Volatility is the inclination of market prices to change as measured bar-by-bar over a period of time.  A low volatility market has, on average, short price bars while a high volatility one (what we mean when we say a volatile market) has tall ones.

Volatile should not be conflated with choppy, a tendency towards repeated directional reversals.  Volatile and choppy often go together, but not always. The bull market that ended in mid-2007 first turned volatile/choppy as it topped but began plunging downwards in a very volatile fashion until it bottomed in early 2009.

The chart at the top of the page shows the unfolding of that sequence in the S&P 500 ETF, SPY.  The indicators show the 21 day versus 200 day volatility (AtrM/AtrN) as well as the 200 day momentum (Momentum Atrs).  When AtrM/AtrN is greater than 1.1 it is colored green, less than 0.9, red.  When MomA is geater than 7.5 Atrs it is colored green, less than -7.5 Atrs, red.

The financial press uses the VIX as its go-to volatility measurement, with readings above 20 interpreted as strong fear.  But they are often in disagreement as to whether the fear represents a buying opportunity or a signal to run for the hills.

During longer-term uptrends spikes in the VIX are generally profitable buys.  During tops, bottoms, and downtrends VIX peaks can mean pretty much anything that selling has been exhausted, that selling will continue, sitting on the sidelines flat is a wise choice, etc.

Financial journalists rarely backtest their evaluations of anything, including those for market direction versus the VIX.  As traders and money managers we cannot afford to avoid testing our ideas before money is put at risk, whether ours or clients.

We also need to have a deep understanding of what we are measuring and how to interpret it in different market contexts.  The VIX is the implied volatility of a set of S&P 500 options it is the expected future volatility of options prices calculated by an options pricing formula.

So the VIX is a derived value and not really even a measurement of the volatility we are experiencing now or in the past in the underlying market.  And it only pertains to the S&P 500.  Even if we are aware of those limitations the VIX reading is not directly actionable.  What do we do to exploit high or low readings?

Exploiting the VIX

Any market whether an index, commodity, stock, etc. that has an options market can have its implied volatilities calculated for the options and use them to determine a VIX.  Other indexes, such as the NASDAQ, actually have their own VIXs.  But the same issue exists there as it does for the S&Ps VIX.

The most obvious decision for interpreting index volatilities pertains to portfolio structuring.  Buy more shares of individual stocks, sell existing stock positions, sell short, etc.

But we could use the same interpretation to play the index directly.  As mentioned above, VIX spikes during uptrends allow us high probability bets the uptrend will continue, as the spikes usually occur during sharp, short selloffs in the index.

The VIX itself is tradeable.  Via VIX futures, options, or ETFs we can make short-term bets that the VIX will revert to its longer-term average or trend, or take the position that a recent VIX short-term trend will persist.

Exploiting Options Implied Volatility

Options players have non-directional volatility trades available to them.  Options prices are a result of a number of factors with price movement and volatility and time chief among them.

If volatility rises so will option prices, all other things being equal.  If prices of the underlying market are expected to remain fairly stable high volatility options can be sold. The supposition is that volatility and time remaining will decrease leading to a decrease in options prices and a profit for the seller.

If options volatility is low option buyers can make money if volatility increases or prices move.  This can be true even if no directional bias is known buying both calls and puts can be profitable if price moves enough in either direction to cover the costs of all options purchased.

Beware though of using implied volatility as a measure of where price will trade (as many options players do).  The underlying assumption is normal distribution of the underlyings price changes plus no directional bias.

The distribution of price changes is not statistically normal (i.e. a bell curve) there are more outliers than normal predicts.  And volatility is constantly changing along with price distribution expectations.

More important is the assumption of price non-directionality prices rarely stay stationary very long.  If you can identify a directional bias it is much easier to make options profits, with or without a volatility aspect to the trade.

Aspects of Measured Volatility

Not all markets have options associated with them and often they are illiquid when they do exist.  And many traders do not want to deal with the extra complexity of options.

In a strong sense the measured volatility of a market, that from its actual price changes, is the real bread-and-butter application area of volatility.  The primary reason is volatilitys versatility   filtering a stocks universe, finding and analyzing specific setups, ranking setups for portfolio structuring, determining position sizes, identifying where stops should be placed during the life of the trade, etc.

We have several ways to measure actual volatility.  Historical volatility is the annualized standard deviation of price changes and is often used in comparisons between markets (and in options).

Average True Range measures the average price change per bar in dollars over an arbitrary time period.  Atrs are my favorite since they have the same units as price and can be used directly for stops and other purposes where prices/points are used.

Volatilities can also be relative.  Beta measures the relative volatility (via regression analysis) of a stock compared to the market, usually versus the S&P 500.  A value of 1.2 is 20% higher volatility than the market, while .5 is half that of the market.

We can also directly compare the normalized measured volatilities of two markets.  Historical volatility is one way of doing that.  The percent of the markets price that ATR represents is another way to judge the relative volatility between markets.

We can also measure the relative volatility of a market to itself (lets call it its self-volatility to differentiate it).  A 21 day volatility compared to its 200 day counterpart can tell us how volatile the market has been recently versus the longer-term.

Finally, other indicators can be adjusted for volatility.  One example is a moving average whose lookback is decreased to reflect increased volatility.  Another might be a trailing stop which slows down with increased volatility.

Exploiting Measured Volatility

In I showed how I used Toolkit components Average Volume Truncated and Atr Percent to find a universe of stocks from a starting pool of over 8000 stocks, identifying ones that have minimum 65 day average volume of over 800000 shares per day and 21 day Atr Percent of between 2.5% and 7.5%.

This ensures sufficient liquidity and volatility for all stocks in the tradeable universe.  The universe can also be used as an internals measure for the liquidity and volatility of the market as a whole.  The list size varies from under 100 to over 400 as those measures decrease and increase, respectively.  Below is a sample of the various row types seen due to alerts we want the first eight rows with alerts going off for both volume and volatility.

Atr Percent provides three Atr percent measurements for 1 day, for a small number of days, and for a longer number of days.  So it can be used to judge a markets relative self-volatility on a percent basis but can also be used to judge the relative volatility between different markets.

The Toolkit relative volatility component AtrM/AtrN directly compares a shorter-term Atr measurement with a longer-term one, providing a ratio of relative self-volatility.  With these volatility measures we can determine if volatility is high or low for a given market without any reference to options and implied volatility.

A trading system can be tested for its sensitivity to both absolute volatility, such as that for short-term Atr Percent, but also for recent volatility relative to longer-term.  So volatility can be used a trade filter and as a means to order trade signals for portfolio structuring decisions.

Volatility can also be used within trading systems for setups and to identify them for actual trading in the trading universe.  Breakouts or breakout candidates as well as trading range reversals can be found by their proximity to previous highest Highs or lowest Lows, with tookit components Atrs from HiLo or StdDev from HiLo measuring those distances.

Toolkit components Atrs from MA and StdDev from MA measure distances from moving averages.  A short-term retracement system might use downside proximity from a 10 day moving average to find setups, ordering those distances from largest to smallest to find the strongest long candidates.

A short-to-medium-term retracement system might similarly use distance from a 20 day moving average.  The popular Bollinger Band systems find setups that exceed 2.0 standard deviations away from the 20 MA counter to a longer-term trend (a long would be below the 20 MA in an uptrending market).

A medium-term trend following system might interpret deviations in an opposite manner.  A deviation that exceeded a given threshold in the direction of the longer-term trend might be used as an entry point to take positions based on strength.

Below is a snapshot of a simple medium-term screener with Toolkit components.  The relative volume, VolM/VolN (V/V) shows the 1 day volume compared to the 65 day average volume.  The relative volatility, AtrM/AtrN (A/A), compares 21 day to 200 day volatility.  The Momentum Atrs (MomA) calculates the 200 day momentum in units of Atrs (the screener is ordered by MomA descending).  HiLo counts the number of days for which today is the High (positive numbers) or Low.  The Atrs from HiLo shows the number of Atrs above the 200 day lows (positive numbers) or below the highs (whichever is closer).  The Atrs from Moving Average (AfMA) displays the number of Atrs below (negative numbers) or above the 50 day MA.

One volatility application well-known to many traders is volatility stops.  For example a medium-term directional system may trail a stop for a long position 2.5 Atrs below the highest High since entry.

Volatility stops are useful as standalone stops or as checks for support and resistance stops.  For example a long position may have a logical stop just below support.  The volatility stop can tell us if the logical stop is too close or a sufficient distance away that it wont get taken out on expected price variation (we want the volatility stop to be the closer of the two).

As I showed in once an entry price, En, and an exit price, Ex, are known a position size can be easily calculated for a given maximum percent equity loss.  The initial exit price is just the entry minus the Atrs trailed in points.

This is a great way to go for both stops and position sizing and may be the best hack for volatility that exists.  For similarly priced markets the more volatile one will have a smaller position size and stops that are further away, which is exactly what you want.

Another approach to position sizing could use a relative volatility measurement.  This time we use a single market as a standard, say the S&P 500.  Anything with the same volatility gets a 10% of equity position size and we scale to other relative volatilities proportionately one with half the S&Ps is 20% of equity, one with twice is 5% of equity, etc.

One other application area is that of indicator normalization.  The distance from HiLo and MA Toolkit components discussed above (Atrs from HiLo, etc) are examples of that.  It simply means that we use units of Atrs or StdDev instead of raw points.

The Toolkit component Momentum Atrs (and breathren Momentum StdDevs) is my most-used strength indicator.  Regular Momentum uses price change versus the beginning of the lookback period.  But if two similarly-priced markets have a momentum of 10 they are not directly comparable if they have different volatilities.  Momentum Atrs allows that comparison and would show them to be unequal.

One final but crucial point is worth considering volatility is always changing.  That is one important reason that Atr Percent uses three time periods.  If volatility changes a good or great deal while we are in position our position sizing or our stops may need to be reconsidered.