A crossover system replaces judgment with a single question: is the faster moving average above the slower one? When it moves above, the system buys; when it moves back below, it sells. Nothing in between needs to be decided. This simplicity makes crosses the most familiar mechanical trend system, and it also makes their weaknesses easy to see. This lesson uses the 10-day and 20-day exponential moving averages, a common pairing for swing trading; the 9/21 pairing behaves almost identically.
What a 10/20 EMA cross captures
An EMA weights recent closes more heavily, with weights that fade gradually (see moving averages). The 10-day EMA reflects roughly the last two weeks; the 20-day EMA, roughly the last month. When the 10-day rises above the 20-day, the most recent prices are, on average, higher than those of the somewhat longer window: short-term momentum has turned up. When it falls back below, momentum has turned down.
Two consequences follow from that definition:
- The cross always comes late. Both averages need several higher closes to turn, so the entry arrives after the swing low and the exit after the swing high. A crossover system gives up the first and last part of every move by design.
- It needs no forecast. The system does not try to pick the low. It waits until a move has begun and assumes that moves already in progress tend to continue for a while.
The 10/20 cross is a swing-trading tool. The better-known crossing of the 50-day and 200-day averages, often called the golden cross, works on the same principle but describes trends lasting months and is more useful as a market-regime gauge than as an entry signal.
The cost: whipsaws
In a sideways market, the two averages flatten and braid around each other. Each small rally produces a cross up, each small dip a cross down, and every round trip loses a little: the system buys after a rise and sells after a fall, inside a range that goes nowhere. This is called a whipsaw.
Whipsaws give crossover systems their characteristic profile: a win rate that is often below half, many small losses, and a minority of large winners from sustained trends that pay for everything else. That profile is psychologically demanding. A trader who abandons the system after six small losses in a row will often miss the one trend that would have made the year.
Filters that reduce whipsaws
A filter is an extra condition that must hold for a cross to count. Common choices:
- Long-term trend. Take long crosses only when the close is above the 200-day average. This avoids buying short rallies inside a larger decline.
- Slope of the slow average. Require the 20-day or 50-day average to be rising, so the cross aligns with the intermediate trend.
- Confirmation. Require the fast average to stay above the slow one for a second close, or to clear it by a minimum margin, such as a fraction of ATR.
- Trend strength. Wilder's ADX measures how directional a market is regardless of direction; some traders ignore crosses while it is low.
- Market regime. Take crosses in individual stocks only while the index itself is in an uptrend.
Every filter removes some losing trades and some winning ones, and adds delay. The test is whether, across many trades, the losers removed outweigh the winners lost. Adding filters until a backtest looks perfect is a form of overfitting, discussed in lesson 7.
Crosses and pullbacks together
Within an uptrend, a pullback often pushes the 10-day EMA briefly below the 20-day. The cross back up then marks the end of the pullback, so a crossover system naturally re-enters a trend after each meaningful dip. Compared with the pullback setup in lesson 2, it enters later and without a defined pullback low, but it needs no judgment about where the pullback should stop. Some traders use the two together: the pullback rules for the entry, the opposite cross as the exit that lets the trend run.
How to apply it
- Entry: the 10-day EMA closes above the 20-day EMA after being at or below it the session before, and the close is above the 200-day SMA. Buy at that close.
- Stop: the higher of the entry day's low and the entry minus 2 × ATR(14), but never more than 10% below the entry. Set it on the entry day and do not lower it.
- Exit: the 10-day EMA crosses back below the 20-day EMA; sell at that close. There is no fixed target: the opposite cross lets winners run.
- Invalidation: skip crosses when the 200-day average is falling, when the stock has just gapped far above both averages, or when earnings fall within the next few sessions.
An illustrative example: the cross happens on a day that closes at $80.00 with a low of $78.60, and ATR(14) is $1.50. The ATR-based level is $80.00 − $3.00 = $77.00; the entry day's low, $78.60, is higher, so it becomes the stop and 1R is $1.40. The 10% cap, $72.00, does not bind. Had the entry day been a wide bar with a low of $74.50, the ATR level of $77.00 would have been the stop instead.
Note the consequence: the stop is often close to the entry, while the exit rule waits for the opposite cross. Many trades will therefore end at the stop on ordinary noise before the averages turn, and the system relies on the trades that survive to run far. Position size should be calculated from this stop, not from the distance to the opposite cross.
In Clear Tape
The preset 10/20 EMA Cross in trading systems applies the rules above exactly: the 10-day EMA above the 20-day EMA at today's close, at or below it the session before; the close above the 200-day SMA; entry at that close; stop at the higher of the entry day's low and entry − 2 × ATR(14), capped at 10% below entry; exit when the 10-day EMA crosses back below the 20-day, at that close. The system has no target of its own; the 2R level shown with a signal is for reference only and does not close the trade. EMAs are seeded with the simple average of their first values, matching TradingView, so the levels agree with the chart. The rule page notes when it struggles: sideways markets, where small losses accumulate, and sharp V-shaped turns, where the signal comes late.
In the system builder, the operators "crosses above" and "crosses below" let you write variants, such as a 9/21 pair or an added condition that the 50-day SMA be above the 200-day SMA, and compare them with the preset in a backtest. Screening and backtests are available to Tools and Pro members.
Common mistakes
- Judging the system by its last few trades. A run of whipsaw losses is the normal cost of a crossover system, not proof that it has stopped working.
- Trading crosses without a trend filter in a falling market, where every short rally produces a buy signal.
- Overriding the exit by selling early on the first red day. Cutting winners short removes the large trades that the profile depends on.
- Searching for the perfect pair. A 10/20 pair, a 9/21 pair and an 8/21 pair behave much the same; picking whichever tested best on the past is overfitting.
- Sizing from the wrong distance. Risk per trade is set by the stop, not by where the opposite cross might occur.
Summary
A 10/20 EMA cross turns a shift in short-term momentum into a mechanical entry and exit. It always enters after the low and exits after the high, makes its money in sustained trends and loses in small steps in sideways markets. A 200-day filter, slope or confirmation rules reduce whipsaws at the cost of delay. The next lesson turns from following trends to fading short-term extremes: short-term mean reversion.
For educational purposes only. Not investment advice.