Strategy Patterns Playbook¶

Approved Patterns¶

These patterns have been validated and are approved for use. Follow the implementation guidelines exactly.

Pattern 1: ICT Quasimodo (QM) Reversal¶

Edge Source¶

Institutional order flow creates liquidity sweeps before major reversals.

Setup Sequence¶

1. Trend established (BoS)
2. Liquidity sweep (Sweep)
3. Structure shift confirmation (MSS)
4. Entry at Order Block retest

Implementation¶

// State machine
var int state = 0  // 0=looking, 1=BoS, 2=sweep, 3=MSS

// State transitions
if state == 0 and detectBoS()
    state := 1
else if state == 1 and detectSweep()
    state := 2
else if state == 2 and detectMSS()
    state := 3
    // Setup complete, wait for retest

// Entry on retest
if state == 3 and priceRetestsOB()
    enterLong()
    state := 0

Risk Parameters¶

Known Weaknesses¶

• Fails in strong trends without pullbacks
• Requires clear structure (choppy markets fail)
• Session-dependent (best in NY AM)

Pattern 2: Session Breakout with Confirmation¶

Edge Source¶

Asian range compression releases during London/NY open.

Setup Sequence¶

1. Define Asian range (20:00-00:00 UTC)
2. Wait for London/NY session
3. Breakout of Asian high/low
4. Confirmation candle close beyond range
5. Entry on pullback to range edge

Implementation¶

def session_breakout_signal(df, session='london'):
    asian_high = df[df['session'] == 'asia']['high'].max()
    asian_low = df[df['session'] == 'asia']['low'].min()

    current = df.iloc[-1]
    if current['session'] != session:
        return None

    # Bullish breakout
    if current['close'] > asian_high:
        return {
            'direction': 'long',
            'entry': asian_high,  # Pullback entry
            'stop': asian_low,
            'target': asian_high + (asian_high - asian_low)
        }
    # Bearish breakout
    elif current['close'] < asian_low:
        return {
            'direction': 'short',
            'entry': asian_low,
            'stop': asian_high,
            'target': asian_low - (asian_high - asian_low)
        }
    return None

Risk Parameters¶

Known Weaknesses¶

• False breakouts common on NFP/FOMC days
• Range too wide = poor R:R
• Range too narrow = noise

Pattern 3: Mean Reversion with Vol Filter¶

Edge Source¶

Overextended moves revert when volatility is contracting.

Setup Sequence¶

1. Price extends beyond 2 ATR from mean
2. Volatility is below average (contracting)
3. RSI/Stochastic shows divergence
4. Entry toward mean with tight stop

Implementation¶

def mean_reversion_signal(df, atr_mult=2.0, vol_percentile=50):
    current = df.iloc[-1]
    sma = df['close'].rolling(20).mean().iloc[-1]
    atr = df['atr'].iloc[-1]
    vol_pct = (df['atr'].iloc[-1] / df['atr'].rolling(90).mean().iloc[-1]) * 100

    # Only trade in low vol
    if vol_pct > vol_percentile:
        return None

    distance = (current['close'] - sma) / atr

    if distance > atr_mult:  # Overbought
        return {
            'direction': 'short',
            'entry': current['close'],
            'stop': current['close'] + atr,
            'target': sma
        }
    elif distance < -atr_mult:  # Oversold
        return {
            'direction': 'long',
            'entry': current['close'],
            'stop': current['close'] - atr,
            'target': sma
        }
    return None

Risk Parameters¶

Known Weaknesses¶

• Fails spectacularly in trending markets
• Requires strict vol filter
• Max 1 position at a time

Pattern 4: Momentum Continuation¶

Edge Source¶

Strong moves often continue after brief consolidation.

Setup Sequence¶

1. Strong move (>1.5 ATR candle)
2. 2-5 bars of consolidation (<0.5 ATR range)
3. Break of consolidation in original direction
4. Entry on break, stop below consolidation

Implementation¶

// Detect impulse
impulse = math.abs(close - open) > ta.atr(14) * 1.5
impulseDir = close > open ? 1 : -1

// Detect consolidation (2-5 bars, range < 0.5 ATR)
consol = ta.highest(high, 5) - ta.lowest(low, 5) < ta.atr(14) * 0.5

// Entry on break
longEntry = impulseDir == 1 and consol[1] and close > ta.highest(high[1], 5)
shortEntry = impulseDir == -1 and consol[1] and close < ta.lowest(low[1], 5)

Risk Parameters¶

Known Weaknesses¶

• Whipsaws in choppy markets
• Requires clear impulse
• Time-sensitive (stale setups fail)

Anti-Patterns (DO NOT USE)¶

Anti-Pattern 1: Indicator Stacking¶

Description: Combining 5+ indicators hoping for confluence.

Why It Fails: - Overfitting to historical data - Conflicting signals cause indecision - Each indicator adds lag

Instead: Use max 2-3 complementary indicators.

Anti-Pattern 2: Optimization to Perfection¶

Description: Parameter optimization until backtest Sharpe > 3.

Why It Fails: - Curve fitting to noise - OOS performance collapses - Parameters break on new data

Instead: Accept Sharpe 1.0-2.0 with robust parameters.

Anti-Pattern 3: Ignoring Costs¶

Description: Backtesting without spreads, commissions, slippage.

Why It Fails: - False profitability - High-frequency strategies destroyed by costs - Reality check at deployment

Instead: Include 1.5x expected costs in backtest.

Anti-Pattern 4: Single Regime Specialist¶

Description: Strategy only works in trending markets.

Why It Fails: - Markets change regimes - No warning when regime shifts - Catastrophic losses possible

Instead: Either filter for regime OR be robust across regimes.

Pattern Selection Guide¶

Pattern Combination Rules¶

1. Never run opposing patterns simultaneously on same instrument
2. Correlation check: Patterns on same instrument should have correlation < 0.5
3. Risk aggregation: Combined exposure across patterns ≤ 2x single pattern risk
4. Regime filter: If using regime-specific pattern, must have regime detector active