Trend

Mechanism

A trend EA confirms direction with a moving-average relationship, a channel or ATR-band break, or a momentum reading above a threshold, then enters with the move and manages the position until the trend ends. The exit is what varies most: a trailing stop or signal reversal holds winners open for as long as the move runs, while a fixed take-profit closes them at a set distance. Position size is fixed or scaled to ATR, and trade frequency is low — the five trend EAs published here run 2.9 to 7.1 trades a month.

Suitability

Best where moves persist for many bars: major FX on H1 and H4, index CFDs, and macro-driven metals or crypto. Hostile to range-bound conditions, where entries are triggered by moves that immediately fold back. Cost per trade matters far less than for scalping because holds are long, but overnight swap becomes a real expense on multi-day positions. Suitable for a trader who can leave positions alone for hours or days and sit through a losing streak without switching the EA off.

Notes

Trend-following is the oldest systematic edge that still works, and the one most often misdescribed. The mechanism is not a forecast: the EA waits until a direction is already established, joins it, and stays in while the condition that identified it persists. The money is made by holding, which is also why it is the hardest strategy for a human to run manually and the most natural one to hand to an expert advisor.

The description usually attached to it — trend systems lose most of their trades and make it back on a few big winners — is a claim about numbers. That means it can be checked rather than repeated. Checked against our own data, the trend EAs published on this site, it holds for one of the five. This page covers how a trend EA is built inside MetaTrader 5, what those records show about the shape of its results, and how to stress-test one before any capital is exposed.

How it works: an entry that follows, an exit that decides

The entry side is well-travelled and largely interchangeable. Three constructions cover most of what is sold:

• Moving-average relationship. A fast average crossing a slow one, or price holding above an average, used as a statement that direction is established. Cheap to compute, late by construction.
• Channel or ATR-band break. Entry when price closes beyond the highest high of n bars or beyond an ATR band around a mean. This scales the trigger to current volatility instead of a fixed pip distance.
• Momentum threshold. A directional oscillator past a level, usually combined with a higher-timeframe filter so the EA does not take a momentum reading inside a range.

The exit is where trend EAs actually differ, and it is the part buyers read last. Expectancy per trade is win rate × average win − (1 − win rate) × average loss. That is arithmetic, not a finding. The useful consequence is that the two terms stop being independent once an exit rule is fixed. An exit that trails a stop keeps winners open for as long as the move continues. That raises the average win and lengthens holding time on the winning side, and it lowers the win rate, because ordinary noise takes the trade out before the move develops. An exit at a fixed take-profit does the reverse. It closes winners at a set distance and lets losers run to the stop, which produces a high win rate and an average win smaller than the average loss.

Neither is wrong. Both are trend-following if the entry follows a trend. But they produce records that look like completely different strategies, which is why win rate on its own tells a buyer almost nothing about what they are buying.

What the published trend records show

Every figure in the table below is computed from the closed-trade lists published on each EA page — 1,498 trades in total, from records that begin in 2019 or 2021 and run to March 2026. The full runs, parameters and testing conditions are published under the site’s methodology.

EA	Symbol / TF	Trades	Win rate	Payoff (avg win ÷ avg loss)	Median hold, winners ÷ losers	Worst losing run
Kestrel Hover	USD/JPY M15	284	30.99%	2.95	5.00×	9
Tessera	USD/JPY H1	624	50.64%	1.25	1.10×	8
Orrery	JP225 H1	195	55.90%	1.01	1.45×	4
Zerqon	US30 H4	232	78.45%	0.42	0.84×	2
Peregrine	BTCUSD H4	163	78.53%	0.33	0.41×	3

Three things follow from it, and none of them is the folklore.

Win rate does not identify the strategy. The five span 30.99% to 78.53%. Their median, 55.90%, is not below the median of the mean-reversion EAs published here (54.50%). Screening a catalogue by win rate sorts EAs by their exit rule, not by what they trade.

The payoff ratio tracks how long winners are held relative to losers. Read the last three numeric columns together. The EA that holds winners five times longer than losers has a payoff ratio near 3.0; the one that holds winners less than half as long as losers has a payoff near 0.33. The ordering is monotonic in four of the five — Tessera and Orrery swap places, and they sit within 0.24 of each other. This is the exit rule showing up directly in the arithmetic of the record.

The smoothest-looking records are the most concentrated. The two highest win rates in the table are also the two whose profit depends on the fewest trades. Peregrine’s five largest wins amount to 106.1% of its net profit — remove them and the remaining 158 trades are net negative. Zerqon’s five largest are 39.8% of net. The EA that loses most often, Kestrel Hover, is less dependent on its best trades than either.

Market conditions: when a trend EA wins and when it breaks

The strategy’s bad environment is not a losing market — it is a directionless one. In a range, the entry condition is satisfied repeatedly by moves that fold back, and each attempt pays entry cost plus the stop. This is the mirror image of how mean reversion fails, which is the honest reason to consider the two alongside each other rather than as substitutes: they lose money in different weather.

The second condition worth stating plainly is holding cost. In the records here the median winning trade is held for between about 7 hours and about 8 days, and a position held over nights and weekends pays financing every night. A strategy whose average winner is modest and whose winners are held for a week can hand a meaningful share of its edge to swap without a single thing going wrong in the signal.

Parameters and settings in MT5

Input	Typical starting point	What it controls
MA_Fast / MA_Slow (or channel period)	Long enough that the signal is a trend, not a wiggle	How much movement must accumulate before the EA calls a direction established
ATR_Period / ATR_Mult	Volatility-scaled rather than a fixed pip distance	Stop distance and, in band constructions, the entry trigger itself
TrailStart / TrailStep	Start beyond the noise band, step in volatility units	The whole payoff-versus-win-rate trade-off lives here
TakeProfit	Zero if the exit is meant to trail	A non-zero cap converts the record to high win rate and payoff below 1
MaxSpreadPoints	Capped just above the normal spread for the symbol	Blocks entry when the spread is abnormal — less critical here than for scalping, but still the cheapest guard
OncePerBar	true for bar-close logic	Prevents the same signal firing repeatedly inside one bar and changing the strategy silently
Session / day filter	Off unless the symbol demands it	Index and metal symbols have hours where the “trend” is an illiquid drift

Three MT5-specific realities decide whether a promising backtest survives contact:

• Stops have a minimum distance. A trailing stop that tries to sit closer than the broker’s stop level is rejected, and the EA carries on as if it had been set. Read SymbolInfoInteger(symbol, SYMBOL_TRADE_STOPS_LEVEL) and clamp the trail to it, and treat SYMBOL_TRADE_FREEZE_LEVEL the same way for modifications near the current price.
• Swap is a strategy cost, not a footnote. SymbolInfoDouble(symbol, SYMBOL_SWAP_LONG) and SYMBOL_SWAP_SHORT are the values that apply to your account; triple-swap day differs by broker. A multi-day trend EA needs them in the test, not assumed away.
• Bar-close logic must be tested the way it runs. An EA that evaluates once per bar behaves differently under “every tick” than under “open prices only”. The difference is largest exactly where the trailing stop lives. Test on real tick history for the symbol you will trade, and read the modelling quality of the run before its profit.

Failure modes: how a trend EA loses money

• The market stops trending. Entries keep triggering on moves that fold back, and the account bleeds by stop-out rather than by any single disaster. This is the normal way a trend EA loses, and it can persist for months while the strategy remains intact.
• The trail is tighter than the noise. A trailing stop set inside the symbol’s ordinary retracement converts winners into scratches. The equity curve flattens and the profit factor falls without a single rule having failed — the exit simply removed the trades that pay for everything else.
• The strategy survives the losing streak; the trader often does not. Nine consecutive losses is an ordinary run at a 31% win rate. Switching the EA off during one keeps every loss and forfeits the recovery. Read the worst streak before the drawdown, not during it.
• A record that rests on a handful of trades. When the top five trades are the whole profit, the backtest is a statement about those five trades, not about the rule. Check what the curve looks like with the best trades removed.
• Parameters fitted to one regime. Averages and channel lengths optimised over a directional period will be the wrong length in the next one. Neighbouring parameter values producing wildly different results is the tell that the number was fitted rather than chosen.
• Cost that only appears live. Swap on multi-day holds, weekend gaps against an open position, a trailing stop the broker rejected for sitting too close. None of these announce themselves in a tester report configured optimistically.

How to build and stress-test a trend EA in mt5depot

You do not need to write MQL5 to build the machine described above: a directional filter, an entry on confirmation, a volatility-scaled stop and a trailing exit are blocks in the no-code builder. The reason to assemble it yourself is that the test that matters is not “does it make money”. It is the comparison between the two exit rules on the same entry, and no vendor page will run that for you.

Trend versus mean reversion

	Trend-following	Mean reversion
What it assumes	An established direction continues	A stretched price returns to its average
Entry relative to the move	With it, after confirmation	Against it, at an extreme
Loses when	The market ranges and moves fold back	The market trends and the stretch keeps stretching
Typical exit	Trailing stop or signal reversal	Fixed target at the mean
Worst case	A long series of ordinary losses	One position held against a move that does not return
Complexity	Beginner — few inputs, hard to sit through	Intermediate — the exit and the stop carry the risk

Neither is safer in general, and their failure modes are close to complementary: one needs movement, the other needs stillness. That is a genuine argument for running both, with one caveat that the table cannot show. A trend EA with a fixed take-profit and a mean-reversion EA are not as different as their labels suggest, because both then depend on price stopping where the rule says it should. Check what each one’s exit actually does before assuming that holding two EAs has diversified anything.