The 20 EMA Pullback: A Simple Trend Strategy Defined Precisely

[Fairman]

The 20 EMA Pullback: A Simple Trend Strategy Defined Precisely

Many traders describe buying pullbacks to the 20-period exponential moving average. Writing it as exact rules shows what is and is not testable.

diagram.png (39.01 KiB) Viewed 24 times

Define the trend: for example, price above a rising 50-period average on the four-hour chart. Define the pullback: price retraces to touch or come within a set distance, say 5 pips, of the 20 EMA on the hourly chart. Define the trigger: a candle that closes back above the previous candle's high after touching the average. Define the stop: below the pullback low, with a small buffer. Define the target: the previous swing high, or a multiple of the risk, such as two times.

These details matter. Without them, "buy the pullback" becomes whatever the trader feels in the moment.

Observations from testing such systems are consistent. The strategy performs well in steady trends and poorly in choppy markets, where the average is crossed repeatedly. Filtering with a trend strength measure like ADX helps. The choice of 20 is arbitrary, and similar results appear with 18 or 25.

Costs matter, since stops are often 15 to 30 pips, and the risk-to-reward ratio after spread decides whether the system works.

Use this as a template: any strategy becomes testable when each element has a number.

Practical step: write your own pullback strategy with a number for each rule, and backtest fifty trades.

[Fairman]

Writing it with a number for every rule is the whole lesson. Once each part is a number, you can test it, and once you test it you find out most of the edge comes from the trend filter. Try running the same entry rules with the trend filter switched off and compare. On most pairs the results get much worse, because the pullback entry is basically a way of joining a trend that's already there. That tells you where to focus. Spend your effort on defining trend conditions well, and keep the entry simple. A perfect trigger in a bad trend still loses, while a plain trigger in a clean trend usually does fine.

[Fairman]

Try it with 18 and 25 as well as 20. If they all work about the same, good. If only 20 works, it's luck. I'd do the same with the trend filter period and the distance allowed from the average. Change one at a time and write the results in a small table. What you want to see is a broad area where results are similar, not one sharp peak. Also test across different years. A 20 EMA pullback system that only worked in a strong trending year like a big dollar rally is really a bet on that kind of year coming back, which is fine to know, but it's not the same as a reliable edge.