The Average True Range (ATR) indicator is one which falls in the general category of
volatility-based technical analysis tools. It is so because like Bollinger Bands, another
volatility-based study, it does not focus on direction in any way, but rather how much raw
movement there is in price. To understand this a little better, it is worth taking a look at how
the indicator is calculated.

ATR Calculation

The ATR calculation starts with determining the True Range (TR). The TR for a given
period is defined as being the largest of:

Current Period High minus Current Period Low
Current Period High minus Previous Period Close
Previous Period Close minus Current Period Low

Introducing Technical Analysis 101 by DayTradersGroup.com

ATR is the average of the True Range (TR) over the past n periods. This is calculated the same as any standard simple moving average. The default setting is generally 14 periods.

As can be seen by examining the determination of TR, the study is attempting to capture the amount of actual price moment which has taken place during a trading period, which differs somewhat from the trading action. The latter could be defined as where the market actually traded during the course of a trading period, but the former actually takes in to account price gaps between periods.

It should be observed that ATR as calculated is not in any way normalized.  That is to say because it is purely a price measure, ATR cannot easily be used to compare different securities as the indicator's readings would expected to be different for instruments with different prices.  For example, one would expect ATR to be significantly higher for a stock trading at 100 than for one trading at 20.

In order to compare ATR across securities, or even across timeframes for the same security, it would have to be normalized.  This can be accomplished by dividing the ATR by some current price measure such as the most recent close or a moving average.  That would given a reading of ATR as a percentage of whatever that standard divisor is.  So if ATR on a 100 stock is 5, then the normalized ATR would be 5% (assuming the close is used for the calculation).

The best use of ATR is in Pair trading but it often not mentioned in any literatures,

Let say you want to go LONG GOOG 100 Shares and you looking to hedge another stock such as YHOO against it.

Number of Shares in YHOO = ATR GOOG / ATR YHOO * number of shares in GOOG

You do this just before taking your the position

On Friday the correct number of sharing hedging YHOO against GOOG would have been 800 YHOO against 100 GOOG . Remember these numbers vary every day due to stock's
daily volatility .

There are other uses of ATR in Multi Time frame analysis but that only applies to algorithmic program trading .. ( you can hedge multiple assets across multiple time frames using ATR
For those trying this at home, a few caveats apply. First, ATR values depend on where you begin. The first True Range value is simply the current High minus the current Low and the
first ATR is an average of the first 14 True Range values. The real ATR formula does not kick in until day 15. Even so, the remnants of these first two calculations linger to slightly
affect ATR values. Spreadsheet values for a small subset of data may not match exactly with what is seen on the price chart. Decimal rounding can also slightly affect ATR values.

ATR is based on the True Range, which uses absolute price changes. As such, ATR reflects volatility as absolute level. In other words, ATR is not shown as a percentage of the
current close. This means low priced stocks will have lower ATR values than high price stocks. For example, a $20-30 security will have much lower ATR values than a $200-300
security. Because of this, ATR values are not comparable. Even large price movements for a single security, such as a decline from 70 to 20, can make long-term ATR comparisons
impractical. Chart 4 shows Google with double digit ATR values and chart 5 shows Microsoft with ATR values below 1. Despite different values, their ATR lines have similar shapes.
ATR is not a directional indicator, such as MACD or RSI. Instead, ATR is a unique volatility indicator that reflects the degree of interest or disinterest in a move. Strong moves, in either
direction, are often accompanied by large ranges, or large True Ranges. This is especially true at the beginning of a move. Uninspiring moves can be accompanied by relatively
narrow ranges. As such, ATR can be used to validate the enthusiasm behind a move or breakout. A bullish reversal with an increase in ATR would show strong buying pressure and
reinforce the reversal. A bearish support break with an increase in ATR would show strong selling pressure and reinforce the support break.

Absolute values are used to ensure positive numbers. After all, Wilder was interested in measuring the distance between two points, not the direction. If the current period's high is
above the prior period's high and the low is below the prior period's low, then the current period's high-low range will be used as the True Range. This is an outside day that would use
Method 1 to calculate the TR. This is pretty straight forward. Methods 2 and 3 are used when there is a gap or an inside day. A gap occurs when the previous close is greater than the
current high (signaling a potential gap down or limit move) or the previous close is lower than the current low (signaling a potential gap up or limit move). The image below shows
examples of when methods 2 and 3 are appropriate.

During the summer months the bollinger bands will range outside of historical mean reversion as it is a time when many traders losses outweigh their winnings as the months between
May thru September that have historically proven to be some of the toughest months to trade due to intense volatility coupled with low volume. Seasoned veterans, financial analysts
and some of the worlds largest hedge fund managers will employ Professional Wall Street traders to guide them through the toughest season of the year.