Breakout trend following is one of the oldest systematic trading ideas: buy when price pushes to a new high, keep losses small with a predefined stop, and let the occasional large move pay for the many small ones. It is simple to describe and hard to follow, because most individual trades are unremarkable or lose. This guide explains how the approach works, the filters that make it more robust, and the conditions in which it reliably fails.

What a breakout is

A breakout happens when price moves beyond a level it has respected for a while. The most common systematic definition is:

breakout = the candle closes above the highest high of the previous N candles

With N = 20 on 4-hour candles, that means a close above the highest price of the last three and a bit days. Using the close rather than a wick means a brief spike that is sold back within the same candle isn't counted. The idea is simple: trending markets tend to keep making new highs.

On its own, though, a breakout rule fires constantly, including in sideways markets where price pokes above a range and falls straight back in. Those false breakouts are the main cost of the strategy, which is why practical versions add filters.

Trend filters: only trade with the bigger picture

A long moving average

A simple first filter is to take a long breakout only when price is above a slow moving average, such as the 200-candle EMA. A new high inside a long-term downtrend is often just a bounce; above a rising 200 EMA it is more likely to be part of an established move. Our RSI and EMA guide explains how the EMA is calculated.

Trend strength with ADX

The Average Directional Index (ADX), another of J. Welles Wilder's indicators, measures how strongly price is trending on a 0-100 scale, regardless of direction. Readings below about 20 suggest a range; readings of 25 or more are commonly treated as a trending market. Requiring ADX ≥ 25 removes many breakouts from flat, choppy periods, which is exactly where false breakouts cluster.

The market regime filter

Crypto assets move together. When Bitcoin is in a bear market, most altcoins fall with it, and their breakouts tend to fail within days. A regime filter addresses this at the level of the whole market: only open new long trades while Bitcoin's daily close is above its 200-day average.

2022 is the clearest illustration. Bitcoin spent most of the year below its 200-day average, and long breakouts across the market were repeatedly bought at local highs and stopped out as prices kept sliding. A regime filter doesn't make a strategy profitable in a bear market; it mostly keeps it out of one. The cost is that some genuine early rallies are missed while the filter is still switched off.

Volatility-based stops and targets

A fixed percentage stop ignores the fact that some coins routinely move 2% in four hours and others 8%. The Average True Range (ATR), commonly averaged over 14 candles, measures how far a coin typically moves per candle, so a stop based on it adapts to each market.

A common set of rules:

  • Stop loss: 3 ATR below the entry. Wide enough that ordinary noise doesn't trigger it.
  • Risk unit (R): the distance from entry to stop.
  • First target (TP1): entry + 2R.
  • Second target (TP2): entry + 4R.

Worked example

Price breaks out and closes at $60,000, and the 4-hour ATR is $900. The stop goes 3 × $900 = $2,700 lower, at $57,300, so R = $2,700 (4.5%). TP1 is $60,000 + 2 × $2,700 = $65,400 and TP2 is $60,000 + 4 × $2,700 = $70,800. Many traders take part of the position off at TP1 and move the stop to the entry, so the rest of the trade can no longer turn into a loss. For more on these levels see entry, stop loss and take profit explained.

Some versions also add a time limit: if neither the stop nor a target is reached within a set period, the trade is closed at the market price, so many trades end with a small gain or loss rather than at a level.

Why a low win rate can still be profitable

With a stop 3 ATR away and targets several R beyond it, the first target is a long way off, and many trades never reach it. In our own five-year backtest of this approach, only about one trade in six reached TP1. That sounds poor until you look at how the other trades ended: under a third hit the full stop, and over half were closed at the time limit with results split roughly evenly between small gains and small losses.

What decides profitability is expectancy, the average result per trade:

expectancy = (win rate × average win) − (loss rate × average loss)

An illustrative example, not real data: if 35% of trades close in profit averaging +2R and 65% close at a loss averaging −0.8R (because many losers are time-limit exits, not full stops), expectancy is 0.70R − 0.52R = +0.18R per trade. A strategy can be right less than half the time and still make money, provided the winners are clearly bigger than the losers. The reverse is also true, which is why a headline win rate tells you little on its own.

The practical consequence is psychological: long strings of stopped and flat trades are normal, and you can't know in advance which breakout will become the large move.

When it fails

  • Sideways markets. Ranges produce repeated breakouts that reverse. Trend and ADX filters reduce this but can't eliminate it.
  • Bear markets. Without a regime filter, a long-only breakout strategy keeps buying rallies that fail. With one, it mostly sits out, which preserves capital but earns nothing.
  • Sharp reversals. News or liquidations can push price through a stop in one candle, so the actual loss is larger than planned.
  • Flat years. Even with filters, there are stretches of months where gains and losses roughly cancel out. Trend following earns most of its returns in a minority of strong trending periods.
  • Over-tuning. Settings picked because they looked best on past data usually do worse afterwards. Our guide on backtesting without fooling yourself covers this.

Fees and slippage

Every trade pays fees on entry and exit, typically around 0.1% per side on spot exchanges before discounts, so 0.2% per round trip. Breakout entries also tend to fill slightly above the signal price, because you are buying while price is rising, and stops in fast markets fill slightly below. On a strategy whose average edge is around one percent per trade, these costs are a meaningful share of the result. Any backtest or track record you rely on should state that it includes them.

Sizing the trades

Because the stop distance is set by volatility, it varies from trade to trade: perhaps 4% on Bitcoin and 12% on a volatile meme coin. The position size should vary with it, so that a stopped trade always costs the same small share of your account. The formula and worked examples are in our position sizing guide. With a low hit rate and normal losing streaks, risking 1% or less per trade is sensible; several open breakouts at once are also strongly correlated, so cap your total open risk.

How SignalEye uses this approach

SignalEye's signals come from a rule-based breakout strategy of this kind, chosen after a five-year backtest. It is buy-only, scans 4-hour candles shortly after each one closes, requires the trend and regime filters described above, and sets the stop and targets from ATR. Each signal is valid for 72 hours, after which an open trade is closed at the market price. The backtest showed a modest edge after fees that was uneven from year to year, with 2022 losing slightly and 2025 roughly flat; past results don't guarantee future ones. You can check how live signals actually turned out on the public track record.