MA Cross

The classic trend-following pattern. Long when a fast moving average crosses above a slow one; exit when it crosses back below. Two stop layers protect each trade.

class MACross(Strategy):
    """Trend following with stop loss and take profit."""
    fast_period = Param(7, min=2, max=50)
    slow_period = Param(21, min=10, max=200)
    trade = Symbol()
    sizing = FixedCapital()

    def execute(self):
        fast = SMA(self.fast_period, on=self.trade)
        slow = SMA(self.slow_period, on=self.trade)

        self.stop_loss(2.0)
        self.take_profit(5.0)

        if fast.crossed_above(slow) and self.trade.is_flat:
            self.buy(self.trade, reason="Bullish crossover")

        if fast.crossed_below(slow) and self.trade.is_long:
            self.sell(self.trade, self.trade.position, reason="Bearish crossover")

Walkthrough

Parameters

fast_period and slow_period define the two SMAs. Both have hard bounds — the optimizer will respect them. Defaults of 7 and 21 are common starting points for daily bars.

Indicators

Both SMAs are declared at the top of execute(). Even when the buy/sell branches don’t fire, the SMAs receive each bar’s close and stay in sync.

Exit rules

Two rules registered every bar:

• stop_loss(2.0) — exit if drawdown ≥ 2% from entry.
• take_profit(5.0) — exit if gain ≥ 5% from entry.

The first to trigger wins. Both are checked tick-by-tick, not just at bar close.

Entry

crossed_above returns True for the single bar where fast transitions from ≤ slow to > slow. The is_flat guard prevents the strategy from re-entering when already long (which would silently no-op anyway, but the explicit check makes intent clear).

Exit

Same pattern in reverse — crossed_below for one bar, exit only if currently long. The exit verb requires the explicit quantity (self.trade.position).

What to tune

Run an optimization sweep. Three dimensions to play with:

• fast_period — usually 3..15 for daily bars, 5..30 for intraday.
• slow_period — usually 10..50 for daily, 20..100 for intraday. Slow must be greater than fast.
• stop_loss(...) / take_profit(...) — currently hardcoded. Promote to Params for the optimizer:

sl_pct = Param(2.0, min=0.5, max=5.0)
tp_pct = Param(5.0, min=1.0, max=20.0)

Common modifications

Symmetric short side

class LongShortMACross(Strategy):
    """Golden-cross trend follower, both directions."""
    fast_period = Param(7, min=2, max=50)
    slow_period = Param(21, min=10, max=200)
    trade = Symbol()
    sizing = FixedCapital()

    def execute(self):
        fast = SMA(self.fast_period, on=self.trade)
        slow = SMA(self.slow_period, on=self.trade)

        if fast.crossed_above(slow) and self.trade.is_flat:
            self.buy(self.trade, reason="Bullish crossover")

        if fast.crossed_below(slow) and self.trade.is_flat:
            self.sell_short(self.trade, reason="Bearish crossover")

        if fast.crossed_above(slow) and self.trade.is_short:
            self.buy_to_cover(self.trade, self.trade.position, reason="Cover short")

        if fast.crossed_below(slow) and self.trade.is_long:
            self.sell(self.trade, self.trade.position, reason="Exit long")

Note: in a flat → long → flat → short pattern, you’ll need two bars (one to exit, one to enter the opposite side) — positions don’t flip automatically.

EMA instead of SMA

Swap SMA for EMA to react faster:

fast = EMA(self.fast_period, on=self.trade)
slow = EMA(self.slow_period, on=self.trade)

Whipsaw will increase. Pair with a longer-period trend filter (Aroon, for instance) to reduce false crossovers.