What's Working
Putting The CBOE Volatility Index, the VIX, To Work

You may have noticed that the CBOE Volatility Index, the VIX, is now part of Portfolio123 and accessible through the GetSeries function. Not only is this a useful index in and of itself, as some users are already seeing, but it can also be used to help you decide how to use other tools.

Introducing the VIX

The media has lately become aware of the VIX. That's not surprising. Quants have been strutting on Wall Street for a while now, so the word "volatility" has become hip. That may be a good thing. Volatility is an important concept and widespread awareness of it is probably long overdue.

The VIX has roots in a landmark option pricing model known as Black Scholes. The inputs are the price of the underlying stock, the option's exercise price, the risk-free interest rate, the time to expiration, and the stock's volatility (standard deviation expressed as a percent). The output is the fair value of the option contract. As a theoretical doctrine, this was massive, enough so to generate a Nobel Prize.

In practice, don't even think of committing your life savings based on the fair values you get from this model. We know some of the inputs with certainty; the stock price, the time to expiration and the exercise price. We can make pretty good assumptions about the risk free rate (different users will choose different instruments, but all assumptions are likely to cluster in a narrow range). The problem is volatility. Nobody can come up with reliable assumptions for how volatile a stock will be between now and the time when an option expires. Hence in the real world, the model falls prey to GIGO, "Garbage in Garbage Out."

But if you think it's hard to estimate volatility, imagine the difficulty in trashing a Nobel Prize winning model! As it turns out, the investment community found a great use for Black Scholes. Instead of having the fair value of an option be the output, we turn things around and use the observed price of the option (which we know simply by looking) as an input. Using some (a lot, actually) of algebra, we reshuffle the grand equation and move volatility, the only thing that is truly unknown, into the output position, and we give it a new name: implied volatility.

In other words, the resuscitated Black Scholes model is now telling us: "Based on the price at which the option is trading, and some other simple inputs, we can conclude that the market expects the stock to experience volatility of X% between now and the time when the option expires." If you think the stock will be more or less volatile than the market assumes, you would assume the option in mispriced and act accordingly.

The VIX, by amalgamating implied volatilities for S&P 500 stocks, provides a great window into investor emotion. Low VIX levels indicate that investors are confident in what they're doing; they expect low volatility. High levels of VIX suggest investors have less confidence in their decisions, and very high VIX levels suggest that most are completely clueless.

Interpreting VIX levels and trends

Truly understanding VIX and how to use it may involve a bit of culture shock. We're all well conditioned to looking at price charts, index charts, indicator charts, and so forth and assuming high or low means good or bad. Which is which depends on what we're looking at; in many contexts, high means good, but there are some where high is bad.

VIX is a very different animal. It measures exactly what it says it measures: volatility, more specifically implied volatility on options contracts for S&P 500 stocks.

We cannot associate always concepts of good or bad with high or low levels of the index. Volatility is a two-way street. Sometimes, it's good to have a lot of it (i.e. when we own stocks in a rally). Other times, it's bad.

That alone can be a challenge. But there's more.

As it turns out in the recent past, the highest volatility readings we've seen have been associated with downturns in the stock market. This is not inevitable. We can have huge volatility tallies accompanying bull markets. The fact that we haven't been seeing that lately does not result from the nature of volatility, but from investor psychology.

When we've been scared, we have tended to act (dump stocks) quickly and pretty much all at once. When it comes to getting back in, we have tended to be gradual, each of us edging our way back at different times and at different paces. That's why volatility has tended to be less extreme during the rallies that have been experienced since the CBOPE started computing this index.

I've often wondered, though, what the VIX might have looked like had it been computed in 1982. The August rally that year, the one that launched the nearly 20-year bull market, amounted to a "panic buying" that left many gurus of the day shocked and perplexed. I suspect we'd have seen very high VIX readings. Ditto with the buying onslaught that followed the 1987 crash. Both of those bullish periods were characterized by the repertoire of dissent, verbal argument, name calling, etc. that would be consistent with expectations of high volatility, some believing that stocks would continue to roar ahead and others expecting a new crash. I suspect we'd have seen the same at other major buying opportunities; 1973, 1932, etc.

But because the VIX calculations don't go back that far, we're presently stuck with a situation in which basic visual and/or statistical analysis of VIX data will point to high levels as being bearish. Case in point: see figure 1.

Figure 1

The Subtleties

Look again at Figure 1. In the early 2000s, high VIX readings could have caused you to do some well-timed selling. But had you waited until VIX returned to more comfortable levels, below 20, you'd have missed a lot of upward movement. You'd have had a great pre-2008 warning, but as to getting back in, you'd have missed every step of the approximately yearlong super rally that began at the end of 3/09.

When working with VIX, the numeric level is obviously important but needs to be supplemented with other factors. Trends in the VIX, whether its rising or falling, and how fast the trend is moving, need to be considered. Usually, declines in VIX indicate increasing confidence, and with investors being chronic worriers, that's not usually going to happen absent bullish sentiment. But even this is not a perfect rule. Investors in late 2008 lost a lot of money while VIX was falling. Note, though, that this decline was from a level that could be characterized as insanely frightening (around 80) to one that was merely dreadful (the 40s).

That's an important cue. A decline in VIX from 80 to 40, even if the same in percentage terms as a decline from 18 to 9, is a completely different phenomenon. The reverse is also true. A rise from 9 to 18 would not necessarily be as bearish as a rise from 15 to 30, or 20 to 40, or heaven forbid, another jump from 40 to 80.

To use VIX effectively, be sensitive to three things:

1. The level of the index; it's numeric value
   
   
2. The trend of the index; whether its moving up or down and how quickly it's moving
   
   
3. One or more additional indicators that give you a sense of how stock prices are moving; if VIX is in the process of jumping from 40 to 80 and other indicators are bullish, back up the truck!

An Example

Figure 2 shows what would have happened had one just bought and held the S&P 500 SPDR ETF (SPY) from 3/31/01 through 5/25/10. (The difference between SPY and the index reflects slippage. Even SPY isn't immune!)

Figure 2

Figure 3 reflects the addition of a simple timing rule that puts me on the sidelines if the 50-day moving average of the S&P 500 is below the 200-day moving average. (Assume weekly rebalancing.)

Figure 3

Here's the screen:

ticker("spy")
sma(50,0,#bench)>sma(200,0,#bench)

Under bullish conditions, I get one ETF in my list. When conditions are bearish, the result set is zero.

That's actually not bad overall. But notice that as horrible as things have been in recent weeks, we're still in the market. Waiting for twists and turns between 50- and 200-day moving averages may be fine for most occasions, but perhaps we've encountered one right now where that will produce more pain than we can tolerate.

Let's speed things up to catch turns more quickly. Figure 4 shows a strategy in which the timing signal pushes me out of the market when the 5-day moving average drops below the 20-day average.

Figure 4

That got me out of the market. It worked . . . sort of. The problem, here, is that a chart like this leaves me to wonder if I should have bothered with stocks at all. Why not simply pick a good short-term treasury ETF and just hold that. Figure 5 shows what would have happened had I just held onto the iShares 1-3 Year Treasury ETF (SHY).

Figure 5

If we want equity-like returns, we're going to have to give ourselves a chance to let the stock market work. We can't constantly get shaken out by an overly jumpy timing model. We might appreciate jumpiness in May, 2010, but more often than not, we won't.This dilemma is probably very familiar to most who work with technical analysis, especially moving averages. The shorter the averages, the more quickly you can catch major turn, but shorter averages will also catch a lot of trivial turns. It's always a struggle to find the right balance.

It just seems that there are certain times when the market is especially antsy and you want to move fast and other times, most times, when the market is normal and you want to give trends a chance to unfold properly.

VIX might help decide which strategy to use at a particular point in time.

Here's a first stab at a moving-average timing model that mainly uses the 50-200 day combination, but switches over to a 5-20 combination if VIX is especially high.

ticker("spy")
Eval(Close(0,GetSeries("$VIX")) >=30, sma(5,0,#bench)>sma(20,0,#bench),sma(50,0,#bench)>sma(200,0,#bench))

Eval is the portfolio123 if-then function. The syntax is Eval(condition, rule 1, rule 2). If the condition is met, apply rule 1. If the condition is violated, apply rule 2.

In this case the condition is a very volatile market, with VIX being 30 or higher. If the condition is met - if the market is very volatile - apply rule 1, which tells us we can buy only if the 5-day S&P 500 moving average is above the 20-day average. If the condition is violated (i.e., if the market is not horribly volatile), use rule 2, the regular one, which tells us to be bullish of the 50-day moving average is above the 200-day average.

Figure 6 shows the result.

Figure 6

For the most part, this looks like Figure 3, our basic 50-200 day protocol. Notice, though, the most recent past. We went bearish (on May 8th) because the current high VIX level caused the screen to switch to a 5-20 day moving average signal.

This is by no means the last word. VIX just joined the portfolio123 family, so I'm sure I and other users will do quite a bit with it in the future. The idea, here, is to demonstrate how VIX might help you detect what kinds of indicators might be more useful at different points in time.