Strategies we backtested Keltner Channel Breakout

Tested by BacktestingMax

Keltner channel breakout strategy backtest

Does the Keltner channel breakout strategy make money?

On some markets, yes. A Keltner channel is three lines. The middle line is the 20-candle EMA, a moving average of the price. The other two lines are 2 times the ATR above and below the middle line. ATR is how far price moves in an average candle. The breakout strategy buys when a candle closes above the channel. It sells short (sells first, to make money if price falls) when a candle closes below the channel. We ran 30 backtests from 2020 to 2025, after trading costs, each with a $10,000 account. They cover 10 markets (5 currency pairs, gold, silver and 3 US stock indices), each on 15-minute, 1-hour and 4-hour charts. With the standard rules (no changes), 10 out of 30 backtests made money, and 8 made money in both 2020–2023 and 2024–2025. On the Nasdaq 100, all 3 backtests made money, and the 15-minute backtest grew the account by 244.2%. On EUR/USD, GBP/USD, AUD/USD and USD/CAD, the strategy lost money on every chart, even before trading costs. Trading the other way (the fade) made money on the 4-hour charts of AUD/USD, GBP/USD and EUR/USD. On 1-hour charts, the strategy made money on 6 out of 10 markets instead of 3 when it waited for a squeeze, a quiet time before the breakout.

30 backtests 2020–2025 Trading costs included Tested 9 November 2026

Where it worked 30 backtests
Made money
10
of 30 tests, with the standard rules
Made money in both periods
8
2020–2023 and again 2024–2025
Best backtest
+244%
Nasdaq 100, 15-minute chart: account growth 2020–2025
With a squeeze filter
4
tests made money in both periods, compared with 8 with the standard rules

Made money in both periods

  • USD/JPY, 1-hour
  • Gold, 4-hour
  • Silver, 4-hour
  • Nasdaq 100, 15-minute
  • Nasdaq 100, 1-hour
  • Nasdaq 100, 4-hour
  • S&P 500, 15-minute
  • Dow Jones, 4-hour

The results

Results on every market

Each box shows one backtest of the Keltner Channel Breakout with the standard rules, from 2020 to 2025, after trading costs: the % result, and a small chart of the account balance. The solid line is the account balance and the dashed line is the starting balance. Green results made money; red results lost money. Under each result: whether it made money in both 2020–2023 and 2024–2025, or passed all four of our checks (see "How a test passes" further down).

Market 15-minute 1-hour 4-hour
EUR/USD −99% Lost money. Account down 90% by Dec 2022 −47% Lost money −18% Lost money
GBP/USD −97% Lost money. Account down 90% by Oct 2023 −39% Lost money −35% Lost money
USD/JPY −65% Lost money +152% Made money in both periods +21% Made money overall, lost in one period
AUD/USD −99% Lost money. Account down 90% by Dec 2022 −56% Lost money −42% Lost money
USD/CAD −99% Lost money. Account down 90% by Jul 2023 −49% Lost money −14% Lost money
Gold −92% Lost money. Account down 90% by Sept 2023 +43% Made money overall, lost in one period +17% Passed all four checks
Silver −97% Lost money. Account down 90% by Feb 2022 −26% Lost money +35% Passed all four checks
Nasdaq 100 +244% Made money in both periods +73% Made money in both periods +3% Made money in both periods
S&P 500 +12% Made money in both periods −5% Lost money −14% Lost money
Dow Jones −64% Lost money −11% Lost money +7% Made money in both periods
All the numbers, test by test
Market Chart Trades Win rate Profit factor Result Max drawdown 2020–2023 2024–2025 Monte Carlo drawdown Trading costs
EUR/USD 15-minute 3,206 29.4% 0.76 −99.0% −99.0% −96.8% −67.6% −99.2% $5,013
EUR/USD 1-hour 897 32.8% 0.87 −47.4% −52.4% −18.5% −35.5% −64.1% $2,191
EUR/USD 4-hour 236 31.4% 0.84 −17.8% −28.9% −1.2% −16.8% −33.7% $326
GBP/USD 15-minute 3,426 29.7% 0.83 −97.2% −97.4% −91.6% −66.2% −98.2% $7,214
GBP/USD 1-hour 878 31.4% 0.89 −38.9% −53.7% −26.5% −16.9% −60.1% $2,324
GBP/USD 4-hour 237 29.1% 0.64 −35.2% −36.3% −26.5% −11.7% −43.6% $289
USD/JPY 15-minute 3,411 31.3% 0.94 −64.9% −79.1% −58.2% −16.1% −87.5% $12,701
USD/JPY 1-hour 845 37.6% 1.21 +151.5% −23.7% +92.0% +31.0% −34.4% $3,869
USD/JPY 4-hour 226 34.5% 1.16 +20.7% −15.5% +21.5% −0.7% −22.9% $332
AUD/USD 15-minute 3,348 29.6% 0.77 −99.2% −99.3% −94.9% −83.7% −99.4% $6,119
AUD/USD 1-hour 855 32.2% 0.85 −55.9% −63.3% −20.9% −44.3% −67.0% $2,858
AUD/USD 4-hour 225 26.7% 0.56 −41.8% −48.8% −32.1% −14.3% −46.8% $335
USD/CAD 15-minute 3,241 27.7% 0.75 −99.4% −99.4% −94.3% −90.0% −99.6% $6,216
USD/CAD 1-hour 858 33.3% 0.84 −49.0% −56.5% −33.2% −23.7% −63.6% $2,517
USD/CAD 4-hour 220 30% 0.85 −14.3% −25.5% −8.5% −6.3% −30.7% $417
Gold 15-minute 2,705 29.6% 0.82 −91.9% −93.5% −88.9% −27.0% −94.6% $6,373
Gold 1-hour 800 33% 1.08 +43.1% −27.8% +45.3% −1.5% −38.4% $2,909
Gold 4-hour 220 33.2% 1.16 +16.8% −23.6% +4.2% +12.1% −24.0% $305
Silver 15-minute 1,486 24.6% 0.62 −96.5% −96.7% −96.4% −3.6% −97.4% $6,771
Silver 1-hour 721 32.6% 0.93 −26.3% −52.2% −16.6% −11.6% −53.8% $4,981
Silver 4-hour 189 41.8% 1.39 +35.4% −9.1% +26.3% +7.2% −14.9% $776
Nasdaq 100 15-minute 3,301 32.1% 1.06 +244.2% −40.1% +147.5% +39.1% −67.1% $25,802
Nasdaq 100 1-hour 883 36.6% 1.13 +73.4% −24.3% +33.6% +29.8% −36.2% $1,776
Nasdaq 100 4-hour 253 37.2% 1.02 +2.7% −10.0% +2.6% +0.1% −22.9% $174
S&P 500 15-minute 3,309 33% 1.01 +12.1% −55.1% +0.9% +11.1% −77.6% $16,956
S&P 500 1-hour 914 35.6% 0.99 −5.0% −30.2% +11.3% −14.7% −49.0% $1,868
S&P 500 4-hour 253 33.6% 0.88 −13.9% −25.8% −8.7% −5.7% −31.7% $203
Dow Jones 15-minute 3,252 31.6% 0.95 −64.1% −75.1% −52.9% −23.6% −86.0% $7,305
Dow Jones 1-hour 896 33% 0.97 −10.7% −40.1% −21.8% +14.1% −46.8% $922
Dow Jones 4-hour 231 32.5% 1.07 +6.8% −15.8% +5.5% +1.3% −21.4% $116

Profit factor is the money won divided by the money lost. Above 1 means the test made money. Max drawdown is the biggest fall from the highest balance. 2020–2023 and 2024–2025 show the result in each period on its own. Monte Carlo drawdown: we shuffled the trades into a random order 1,000 times; in 95% of those runs, the worst drawdown was this number or less. Result is the change in the account, with the risk of each trade set to 1% of the account at that time.

The account balance

The best test and the middle test

Best test Nasdaq 100, 15-minute chart +244.2% from 2020 to 2025
Middle test (median): half of the 30 did better, half did worse EUR/USD, 4-hour chart −17.8% from 2020 to 2025

What we found

What the 30 backtests show

  1. With the standard rules, 8 out of 30 backtests made money in both 2020–2023 and 2024–2025, after trading costs. The Nasdaq 100 made money in both periods on all 3 charts: +244.2% on 15-minute charts, +73.4% on 1-hour charts and +2.7% on 4-hour charts. Two more were USD/JPY on 1-hour charts (+151.5%) and the S&P 500 on 15-minute charts (+12.1%). The last three were gold (+16.8%), silver (+35.4%) and the Dow Jones (+6.8%), all on 4-hour charts. Gold and silver on 4-hour charts also passed all four of our checks, so the results grid marks them as passed. But both made most of their profit in one year, 2020. To compare buys and sells, we measure each trade in R. R is the trade's result divided by 1% of the account at that time, the amount each trade was set to risk. So 0R is break-even. On the Nasdaq 100's 15-minute chart, the buys made money in every year from 2020 to 2025, +166.1R in total. The sells made +65.5R in 2022, when the Nasdaq 100 fell 32.9%, but lost 56.6R in the other 5 years together.

  2. On EUR/USD, GBP/USD, AUD/USD and USD/CAD, the breakout strategy lost money on all 3 charts, even before trading costs. On these 4 pairs, trading the other way did better. Trading the other way is called the fade. The fade sells when a candle closes above the channel, and buys when a candle closes below the channel. It closes the trade when price closes back across the middle line. Before trading costs, the fade made money in 11 out of 12 backtests on these 4 pairs. After trading costs, it made money on the 4-hour charts of AUD/USD (+36.4%), GBP/USD (+30.7%) and EUR/USD (+1.8%), in both periods. On the 1-hour chart of USD/CAD, it made +16.4%, but it lost money in 2024–2025. On the other 6 markets, the fade lost money in 16 out of 18 backtests (see the comparison below).

  3. On 1-hour charts, breakouts that came soon after a squeeze did better. A squeeze is a quiet time, measured with Bollinger Bands. Bollinger Bands are two lines around the average of the last 20 closes, 2 standard deviations above and below that average. The standard deviation measures how spread out those closes are. In a squeeze, the Bollinger Bands are inside a narrower Keltner channel: the same middle line, with the outer lines 1.5 times the ATR away from it, instead of 2. John Carter's TTM Squeeze indicator uses the same rule. With the standard rules on 1-hour charts, 3,062 trades, counting all 10 markets, came after a squeeze on at least one of the 5 candles before the breakout candle. These trades made +0.054R each on average. The other 5,485 trades lost 0.039R each on average. On 15-minute and 4-hour charts, both groups lost money (see the chart below).

  4. On daily charts, the standard rules made money on only 2 out of 10 markets from 2014 to 2025: gold (+14.5%) and USD/JPY (+12.9%). Both made money in 2014–2021 and again in 2022–2025. These daily charts are 10 extra backtests, separate from the 30 above, with more years because a daily chart gives few trades. Each market made only 60 to 79 trades in those 12 years, about 6 a year. On the Nasdaq 100, the account lost 6.9%, while the Nasdaq 100's price rose 603.2%. The fade made money on 3 markets on daily charts, but on none of them in both periods (see the table below).

Breakout or fade

On 4 currency pairs, trading the other way did better than the breakout

Trades of the standard rules (the breakout) and of the fade, on 1-hour and 4-hour charts from 2020 to 2025, after trading costs. 15-minute charts are in the note below. The 4 pairs, where the breakout lost money, are EUR/USD, GBP/USD, AUD/USD and USD/CAD. The other 6 markets are USD/JPY, gold, silver, the Nasdaq 100, the S&P 500 and the Dow Jones. Each number is the average result of one trade, in R.

The 4 pairs: breakout
−0.082R
4,406 trades. 0 out of 8 backtests made money.
The 4 pairs: fade
+0.008R
5,479 trades. 4 out of 8 backtests made money.
The other 6 markets: breakout
+0.040R
6,431 trades. 8 out of 12 backtests made money.
The other 6 markets: fade
−0.046R
8,220 trades. 2 out of 12 backtests made money.

We added up the results of all the trades in each group and divided by the number of trades. Before trading costs, the breakout lost 0.050R per trade on the 4 pairs and made +0.064R per trade on the other 6 markets. Before trading costs, the fade made +0.041R per trade on the 4 pairs and lost 0.023R per trade on the other 6 markets. On 15-minute charts, the breakout lost money on 8 out of 10 markets, and the fade lost money on all 10. On 15-minute charts, the fade made +0.024R per trade before trading costs, but trading costs took 0.066R per trade.

The squeeze

On 1-hour charts, trades taken soon after a squeeze made money on average, and the other trades lost money

Trades of the standard rules, after trading costs, with all 10 markets added together on each chart. The years are 2020 to 2025 on 15-minute, 1-hour and 4-hour charts, and 2014 to 2025 on daily charts. 'After a squeeze' means the Bollinger Bands were inside the narrower Keltner channel on at least one of the 5 candles before the breakout candle. Each bar is the average result of one trade, in R.

  • 15-minute: after a squeeze11,491 trades −0.054R
  • 15-minute: other trades19,194 trades −0.078R
  • 1-hour: after a squeeze3,062 trades +0.054R
  • 1-hour: other trades5,485 trades −0.039R
  • 4-hour: after a squeeze649 trades −0.021R
  • 4-hour: other trades1,641 trades −0.025R
  • Daily: after a squeeze194 trades, 2014–2025 +0.029R
  • Daily: other trades495 trades, 2014–2025 −0.030R

We added up the results of the trades in each group and divided by the number of trades. On 1-hour charts, the trades after a squeeze did better than the other trades on 9 out of 10 markets, and made money on 6 of the 10 markets. On 4-hour charts, they did better on 6 out of 10 markets, including the 3 stock indices, but worse on gold, silver, USD/JPY and EUR/USD.

Daily charts

On daily charts from 2014 to 2025, the standard rules made money only on gold and USD/JPY

Each result is the change in the account from January 2014 to December 2025, after trading costs, risking 1% of the account on each trade. The fade and the exit back inside the channel are two of the changes to the standard rules. They are explained under 'Can the Keltner Channel Breakout be improved?' below. The last column is the change in the market's price over the same years. The last row adds up the results of the trades on all 10 markets, in R instead of %.

Market Trades, standard rules Standard rules Fade Exit back inside the channel Price change
EUR/USD 70 −3.8% −0.7% +0.4% −14.1%
GBP/USD 77 −9.8% −2.5% +0.6% −18.1%
USD/JPY 65 +12.9% −3.0% +7.3% +49.5%
AUD/USD 68 −6.9% −4.8% −5.8% −25.1%
USD/CAD 67 −2.4% +6.9% −2.2% +28.6%
Gold 66 +14.5% −10.1% +9.5% +252.8%
Silver 63 −0.2% −8.6% +4.9% +257.8%
Nasdaq 100 79 −6.9% −9.3% −10.5% +603.2%
S&P 500 74 −6.0% +1.5% −11.2% +270.6%
Dow Jones 60 −2.4% +0.9% +5.2% +190.2%
All 10 markets, in R 689 −9.0R −28.5R −2.2R —

Most daily backtests made fewer than 100 trades in 12 years: 60 to 79 per market with the standard rules, and 69 to 97 with the fade. The exit back inside the channel made 108 to 168. So the daily results are less certain than the results on shorter charts. We left out the squeeze filter, because it made only 14 to 30 trades per market on daily charts. On the Dow Jones, the standard rules skipped 21 out of 81 signals. For those 21 signals, a trade that risked 1% was smaller than the smallest trade size the market allows.

The strategy

What the Keltner Channel Breakout is

A Keltner channel is an indicator made of three lines around the price. The middle line is a moving average of the price. Here it is the 20-candle EMA (exponential moving average), which gives more weight to the newest candles. The upper line is 2 times the ATR above the middle line, and the lower line is 2 times the ATR below the middle line. ATR (average true range) is how far price moves in an average candle. So the channel gets wider when price moves a lot (high volatility), and narrower when price is quiet. Chester Keltner described the first version in 1960. The version with the EMA and the ATR is the one most charting platforms show today. Traders use the channel in two opposite ways. A breakout trader reads a close outside the channel as the start of a strong move, and trades in the direction of the move. A mean reversion trader expects price to come back to the middle line, and trades against the move. The standard rules on this page trade the breakout. How often does price close outside the channel? On 1-hour charts, 15.1% of the candles in our data closed outside the channel, about 1 candle in 7. Most of these closes came right after another close outside the channel. A new breakout is a close outside the channel after a close inside it. On 1-hour charts, each market had 222 to 296 new breakouts a year. About half of these new breakouts came while a trade was still open, so they were not traded.

Keltner channel
The middle line is the 20-candle EMA of the close. The upper and lower lines are 2 × ATR(10) above and below the middle line. ATR(10) is the standard average true range (Wilder's) of the last 10 candles: how far price moves in an average candle, including any jump from the previous close. These are the default settings on TradingView. We worked out the channel ourselves, on the same candles the rules use (see 'Price data').
Stop-loss
2 × ATR(14) from the close of the breakout candle. The stop-loss does not move.
Exit
A close back across the middle line: below it for a buy, above it for a sell. The trade closes at the next candle's open.
Take-profit
None
Risk per trade
1% of the account, which started at $10,000. The trade size is worked out at the close of the breakout candle, and rounded down to a size the market allows. The trade opens at the next candle's open. So a loss at the stop-loss was usually close to 1%, but not exactly 1%. When price jumped past the stop-loss, the loss was bigger. With the standard rules, 24 trades on 15-minute charts, 5 on 1-hour charts and 1 on 4-hour charts lost more than 1.5% of the account. No daily trade did. The biggest loss, 7.5%, was an S&P 500 trade on the 15-minute chart on 12 January 2023. The US inflation report (CPI) came out at 08:30 New York time. The price fell more than 50 points in one minute, far past the stop-loss.
Trading costs
The spread, which is the gap between the buy price and the sell price. On currency pairs, the spread is in pips: a pip is 0.0001 on EUR/USD, GBP/USD, AUD/USD and USD/CAD, and 0.01 on USD/JPY. The spread is 0.8 pips on EUR/USD, 1.2 on GBP/USD, 0.9 on USD/JPY, 1.0 on AUD/USD and 1.3 on USD/CAD. It is $0.30 an ounce on gold and $0.020 on silver. On the indices, it is 0.4 points on the S&P 500, 1.5 on the Nasdaq 100 and 2.0 on the Dow Jones. For silver, the Nasdaq 100 and the Dow Jones, these spreads are wider than the spreads BacktestingMax uses by default, to match a typical retail broker. These rules use only market orders and stop-losses, so every trade pays half the spread when it opens and half when it closes. With the standard rules, the spread cost a trade 6.5% of the money it risked (0.065R), on average, on 15-minute charts. It cost 3.1% on 1-hour charts, 1.4% on 4-hour charts and 0.7% on daily charts. There is no commission.
Smallest trade size
Sometimes the trade size that risks 1% is smaller than the smallest trade the market allows. The smallest trade is 1,000 units of a currency pair, 1 ounce of gold, 50 ounces of silver and 0.1 of an index contract. Then the backtest skips the signal. With the standard rules, the backtests skipped 4,324 signals on 15-minute charts, 2,245 of them on silver. All but 9 of the 4,324 came after the account had fallen below $1,000, more than 90% down. Those 9 were on the Dow Jones. The backtests skipped 5 signals on 1-hour charts, 8 on 4-hour charts and 35 on daily charts. On the Dow Jones daily chart, the backtest skipped 21 signals, about 1 in 4. Our backtest engine also does not allow a trade worth more than 30 times the account balance, and a very close stop-loss can need a bigger trade than that. This happened in 3 trades on 15-minute charts, which then risked a little less than 1%.
Price data
CFD prices: the prices a broker quotes for a CFD (contract for difference), the product most brokers outside the US offer for trading these markets. The strategy skips two kinds of candle, and these candles don't count for the EMA, the ATRs or the Bollinger Bands. First, on some Fridays, the data has a few minutes of prices after the weekly close at 17:00 New York time, which make a tiny extra candle. Second, some candles show no price change at all (the high equals the low), for example on market holidays. In the backtests from 2020 to 2025, these were less than 0.3% of the candles in every backtest. In the daily backtests from 2014 to 2025, they were 5.3% of the Nasdaq 100 candles and 5.7% of the S&P 500 candles. Most of these were tiny Friday candles from 2014 to 2018. On the other markets, they were less than 0.6%.
Wrong prices with no trading
Our data for the DAX and the FTSE 100 has wrong prices at some times with no trading, so we did not test these two markets. We checked our 10 markets for the same problem: minutes with no trading, at a price more than 0.75% away from the last price with trading. From 2020 to 2025, this happened on one day, on silver. On 1 June 2023, 7 of these minutes had prices up to 1.5% above the last price with trading, and they made one extra 1-hour candle. We ran silver's 1-hour backtests again without these 7 minutes. With the standard rules, the account still lost 26.3%. The biggest change was with the squeeze filter: +24.4% instead of +25.7%, and the backtest still passed all four checks. From 2014 to 2018, the same problem came on 137 days on the Nasdaq 100, 70 days on the S&P 500, 4 days on silver and 1 day on gold. Of our backtests, only the daily ones include these years. Without these minutes, the Nasdaq 100's daily backtest with the standard rules lost 6.3% instead of 6.9%. The S&P 500's daily backtest with the standard rules lost 4.7% instead of 6.0%. The other daily backtests on these two markets changed by $160 or less on the $10,000 account. None of them changed from a loss to a profit, or from a profit to a loss. The daily backtests on gold and silver did not change.
Daily charts
Tested on their own, from January 2014 to December 2025, split into 2014–2021 and 2022–2025. A daily chart gives only about 6 trades a year on each market, so it needs more years. The daily results are in the daily chart table above, not in the results grid.
Overnight fees
Not included. Brokers charge a fee (a swap), or sometimes pay one, for each night a trade stays open. With the standard rules, half of the trades were open for 3 hours or less on 15-minute charts, and for 16 hours or less on 1-hour charts. On 4-hour charts, half were open for more than 3 days, and on daily charts for more than 16 days. So overnight fees would matter most on 4-hour and daily charts.

The rules we tested

  1. Work out the Keltner channel. The middle line is the EMA of the last 20 closes. The upper and lower lines are 2 × ATR(10) above and below the middle line. ATR(10) is the average true range of the last 10 candles. We check each candle when it closes.
  2. Buy when a candle closes above the upper line, and the candle before it did not close above the upper line. The trade opens at the next candle's open.
  3. Sell short when a candle closes below the lower line, and the candle before it did not close below the lower line. Selling short means you sell first and buy back later, so the trade makes money if the price falls.
  4. Put the stop-loss 2 × ATR(14) from the close of the breakout candle: below it for a buy, above it for a sell. ATR(14) is the average true range of the last 14 candles. The stop-loss does not move. If price reaches the stop-loss, the trade closes there, or at a worse price if price jumps past it.
  5. Close a buy when a candle closes below the middle line, and close a sell when a candle closes above the middle line. The trade closes at the next candle's open. There is no take-profit.
  6. Special case: sometimes the candle that tells you to close a buy (candle 1) also closes below the lower line. The buy closes at the open of candle 2. If candle 2 also closes below the lower line, a sell opens at the open of candle 3. After a sell, the same applies with the upper line.
  7. Size every trade so that, if the stop-loss is hit, the trade loses about 1% of the account. Only one trade is open at a time: a breakout that comes while a trade is open is not traded.

The method

How we tested it

The test

Markets
EUR/USD, GBP/USD, USD/JPY, AUD/USD, USD/CAD, Gold, Silver, Nasdaq 100, S&P 500 and Dow Jones
Charts
15-minute, 1-hour and 4-hour candles
Dates
January 2020 to December 2025
Account
$10,000, risking 1% of the account on each trade
Trading costs
Included. Every trade pays that market's normal trading cost: the spread, which is the gap between the buy price and the sell price.

How a test passes

We call a test passed only if it meets all four checks. A strategy can make money in one backtest and still fail one of these checks.

  • At least 100 trades. With fewer trades, a good result could be luck.
  • It made money from 2020 to 2023 and again from 2024 to 2025. This shows whether the strategy kept working in the newer years, not only in the older ones.
  • A profit factor (the money won divided by the money lost) of at least 1.2 from 2024 to 2025, after trading costs.
  • In 95% of 1,000 random shuffles of the trades, the account must not fall more than 30% from its highest point.

Improvements

Can the Keltner Channel Breakout be improved?

We tried three changes that traders often make. The fade trades the other way. It is for traders who use the channel for mean reversion: the idea that price tends to come back to its average after a big move. The squeeze filter waits for a quiet market first, as in John Carter's TTM Squeeze. The quicker exit, 'Exit back inside the channel', closes the trade at the first close back inside the channel. We tested each change on its own, on the same 30 backtests. To pick the best change, we looked only at 2020–2023, and picked the change with the most backtests that made money in those years. Then we checked it on 2024–2025.

Version Made money 2020–2023years used to pick the best version Made money 2024–2025years not used to pick it Made money in both periods Passed all four checks
Standard rules The rules under 'The rules we tested', with nothing added. 11 of 30middle test −17.6% 9 of 30middle test −11.6% 8 2
Fade the breakout Trade the other way: sell when a candle closes above the channel, and buy when a candle closes below the channel. The stop-loss is 2 × ATR(14) from the close of the breakout candle: above it for a sell, below it for a buy. The trade closes when price closes back across the middle line, which is now the profit target. 7 of 30middle test −21.8% 6 of 30middle test −15.0% 3 1
With a squeeze filterBest on 2020–2023 Only trade a breakout if there was a squeeze on at least one of the 5 candles before the breakout candle. In a squeeze, the Bollinger Bands are inside a narrower Keltner channel, with lines 1.5 × ATR(10) above and below the 20 EMA. The Bollinger Bands are 2 standard deviations above and below the 20-candle average of the close. This is the squeeze in John Carter's TTM Squeeze indicator. 14 of 30middle test −1.5% 9 of 30middle test −5.4% 4 2
Exit back inside the channel Close a buy when a candle closes at or below the upper line, back inside the channel. Close a sell when a candle closes at or above the lower line. This replaces waiting for a close across the middle line. The stop-loss stays the same. 5 of 30middle test −20.8% 7 of 30middle test −12.3% 3 1

None of the three changes made money in both periods in more backtests than the standard rules, which did it in 8 out of 30. On 2020–2023, the years we used to choose, the squeeze filter made money in the most backtests: 14 out of 30, compared with 11 with the standard rules. On 2024–2025, the years we kept for the check, the squeeze filter and the standard rules each made money in 9 out of 30. Over both periods, the squeeze filter made money in 4 backtests, compared with 8 with the standard rules. With the squeeze filter, the strategy made fewer trades: 16,894 instead of 41,522. Its 4-hour backtests made only 51 to 81 trades each, fewer than the 100 we need to judge a backtest. 3 of these 4-hour backtests made money in both periods, but they are not counted in the 4, because they have too few trades. On 1-hour charts, the squeeze filter did better: 6 out of 10 markets made money instead of 3, and 4 made money in both periods instead of 2. The fade and the quicker exit each made money in both periods in 3 backtests. The fade's 3 were all 4-hour charts of currency pairs. The quicker exit made 71,164 trades instead of 41,522. On 1-hour charts, its average winning trade made +0.66R, compared with +1.44R with the standard rules. So the quicker exit traded more often and made less on each winning trade.

Passed all four checks with a squeeze filter

  • Silver, 1-hour
  • Nasdaq 100, 1-hour

Questions

Questions traders ask

Does the Keltner channel breakout work?

On some markets, in our backtests. With the standard rules (no changes), 10 out of 30 backtests made money from 2020 to 2025, after trading costs. 8 of them made money in both 2020–2023 and 2024–2025. The Nasdaq 100 made money on all three charts we tested, and USD/JPY made +151.5% on 1-hour charts. The strategy lost money on EUR/USD, GBP/USD, AUD/USD and USD/CAD on every chart. On daily charts from 2014 to 2025, it made money only on gold and USD/JPY.

Should you trade a Keltner channel breakout or fade it?

In our backtests, it depended on the market. Fading means trading the other way: selling when a candle closes above the channel, and buying when a candle closes below it. On the Nasdaq 100, the S&P 500, the Dow Jones, USD/JPY, gold and silver, the breakout made money in 10 out of 18 backtests after trading costs. The fade made money in only 2 out of these 18. On EUR/USD, GBP/USD, AUD/USD and USD/CAD, the breakout lost money in all 12 backtests, and the fade made money in 4. These 4 were the 4-hour charts of AUD/USD (+36.4%), GBP/USD (+30.7%) and EUR/USD (+1.8%), and the 1-hour chart of USD/CAD (+16.4%). The USD/CAD backtest lost money in 2024–2025. On 15-minute charts, the fade lost money on all 10 markets.

Does the TTM Squeeze make the Keltner channel breakout better?

Yes, but only on 1-hour charts, in our backtests. We tested the squeeze from John Carter's TTM Squeeze: the Bollinger Bands inside a narrower Keltner channel, 1.5 ATRs above and below the middle line. This happens when price is quiet. We traded a breakout only if there was a squeeze on at least one of the 5 candles before the breakout candle. With this filter, 6 out of 10 markets made money on 1-hour charts, compared with 3 out of 10 without it. On 15-minute charts, only the Nasdaq 100 made money with it (+93.7%). On 4-hour charts, the filter left only 51 to 81 trades per backtest from 2020 to 2025, too few to judge.

What is the difference between Keltner Channels and Bollinger Bands?

Both draw lines above and below an average of the price. Keltner Channels use the ATR, how far price moves in an average candle, so the lines move smoothly. Bollinger Bands use the standard deviation of the last 20 closes, a measure of how spread out those closes are. So the Bollinger Bands spread apart quickly after a big move, and come close together when price is quiet. In our data, price closed outside each of them about as often. On 1-hour charts, 15.1% of candles closed outside the Keltner channel (2 ATRs), and 13.5% closed outside the Bollinger Bands (2 standard deviations). The Bollinger Bands were inside the narrower 1.5 ATR Keltner channel, a squeeze, on 15.5% of 1-hour candles.

Keep exploring

Related strategies we tested

ATR ATR trailing stop strategy backtest Made money in 13 of 30 tests, 7 of them in both periods. Best: Nasdaq 100, 15-minute, +238% Supertrend Supertrend (10, 3) strategy backtest Made money in 10 of 36 tests, 4 of them in both periods. Best: USD/JPY, 1-hour, +104% MACD MACD crossover strategy backtest Made money in 6 of 21 tests, 4 of them in both periods. Best: DAX, 1-hour, +294%
The strategy code we ran

This is the exact code from the BacktestingMax strategy engine. Paste it into the strategy editor in the app to run it on any market.

class KeltnerChannelBreakout(Strategy):
    """The Keltner channel breakout, with the default channel most charting platforms use,
    plus three changes traders often make.
    The channel (TradingView's default, and the engine's builtin:keltner):
      middle line = the 20-candle EMA of the close
      upper line  = middle line + 2 x ATR(10)
      lower line  = middle line - 2 x ATR(10)
    ATR is Wilder's average true range, as the built-in ATR. The EMA and the ATRs are worked
    out here, on the candles the strategy uses (see "Skipped candles"), instead of with the
    built-in indicators.
    Standard (all switches off):
      Buy at the next open after a breakout above the channel: a candle closes above the upper
      line, and the candle before it closed at or below its upper line. Sell short at the next
      open after a breakout below the channel (a close below the lower line, after a close at
      or above it).
      Exit at the next open after a candle closes back across the middle line: below it for a
      buy, above it for a sell. If that candle is also a breakout the other way, the opposite
      trade opens one candle later, once the old trade has closed, if the close is still
      beyond the line.
      Stop-loss: 2 x ATR(14) from the close of the breakout candle. It does not move.
      Size: 1% of the account lost at the stop-loss. When that is less than the smallest trade
      size the market allows, the signal is skipped. One trade at a time.
    Skipped candles: a candle that opens on a Friday at 17:00 New York time or later, or on a
    Saturday (some index and silver data has a few minutes of prices after the weekly close,
    which make a tiny extra candle), and a candle with no range (high = low, a market holiday).
    The strategy ignores them: they don't go into the EMA, the ATRs or the Bollinger Bands,
    and they give no breakout and no exit. On a market that trades at the weekend, remove the
    Friday and Saturday rule.
    The changes, one at a time:
      fade = 1         trade the other way: sell after a breakout above the channel, buy after
                       a breakout below it. The stop-loss is 2 x ATR(14) on the other side
                       (above the entry for a sell), and the exit is the same: a close back
                       across the middle line, which is now the target
      squeeze = 1      only trade a breakout that comes soon after a squeeze, as in John
                       Carter's TTM Squeeze: on at least one of the 5 candles before the
                       breakout candle, the Bollinger Bands (20-candle average of the close,
                       +/- 2 standard deviations of the last 20 closes) were inside a
                       narrower Keltner channel (the 20 EMA +/- 1.5 x ATR(10))
      exit_inside = 1  exit at the first close back inside the channel: at or below the upper
                       line for a buy, at or above the lower line for a sell, instead of
                       waiting for a close across the middle line
    """
    params = {"period": Param(type=int, default=20, min=2, max=200),
              "atr_period": Param(type=int, default=10, min=1, max=100),
              "mult": Param(type=float, default=2.0, min=0.1, max=10.0),
              "stop_atr_period": Param(type=int, default=14, min=1, max=100),
              "stop_mult": Param(type=float, default=2.0, min=0.1, max=10.0),
              "fade": Param(type=int, default=0, min=0, max=1),
              "squeeze": Param(type=int, default=0, min=0, max=1),
              "squeeze_bars": Param(type=int, default=5, min=1, max=50),
              "squeeze_mult": Param(type=float, default=1.5, min=0.1, max=10.0),
              "exit_inside": Param(type=int, default=0, min=0, max=1)}
    lookback = 5
    def init(self, ctx):
        self.ny = ctx.clock("America/New_York")
        self.prev_close = None    # close of the previous candle used
        self.ema = None
        self.ema_seed = []
        self.atr = None           # ATR(10), for the channel
        self.atr_seed = []
        self.stop_atr = None      # ATR(14), for the stop-loss
        self.stop_atr_seed = []
        self.closes = []          # the last closes used, for the Bollinger Bands
        self.squeezes = []        # squeeze on (True) or off, for the last candles used
        self.prev_upper = None    # the channel on the previous candle used
        self.prev_lower = None
        self.reverse_to = None    # "up" / "down": a breakout seen while the old trade closed
        self.from_above = None    # True when the open trade came from a breakout above the channel
    @staticmethod
    def wilder(value, seed, x, n):
        """Wilder's average, started with the simple average of the first n values."""
        if value is None:
            seed.append(x)
            return sum(seed) / n if len(seed) == n else None
        return value + (x - value) / n
    def on_bar(self, ctx):
        high, low, close = ctx.high[0], ctx.low[0], ctx.close[0]
        weekday, hour = self.ny.weekday, self.ny.hour
        if (weekday == 4 and hour >= 17) or weekday == 5 or high <= low:
            return  # a tiny candle after the Friday close, or a candle with no range
        # The indicators, on the candles used.
        if self.prev_close is None:
            true_range = high - low
        else:
            true_range = max(high - low, abs(high - self.prev_close), abs(low - self.prev_close))
        prev_close, self.prev_close = self.prev_close, close
        n = ctx.params.period
        if self.ema is None:
            self.ema_seed.append(close)
            if len(self.ema_seed) == n:
                self.ema = sum(self.ema_seed) / n
        else:
            self.ema += (close - self.ema) * 2.0 / (n + 1)
        self.atr = self.wilder(self.atr, self.atr_seed, true_range, ctx.params.atr_period)
        self.stop_atr = self.wilder(self.stop_atr, self.stop_atr_seed, true_range, ctx.params.stop_atr_period)
        self.closes.append(close)
        del self.closes[:-n]
        if self.ema is None or self.atr is None or self.stop_atr is None or len(self.closes) < n:
            return
        mid = self.ema
        upper = mid + ctx.params.mult * self.atr
        lower = mid - ctx.params.mult * self.atr
        average = sum(self.closes) / n
        deviation = (sum((x - average) ** 2 for x in self.closes) / n) ** 0.5
        narrow = ctx.params.squeeze_mult * self.atr
        self.squeezes.append(average + 2.0 * deviation < mid + narrow and average - 2.0 * deviation > mid - narrow)
        del self.squeezes[:-(ctx.params.squeeze_bars + 2)]
        prev_upper, prev_lower = self.prev_upper, self.prev_lower
        self.prev_upper, self.prev_lower = upper, lower
        if prev_upper is None or prev_close is None:
            return
        break_up = close > upper and prev_close <= prev_upper      # a breakout above the channel
        break_down = close < lower and prev_close >= prev_lower    # a breakout below the channel
        pos = ctx.position
        if pos is not None:
            if ctx.params.exit_inside == 1:
                done = close <= upper if pos.is_long else close >= lower
            elif self.from_above:
                done = close < mid     # the trade came from above the channel
            else:
                done = close > mid     # the trade came from below the channel
            if done:
                ctx.close_position()
                self.reverse_to = "up" if break_up else ("down" if break_down else None)
            return
        # A reversal waits one candle for the close to fill. Drop it if price came back inside.
        pending, self.reverse_to = self.reverse_to, None
        up = break_up or (pending == "up" and close > upper)
        down = break_down or (pending == "down" and close < lower)
        if not ctx.gate("keltner_breakout", up or down):
            return
        if ctx.params.squeeze == 1:
            # The 5 candles before the breakout candle (one more back for a delayed reversal).
            k = ctx.params.squeeze_bars
            late = 0 if (break_up or break_down) else 1
            window = self.squeezes[-(k + 1 + late):-(1 + late)]
            if not ctx.gate("after_a_squeeze", any(window)):
                return
        go_long = down if ctx.params.fade == 1 else up
        entry = close
        distance = ctx.params.stop_mult * self.stop_atr
        stop = entry - distance if go_long else entry + distance
        size = ctx.size_by_risk_pct(0.01, stop=stop, entry=entry)
        if size <= 0:
            return  # 1% of the account is less than the smallest trade size: skip the signal
        self.from_above = up
        if go_long:
            ctx.buy(size=size, sl=stop, tag="keltner_long")
        else:
            ctx.sell(size=size, sl=stop, tag="keltner_short")

Your turn

Test the Keltner Channel Breakout yourself

Change the settings, the market or the stop-loss and see the result in seconds. Ask the BacktestingMax AI: Backtest a Keltner channel breakout (20 EMA, 2 × ATR(10)) on the Nasdaq 100, 1-hour chart: buy a close above the upper line, sell short a close below the lower line, only after a TTM Squeeze, exit when price closes back across the middle line, stop-loss 2 × ATR(14), 1% risk per trade.

Try it free in BacktestingMax

Free account. No credit card needed.

See every strategy we tested Browse strategies shared by traders

Backtest on historical data with trading costs, run by BacktestingMax. Past results do not guarantee future results, and nothing here is financial advice. Updated 9 November 2026.