Breakouts, crosses and momentum: three trend followers

5 min · Oct 2026

A trend follower makes no attempt to buy the bottom. It waits until the price has already started moving, joins late, and stays until the move shows signs of ending. It gives up the first and last part of every trend in exchange for not having to predict one. A9's catalog has three classic versions. They differ in how they decide that a move has started.

Donchian breakout: new highs

The Donchian channel is the highest high and the lowest low of the last N bars. The idea, made famous by the turtle traders, is simple: a close above the highest point of recent weeks is a breakout, and a close below the lowest point of a shorter window ends the trade.

A9's version, with the defaults on hourly bars:

Buy when a bar closes above the highest high of the previous 20 bars.
Exit when a bar closes below the lowest low of the previous 10 bars.
The channels are measured before the current bar, since a channel that included the current bar could never be broken by it.
A new bot on the blank form is long only. A setting loaded from the shortlist may have shorting on, in which case a close below the 20-bar low opens a short on the perpetual.

The textbook 20 and 10 outperformed the settings a search tuned on the same data when both were tested on data the search had not seen, which is why they stay the default.

A Donchian breakout: a close above the 20-bar high is the entry, and the exit comes when the price closes under the rising 10-bar low

EMA cross: two averages

An exponential moving average (EMA) weights recent prices more heavily. A fast one follows the price closely; a slow one lags. When the fast average crosses above the slow one, the recent trend has turned up.

With the defaults on hourly bars:

Buy when the 10-bar EMA crosses above the 30-bar EMA.
Exit when it crosses back below.
Shorting is on by default: the cross down that closes a long also opens a short, and the next cross up flips back. A bot with shorting on trades the perpetual in your futures account.
Only a fresh cross opens a trade. A bot started in the middle of a trend waits for the next cross rather than buying one that happened days ago at a worse price.

A fast and a slow average: the fast one crossing above is the entry, and crossing back below is the exit

Momentum: the rate of change

Rate of change (ROC) is the percentage the price has moved over a lookback. It needs no average at all, only today's price against the price N bars ago.

With the defaults on hourly bars:

Buy when the 30-bar change rises above +5%, at the moment it crosses that line.
Exit when the change falls below 0%.
With shorting on, as it is by default, a fall through -5% opens a short, covered when the change rises back above 0%.

The gap between entering at 5% and leaving at 0% is deliberate. Without it, momentum hovering near one line would open and close a position every bar.

What the record looks like

All three share a temperament.

Many small losses. In a sideways market each signal starts a trade that soon reverses, and each costs a little plus fees.
A few large wins. When a real trend arrives, one trade can run for a long way and pay for many false starts.
A low win rate. Under half is common, and it can still be a good result if the winners are much larger than the losers.
Late entries and late exits. Each one buys after the move has started and sells after it has turned, so part of every trend is given up at both ends.

They all trade at market, on bar closes, so they pay the taker fee on every entry and exit.

On tokenized stocks

All three read only bars from the US regular session, 09:30 to 16:00 New York time, when they trade a stock token. Overnight and weekend prices on the token are thin and track a closed market, and they would create crossings and breakouts that the share itself never made. The bot still holds its position around the clock; it just does not take signals from those hours.

Most stock tokens were listed in mid-2026, so their hourly history covers a few months. That is enough to see how a strategy behaved in that stretch, and not enough to know how it behaves through a full cycle.

Choosing between them

Donchian is the slowest to act and the steadiest. It needs a genuinely new high.
EMA cross is smoother but reacts to any drift that turns the averages, and with shorting on it is always in the market one way or the other.
Momentum reacts to the size of a move, not its shape, and the 5% line keeps it out of quiet markets.

Backtest them on the same symbol and window, compare the drawdowns as well as the returns, and prefer the one whose losing stretches you could sit through.

Was this helpful?