The most expensive habit in Moving AverageA moving average smooths out price data by calculating the average closing price over a set number of periods — it filters out noise and identifies the trend direction, with the stock above the MA interpreted as bullish and below as bearishRead definition → Trend FollowingTrend following is a systematic trading approach that identifies an established price trend and enters in that direction, holding the position as long as the trend persists and exiting when defined trend-reversal signals appear — it profitsRead definition → is treating every crossover as a trade. In a DayTradingToolkit simulation, a plain 9/21 EMA crossover took almost five trades on a choppy day and won about 15% of them. It needed 71% of sessions to TrendA trend is a sustained directional movement in price over time — an uptrend is characterized by a sequence of higher highs and higher lows, a downtrend by lower highs and lower lows — and correctly identifying the current trend direction isRead definition → just to break even, and no market gives you that.

The moving averages weren't the problem. The rule was. The same lines work far better as a filter that tells you when to trade, plus a location that tells you where.

This playbook gives you the EMA stack that defines a tradable trend, the PullbackA pullback is a temporary counter-trend price movement within an established trend — in an uptrend, a pullback is a short-term decline that offers a lower-risk entry point before the trend resumes, as opposed to a reversal which signals theRead definition → entry, the stop and the exit, and the numbers behind them.

Strategy snapshot

Trend continuation (moving average pullback)
Timeframe
5-min chart for the setup, 1-min for the trigger
Best time (ET)
10:00 to 11:30 a.m., then 2:00 to 3:30 p.m.
Typical hold
20 to 90 minutes
Difficulty
beginner
Works best when
9 EMA over 21 EMA over 50 EMA, all rising, with a light-volume pullback to the 21 EMA
Markets
Liquid US stocks, ETFs and index futures
Reward to risk2R first target, then a 21 EMA trail
Reward-to-risk is the planned target for this setup. Real results depend on the stock, the market and your execution.

In 30 seconds

  • Moving average trend following means using EMAs to confirm a trend and to mark where to buy pullbacks, instead of trading each crossover.
  • A tradable uptrend has the 9 EMA over the 21 EMA over the 50 EMA, with all three rising. Anything else stays on your watchlist.
  • The entry is one cent over the candle that tests the 21 EMA and turns up; the stop goes just under that candle.
  • A raw moving-average crossover gets chopped up on RangeA range (or trading range) is a price zone where a stock oscillates between a defined support floor and resistance ceiling without making meaningful new highs or lows — price is consolidating rather than trending, and the dominant trade is Read definition → days; buying pullbacks to a stacked set of averages in a confirmed trend filters out most of that chop, and neither works if you trade every session.

What is moving average trend following and why does it work?

Moving average trend following is a method that uses the slope and order of exponential moving averages to confirm a trend, then enters on pullbacks toward those averages. The averages summarize recent price. They don't predict anything, but they turn a messy chart into three easy questions.

Is the line rising? Is price above it? Are the lines in the right order?

Those three questions matter more than the moment two lines cross. A crossover is late by construction, because an average can only turn after price already has. By the time the 9 EMA crosses the 21, a good part of the move is gone. If you need a refresher on how EMAs differ from simple averages, start with our Lesson: moving averages.

Is there any evidence behind moving averages at all? A well-known study of the Dow from 1897 to 1986 found higher returns after moving-average buy signals than after sell signals, before trading costs (Brock, Lakonishok and LeBaron, 1992). A later survey found most modern studies report profits from technical rules, while warning that many suffer from data snooping and weak cost assumptions (Park and Irwin, 2007). So treat moving averages as a useful description of a trend. They aren't an edge by themselves.

Who's on the other side of a 21 EMA pullback? Mostly short-term traders taking profits and people betting the move is finished. On a real trend, they run out of shares while traders who missed the first move step in near a level everyone can see. That's the setup: a pause, at a visible average, inside a trend that's already proven itself.

What does the EMA stack look like?

The EMA stack is the order of the three lines: in an uptrend the 9 EMA sits on top, the 21 EMA in the middle and the 50 EMA at the bottom, and all three slope upward with space between them. In a downtrend the order flips. When the lines tangle together, there's no trend worth following.

Each line has one job. The 9 EMA shows the pace of the move, and it's where the fastest pullbacks stop. The 21 EMA is the main pullback level for a steady trend. The 50 EMA is the trend's floor: if price closes under it, the intraday trend read is finished.

Stack state
Full stack, rising
What you see
9 over 21 over 50, all sloped up, lines spreading
What you do
Trade pullbacks at normal size
Stack state
Forming
What you see
9 over 21, but the 50 is still flat
What you do
Half size, or wait
Stack state
Stretched
What you see
Price far above the 9 EMA, lines fanning wide
What you do
Don't chase; wait for a pullback
Stack state
Flattening
What you see
Lines converging, slope fading
What you do
Stop adding, tighten the exit
Stack state
Tangled
What you see
Lines cross each other repeatedly
What you do
No trades from this playbook

Entry one cent over the candle that tests the 21 EMA, stop six cents under its low, target at 2R.

When does it work, and when should you skip it?

Moving average trend following works on steady, one-way sessions with spacing between the lines, and it fails in chop, where the lines flatten and price keeps slicing through them. The stack does the filtering for you, if you let it.

Trade it when
All three EMAs are stacked and rising
Skip it when
Lines are flat or crossing
Trade it when
Price has stayed above the 21 EMA since the trend began
Skip it when
Price has closed under the 50 EMA
Trade it when
The stock has a catalyst or strong relative volume
Skip it when
Volume is thin and fading
Trade it when
The pullback is on lighter volume
Skip it when
Volume rises as price falls
Trade it when
It's 10:00 to 11:30 a.m. or after 2:00 p.m. ET
Skip it when
It's the first 15 minutes, when the 50 EMA hasn't settled, or the midday lull
Trade it when
The index is moving your way or flat
Skip it when
The index is falling hard against your trade

One timing note: EMAs need bars to mean anything. On a 5-minute chart the 50 EMA is still heavily shaped by the first candles of the day unless your platform includes pre-market or prior-day data. Check which your charts use, and don't compare them with someone else's until you know.

What are the exact rules?

The exact rules are: wait for a full, rising stack, let price pull back to the 21 EMA on lighter volume, buy one cent over the first candle that turns up there, stop under that candle's low, sell half at 2R, and trail the rest under the 21 EMA.

21 EMA stack pullback

10:00 to 11:30 a.m. ET, and 2:00 to 3:30 p.m. ET
Conditions
9 EMA over 21 EMA over 50 EMA, all rising; price has held the 21 EMA since the trend began; relative volume of 1.5x or more
Stock selection
Liquid US stocks over $5 with average daily volume above 1 million shares, or SPY and QQQ
Entry
Buy-stop one cent over the high of the first candle that touches or comes within a dime of the 21 EMA and closes green
Stop
A few cents under that candle's low
Target
Sell half at 2R or the prior swing high, whichever comes first; trail the rest and exit on a 5-minute close under the 21 EMA
Invalid if
A 5-minute close under the 50 EMA, or the 9 EMA crossing below the 21 EMA, before entry
Time stop
Exit if neither the stop nor 2R is hit within 45 minutes
Short version
Mirror every rule for a falling stack with the 9 EMA on the bottom

What does a real setup look like?

A real stack pullback looks like an orderly climb with one calm dip into the middle line. The ticker and prices below are made up so the rules are easy to see, and they match the chart above.

XYZ: pullback to the 21 EMA in a full stack

  1. The trend builds

    XYZ climbs steadily from $50.00 to about $51.60 by 10:45. The 9 EMA moves above the 21 EMA and both pull away from the 50 EMA.

  2. The pullback

    Price drifts lower on small candles while volume drops by about half. The stack stays in order and all three lines still rise.

  3. The test

    A candle dips to $51.14, a dime above the 21 EMA at $51.04, and closes green with a high of $51.31.

  4. Entry

    The buy-stop fills at $51.32. The stop is $51.08, so risk is $0.24 a share and 2R is $51.80.

  5. First target

    XYZ trades through $51.80. Sell half and move the stop on the rest to breakeven.

  6. Trail

    The trend carries on above $53. Hold the second half until a 5-minute candle closes under the 21 EMA.

Position size for the XYZ trade

Amount you're willing to lose$100
divided by Risk per share ($51.32 - $51.08)$0.24
equals Position size416 shares

How do you manage the trade?

You manage a stack pullback by letting the 21 EMA act as your trailing stop once the first partial is banked, and by getting smaller as the lines stretch or flatten. The line gives you a simple rule that takes the guesswork out of holding.

  1. Before 2R: leave the stop where it is. Price often wiggles around the 9 EMA right after entry.
  2. At 2R or the prior high: sell half and move the stop on the rest to breakeven.
  3. Trailing: exit the rest on a 5-minute close under the 21 EMA. Use a close, since wicks through the line are common.
  4. Faster trends: if price has stayed above the 9 EMA for the whole move, trail under the 9 instead. You'll exit sooner and keep more of a sharp run.
  5. Time stop: exit if nothing has happened in 45 minutes.
  6. After the exit: you can re-enter on a later pullback while the stack holds. Our guide to the trend continuation trade explains why you should size down each time.

Is this an A+ setup, a B, or a skip?

An A+ stack pullback has a full rising stack, clear spacing, a catalyst, and a quiet first test of the 21 EMA. Missing one makes it a B at half size. Missing two makes it a skip.

Criterion
Stack
A+
9, 21, 50 in order, all rising
B (half size)
50 EMA still flat
Skip
Lines crossing
Criterion
Spacing
A+
Clear space between lines
B (half size)
Lines close together
Skip
Tangled
Criterion
Test
A+
First touch of the 21 EMA
B (half size)
Second touch
Skip
Third touch or more
Criterion
Volume on the pullback
A+
Under half the push's
B (half size)
Somewhat lighter
Skip
Equal or heavier
Criterion
Catalyst
A+
News or earnings
B (half size)
Sector strength
Skip
None
Criterion
Time
A+
10:00 to 11:30
B (half size)
2:00 to 3:30
Skip
11:30 to 1:30

Where does moving average trend following fail?

Moving average trend following fails in sideways markets, where each cross or touch is a false signal, and at the end of a stretched trend, where a pullback to the line is the start of a reversal. Lag is the root cause in both cases.

Six crosses in one session. Each one is a new trade, and each one pays the spread and costs.

That chart shows a whipsaw day. A crossover trader would take six trades here and pay for all of them, while a stack trader would see tangled lines and sit out.

Trading the crossover itself

You're doing it if
You buy each time the 9 EMA crosses the 21, and most of your losses cluster on the same flat days.
The fix
Treat a cross as a heads-up that a trend may be forming. Wait for the full stack and the first pullback before you risk anything.

Using a stop that's too tight for the line

You're doing it if
You put the stop one cent under the 21 EMA, get tagged by a wick, and watch the trade work without you.
The fix
Place the stop under the test candle's low with a small buffer, then let the position size absorb the wider distance.

Buying the fourth test of a tired trend

You're doing it if
The lines are converging, each new high is smaller, and the pullbacks reach deeper toward the 50 EMA.
The fix
Count your tests. Normal size on the first, half on the second, and stand aside from the third on in a flattening stack.

Stops can also slip. A stop order becomes a market order when triggered, so the fill can be worse than your stop price in a fast market (Investor.gov: stop order).

What are the honest numbers?

The honest numbers show that the stack filter cuts losses in half on bad days, but neither rule survives trading every session. A DayTradingToolkit simulation (fixed seed 10, 1,500 sessions of each type, 78 five-minute bars, a 1R stop on the pullback rule, 0.05R cost per trade) ran both rules on simulated trending and choppy prices.

Rule
Every 9/21 crossover
Day type
Trend
Trades per day
2.9
Win rate
37.9%
Per trade
+0.11R
Per day
+0.32R
Rule
Every 9/21 crossover
Day type
Chop
Trades per day
4.8
Win rate
15.4%
Per trade
-0.17R
Per day
-0.80R
Rule
Stack pullback
Day type
Trend
Trades per day
1.1
Win rate
65.0%
Per trade
+0.28R
Per day
+0.29R
Rule
Stack pullback
Day type
Chop
Trades per day
1.0
Win rate
18.0%
Per trade
-0.25R
Per day
-0.26R

On trend days the two rules earned about the same. The whole gap is on chop days, where the crossover traded nearly five times and lost three times as much. The stack doesn't make you more money when things go right. It stops the bleeding when they don't.

From those numbers you can work out the share of sessions that must trend for each rule to break even.

The filter cuts the share of trend days the rule needs from 71% to 47%. It still needs you to sit out chop.

Even 47% is a high bar. At a mix of 30% trend days and 70% chop, the crossover lost 0.47R a day and the stack pullback lost 0.09R a day. So a stack filter on its own isn't enough: you also need to skip sessions that look like chop before the first trade, using the trend checks in our trend following guide. In this model, the stack rule fired about once a day on chop days too, because brief stacks do appear in noise. That's a good reason to demand a catalyst as well.

Real markets have news, gaps and fat tails that this simple model leaves out, so use it to compare the two rules. It isn't a forecast.

What are the main variations?

The main variations change the lines you watch or how you enter, and all of them keep the stack as the filter.

The 9 EMA pullback. In a fast momentum move, price may never reach the 21 EMA. Our 9 EMA pullback strategy covers that shallower entry.

The 20 EMA bounce. Some traders prefer a single 20 EMA. It behaves almost the same as the 21, and the 20 EMA bounce strategy treats it on its own.

SMA instead of EMA. A simple moving average reacts more slowly, so it gives fewer signals and later ones. Choose one type and stick with it, since testing endless settings is how you fool yourself.

The higher timeframe filter. Check the 15-minute stack before taking a 5-minute setup. You'll take fewer trades, and more of them will run with the bigger trend.

How does it compare with the VWAP pullback?

A 21 EMA pullback uses a line that follows recent price at any time of day, while a VWAP pullback uses the day's volume-weighted average, which gets slower as volume builds. They suit different hours and often agree.

Measures
21 EMA pullback
Recent price
First pullback to VWAP
Average price paid today
50/200 crossover
The slow trend over many bars
Best time
21 EMA pullback
Mid-morning, afternoon
First pullback to VWAP
First 90 minutes
50/200 crossover
Once the trend has run for a while
Reacts to
21 EMA pullback
The latest candles most
First pullback to VWAP
Heavy-volume prices most
50/200 crossover
Changes very slowly
Main use
21 EMA pullback
Trailing a steady trend
First pullback to VWAP
The first test after the open
50/200 crossover
Deciding which side to trade
Weakness
21 EMA pullback
Lag in chop
First pullback to VWAP
Goes flat late in the day
50/200 crossover
Signals come late

If you're trading the open, the first pullback to VWAP is usually the better fit. When the two lines sit at the same price, you have two separate reasons to pay attention. For the daily version of the crossover, see the 50/200 SMA crossover strategy.

Pre-trade checklist

0 of 9 done

Key takeaways

  • A crossover is a heads-up and never an entry. The lag and the chop eat it alive.
  • Trade only a full, rising stack, and only at a pullback to the 21 EMA.
  • Buy one cent over the test candle, stop under it, sell half at 2R, and trail the rest with a 5-minute close under the 21 EMA.
  • Sit out tangled charts. In the simulation, almost all the difference between the rules came from chop days.
  • Add a catalyst or volume filter, since the stack alone still fires too often in noise.

Frequently asked questions

What is the best moving average for day trading?
There's no single best setting, and the 9, 20 or 21, and 50 are simply common choices that are easy to read on a 5-minute chart. Changing a setting rarely fixes a losing strategy. Pick one set, keep it for at least 100 trades, and judge the rules before you touch the numbers.
Should I use EMA or SMA for trend following?
Most day traders use EMAs because they react faster to recent candles, which matters on intraday charts. SMAs are smoother and slower, so they suit longer holds and daily charts. The difference is small compared with the choice of whether to filter for trend days at all.
Do moving average crossovers work for day trading?
Raw crossovers work poorly on intraday charts because they arrive late and fire constantly in chop. In our simulation a 9/21 crossover won about 15% of trades on choppy days. Crossovers are more useful as an alert that a stack may be forming.
What timeframe is best for the 9 and 21 EMA?
The 5-minute chart is the usual choice for setups, with the 1-minute chart for timing entries. A 1-minute stack alone gives too many false signals, and a 15-minute stack gives too few trades for most day traders but works well as a filter.
Can moving averages act as support and resistance?
They can appear to, since many traders watch the same lines, but an average is a moving estimate and has no fixed price where orders must rest. Treat it as a zone. Wait for a candle to show that buyers are active there before you act.
Should moving averages include pre-market data?
It depends on your platform, and you should know which you're using. Including extended hours changes the early-morning values of slower lines like the 50 EMA. Neither choice is right or wrong, but your rules and your paper trades should use the same setting.
How many moving averages should I use on a chart?
Three is plenty, and many traders do well with two. More lines add clutter without adding information, because they're all calculated from the same prices. If you can't explain each line's job in one sentence, remove it.

More trend playbooks are in the Strategies Hub.

Disclaimer

This playbook is educational and is not financial advice. Moving averages lag price, they give false signals in choppy markets, and a stacked trend can reverse without warning. Stops can fill worse than planned in fast markets. The tickers, prices and simulation here are made up to explain the rules; they aren't a record of real trades or a forecast of results. Read the full disclaimer.

Article Sources

Research and regulator material cited in this playbook, with what each source supports.