What it is
A trend pullback trade waits for a stock or index already in an uptrend to dip to a rising exponential moving-average, then buys the resumption. The 20-day EMA catches shallow pullbacks in strong trends; the 50-day catches deeper ones in slower trends. The average is not magic; it is a convenient, widely watched proxy for "the trend's fair value", and the stop under the pullback low keeps the risk defined.
It is probably the most common swing setup on the planet, which is both its strength (the crowd acts at the same level) and its weakness (the edge is thin and well-arbitraged).
The logic
In an uptrend, buyers who missed the last leg are waiting for a discount. A pullback to a moving average is where enough of them agree that the discount is real, and their buying plus short covering resumes the trend. Institutions that are accumulating tend to buy pullbacks rather than chase, so the MA often marks where their orders sit. The setup works because trends persist more often than chance would imply, at least on a multi-week horizon, and because the pullback offers a clearly wrong level to place a stop.
On the other side are traders selling the pullback expecting a reversal, and holders taking profit after the last leg. When the trend is intact, they sell you shares below the price the trend will soon reclaim.
Setup rules
- Market: liquid stocks and ETFs; index futures on daily charts; major forex pairs on 4-hour charts.
- Timeframe: daily chart for the trend and the setup; a 65-minute or 4-hour chart for the entry trigger.
- Trend conditions: price above a rising 50-day EMA; the 20-day EMA above the 50; the last swing high was a 3-month high; the market's index is also above its 50-day.
- Pullback conditions: price touches or closes within 1 percent of the 20-day EMA (or 50-day in a slower trend); the pullback has lasted 3 to 10 days; volume on the down days is below the volume on the prior up leg; no close below the 50-day EMA.
- Disqualifiers: an earnings date within the holding period; a pullback that started with a gap down on news; a stock that has already had three pullbacks to the average in the current trend (later pullbacks fail more often).
Entry, stop, target
Enter on the first daily close above the prior day's high after the touch, or intraday on a break of the prior day's high with a 65-minute close to confirm. Stop 1 ATR below the pullback low. Target 1 is the prior swing high; target 2 is the measured move of the prior leg projected from the pullback low.
| Item | Level | Notes |
|---|---|---|
| Prior swing high | 120.00 | 3-month high |
| 20-day EMA | 112.00 | Rising |
| Pullback low | 111.50 | Volume light |
| Entry | 114.00 | Close above prior day's high |
| Stop | 109.50 | 1 ATR below pullback low, risk 4.50 |
| Target 1 | 120.00 | Prior high, reward 6.00, 1.3R |
| Target 2 | 128.00 | Measured move, reward 14.00, 3.1R |
Sell half at target 1 and trail the remainder under the 20-day EMA on a closing basis. The trailing exit is what produces the occasional 5R outcome that makes the setup profitable; without it the average trade is about breakeven after costs.
Position sizing and risk
Risk 0.5 to 1 percent of equity per trade based on the stop at /tools/position-size. Because trend pullbacks are correlated across stocks in the same sector, cap total open risk in one sector at 3 percent and total portfolio-heat at 6 percent, as described in /learn/risk-management. Holding through earnings is a separate, larger risk that should not be taken by default.
What breaks it
- Trend end. The pullback that breaks the trend looks identical to the ones that did not, until the stop is hit. There is no filter for this; the stop is the filter.
- Bear markets. In a falling market almost every uptrend fails at the moving average. Require the index to be above its own 50-day; when it is not, sit out.
- Costs. Low relative to the stop; commissions and spread are under 2 percent of risk for liquid names. Overnight gaps through the stop are the real cost, and they are not in most backtests.
- Edge decay. The raw "buy the 20 EMA touch" signal has near-zero edge in modern liquid markets; the filters (volume contraction, index trend, first or second pullback only) carry the remaining edge, and it is modest.
- Drawdowns. Win rates of 40 to 50 percent with lumpy winners produce equity curves with long flat periods.
How to test it
Code the trend and pullback rules across a universe of at least 500 liquid stocks over 10 years, including delisted names to avoid survivorship-bias. Record every signal's R outcome under your exit rules with a 0.1 percent per-side cost and overnight gap risk modelled from actual data. Split results by market regime (index above or below its 200-day) and by pullback number (first, second, third in the trend). If the edge only exists in the first pullback in a rising market, that is the rule you trade. Minimum 500 signals; then paper-trade for 3 months to check that you take the signals as they come.
Variations
- 50-day version for slower stocks and for index ETFs, with a wider stop.
- Weekly 10-EMA version for multi-month holds; see also weekly-inside-bar-breakout.
- Anchored VWAP pullback using the anchored-vwap from the trend's start instead of an EMA.
Further reading
moving-average, pullback, trend, higher-highs-higher-lows, atr, portfolio-heat, correlation, survivorship-bias, swing-trading, golden-cross.
Related playbooks: three-day-pullback, abcd-measured-move, base-breakout, relative-strength-rotation