Breakout Indicator

What a breakout indicator is

A breakout is a price move beyond a level that the market has recently stayed inside. A breakout indicator automates the drawing. It picks a level from past bars, shows it on a chart, and flags the first bar that goes past it. The flag is usually a close beyond the level, which is stricter, or a touch (the high or low reaching it), which is looser and fires more often.

The indicator does not know why the price moved. It compares the latest bar with a number computed from older bars. Because the older bars are fixed, the level is known in advance, but the flag can only be raised once the new bar has done the breaking. That timing is the root of most of the problems below.

The common kinds and their formulas

Donchian channel

The upper line is the highest high of the last N bars and the lower line is the lowest low of the last N bars. A break is a close above the upper line or below the lower line. For a rule that can be tested, use the N bars before the current bar, so the bar being tested is not part of its own line.

Range or opening-range box

A fixed window of bars, such as the first part of a session, gives a high and a low. The box is those two numbers held flat. The flag fires when a later bar closes outside the box. See opening range breakout for a full worked version.

ATR band or volatility breakout

The average true range (ATR) is the average of the true range over N bars, and the true range of a bar is the largest of high minus low, the high minus the previous close, and the previous close minus the low. A volatility breakout asks for a move beyond the previous close by a set number of ATRs. The band moves with how active the market has been, so a quiet market needs a smaller move. The Supertrend indicator is built from the same ATR.

Band breaks

Some indicators use a band around a moving average, set by the standard deviation of the price or by a multiple of the ATR. A close outside the band is flagged as a break. These are generic band types, and the Bollinger Bands page shows the arithmetic. A band is not a fixed level, so it can widen away from the price as the move happens. Some indicators also draw a cloud or several lines; the Ichimoku cloud is one example of a multi-line tool whose edges are sometimes read as break levels.

A worked example on 12 made-up bars

The table is made up. It uses N = 5. The Donchian high at a bar is the highest high of the five bars before it, so it can first be read at bar 6. A bar is marked a break when its close is above that line.

Look at bar 9. The five bars before it are bars 4 to 8, and their highest high is 104. The close of 105 is above 104, so this is the first bar the indicator flags. Bars 6 to 8 all closed at or below the line, so nothing was flagged. The five-bar low reads 100, and the box formed by the first five bars is 99 to 103.

The ATR threshold on the same table

Take k = 0.5 and a 5 bar ATR. The true range of each bar is computed from the table, and the ATR known before bar 9 (the average of bars 4 to 8) is 2.00. The threshold is the previous close of 103 plus 0.5 x 2.00, which is 104.00. The close of 105 at bar 9 is above it, so the volatility rule first flags bar 9, the same bar as the Donchian rule. Here the two rules agree. With other settings they can mark different bars, which shows how much the choice of rule shapes the signal.

The "I missed it" problem

An indicator marks the break only when the bar has closed. If you see the mark and then act, the entry is on the next bar or later, at a price that has already moved. Chasing the mark is built into the method.

In the example the flag is at bar 9 with a close of 105. Suppose the entry happens 3 bars late, at the close of bar 12, which is 106. The difference is 106 - 105 = 1 price point paid for nothing. The highest high from the break bar onward is 109, which is 5 points above the Donchian line, so the late entry gave away 1 of that 5. It is made-up arithmetic, and a real market may do more or less, but the direction of the effect is the same every time: a later entry is a worse entry on a move that has already begun. A bot that reads the closed bar and sends the order at once reduces this, and it still pays slippage because the market keeps moving while the order travels.

Settings and the overfitting risk

Every breakout indicator has settings: the length N, the multiplier k, the window of a box. Past data will always show some setting that looks best. If you scan N from 5 to 60 and keep the one with the highest past profit, you have picked it because of the past, and its good result may be luck. That is overfitting. A rule with fewer settings, tested across many markets and periods, with the results shown for neighbouring values as well as the best, is harder to fool. A good sign is a result that changes little when N moves a little. A bad sign is one sharp peak.

What an indicator cannot tell you

• Whether the break will hold. Price often pokes beyond a level and returns. That is a false breakout, covered on the false breakout page.
• How far it will go. The flag has no target. The size of the follow-through has to come from a separate rule.
• Whether the cost is worth it. Each flag is a trade, and the spread, fees and slippage are paid whether the break holds or not.
• Whether the market is in the right state. A break rule can meet many reversals in a quiet range, and the indicator does not warn you.

How a bot would use one, and how to test it

A bot reads a closed bar, compares its close with the line, and if the rule is met, sends an order and a stop. See breakout trading for the strategy side. Code for an indicator can be written in a script language such as Pine Script for charts, or built from a library such as TA-Lib in a bot.

To test it, follow the steps in how to backtest. Use the bar's close to decide and enter on the next bar, so the test does not use information the bot would not have had. Include costs. Count the trades. Then work out the expectancy: the average result per trade after costs. A rule with a handful of trades, or a result that depends on one setting, tells you very little.

What this page does not tell you

It does not say a breakout indicator works or that any setting is good. The prices are made up and this is education, not a signal.