 24 – the level mentioned by Kearns and Tsang,  based on a movement of one standard deviation, the market could be expected to trade with a range (7% over the next 30 days in this instance) – roughly 68% of the time.  Kearns and Tsang mistakenly believe that the number (7%) represents a 'prediction' that the markets will not move by more than that amount. 

I don't know if that's simply a poor conclusion from people who don't grasp statistics, or if it's the result of being given that information by one of the 'experts' interviewed for the article.  But, it's not a valid conclusion. 

What the 24 VIX did suggest was that approximately 32% of the time, the market would move more than 7% over the next 30 days.  More than that, it also suggested that there was a 5% chance the market would move more than twice that amount.  The VIX is not designed to 'predict' the exact movement of the stock market and it cannot be 'wrong.' 

VIX is a measure how options are priced in the marketplace, and there are many factors that go into determining an option's price.  We cannot take those option prices as gospel when estimating how the markets will move.

c) "The so-called VIX also lost 44 percent since
Nov. 20, a bearish signal, even as the S&P 500 rose 18 percent.

It's the VIX, the volatility index.  It's not the 'so-called VIX.'

When VIX decreases, such as 'lost 44%,' that is not a bearish signal.  VIX traditionally 'loses' i.e., decreases when the market rises.  Thus, the VIX  'did what it's supposed to do during rallies.  I have no idea why these authors believe VIX performed its predictive powers poorly.

d) The VIX…"
failed to signal the scope of declines in the
worst year for stocks since 1931"

It did no such thing.  VIX is supposed to move more than two standard deviations about 5% of the time.  These authors think that's never supposed to happen.  And statistics tell us the market will move more than three standard deviations about one time in 200.  That may not be a common event (more than once per year), but it's certainly not rare.

It's true that we did see moves much larger than anticipated, but the VIX steadily rose as the markets fell.  As VIX increased, the so-called predictive power of VIX predicted larger and larger market moves.  But these authors want to focus on that single reading of 24.  That is simply wrong.

There is more to this article and it's not necessary to nit pick every statement they make.  But it goes to my general premise that supports two very trite, but nevertheless true, ideas:

1. A little knowledge is a dangerous thing

2. The blind really do lead the blind.  Or in this case, two under-informed financial writers are trying to educate the public – and are offering only confusion.

To their credit, Kearns and Tsang do quote someone who gets it:

“Saying the VIX isn’t worth looking at because it failed
to predict October is like saying, ‘The weatherman said it was
going to snow and it didn’t, so I’m going to stop looking at the


2 Responses to Journalists Have a Responsibility Not To Get It Wrong

  1. JB 12/22/2008 at 6:49 PM #

    Hi Mark,
    Does VIX have the “fat tails”? If so, wouldn’t VIX move more than two standard deviations about 5% of the time?

  2. Mark Wolfinger 12/24/2008 at 11:50 AM #

    I’m not an expert on statistics and understand just enough to get into trouble.
    Perhaps a reader can help us out.
    SPX is an index of stock prices. Thus, its volatility can be measured. We have seen some very large moves – moves not anticipated by the normal distribution curve – over the past few months. Those are the fat tails.
    VIX is an index that is composed of data based on the implied volatility of a bunch of SPX options.
    VIX is substantially based on human emotions. The actions of panicky option buyers moves VIX substantially higher. I don’t have a clue as to whether it’s ok to use a statistic that’s based on that fear is subject to analysis by using a normal distribution (bell shaped) curve.
    All that means is I don’t know if we can apply statistics to VIX data. It seems we should because market movements are also based on panic and greed and I have no problem using statistics on market prices. But somehow, this ‘feels’ different.
    Despite being trained as a scientist, that’s my best unscientific reply.