July Update - Part 2

Last night I built a back test engine to test the strategies using real world entry and exit points, just like we do now. I went through every strategy and eliminated the ones that we of no value or were not generating enough trades to be useful. This morning I wrote a Meta Strategy focused on taking the best of each strategies to create a better entry and exit points. Here's an explanation of what it is:

Meta Strategy Summary

Overview

The Meta Strategy is an advanced trading system that combines all 7 professional trading strategies into a single, intelligent signal generator. It provides 24x more frequent buy signals than individual strategies while maintaining high quality through multi-strategy consensus.

Core Design Philosophy

Multi-Strategy Consensus

Instead of relying on a single trading methodology, the Meta Strategy combines:

  • Trend Following: Bravo9, 200 Cross, Golden Cross
  • Momentum/Reversal: RSI Divergence, 5/20 EMA Cross
  • Breakout/Support: Turtle Trading, Channel Trading

Swing Trading Optimization

  • Target Hold Period: 2-15 days (achieved: 4.98 days average)
  • Frequency: 130+ signals per year vs. 5.4 for RSI Divergence alone
  • Success Rate: 53% with fast profit-taking
  • Quality Control: Weighted scoring prevents weak signal noise

How It Works

1. Individual Strategy Analysis

Each of the 7 strategies analyzes the same stock data independently:

Stock Data → Bravo9 Strategy → Buy/Sell Signals
 → Turtle Trading → Buy/Sell Signals 
 → 200 Cross → Buy/Sell Signals
 → 5/20 EMA Cross → Buy/Sell Signals
 → RSI Divergence → Buy/Sell Signals
 → Channel Trading → Buy/Sell Signals
 → Golden Cross → Buy/Sell Signals

2. Weighted Signal Combination

Buy Signal Weights (Entry Points):

  • RSI Divergence: 30% (best reversal timing)
  • 5/20 EMA Cross: 20% (momentum confirmation)
  • Channel Trading: 15% (support/resistance)
  • Bravo9: 15% (trend following)
  • Turtle Trading: 10% (breakout confirmation)
  • 200 Cross: 5% (long-term trend)
  • Golden Cross: 5% (institutional confirmation)

Sell Signal Weights (Exit Points):

  • Golden Cross: 35% (institutional trend changes)
  • RSI Divergence: 25% (momentum exhaustion)
  • Channel Trading: 15% (resistance levels)
  • Turtle Trading: 10% (breakout failures)
  • Bravo9: 10% (trend breaks)
  • 5/20 EMA Cross: 5% (momentum loss)
  • 200 Cross: 0% (too slow for swing exits)

3. Score Calculation

For each trading day, the Meta Strategy calculates:

Buy Score = Σ(Strategy_Signal × Weight)

Example:
RSI Divergence: YES (1) × 0.30 = 0.30
5/20 EMA Cross: NO (0) × 0.20 = 0.00
Channel Trading: YES (1) × 0.15 = 0.15
Bravo9: NO (0) × 0.15 = 0.00
Turtle Trading: NO (0) × 0.10 = 0.00
200 Cross: YES (1) × 0.05 = 0.05
Golden Cross: NO (0) × 0.05 = 0.00
 
Total Buy Score: 0.50

Sell Score = Σ(Strategy_Signal × Weight) (Similar calculation with different weights)

4. Signal Generation Thresholds

WEAK Threshold (Default - Optimized for Frequency):

  • Buy Signal: Score ≥ 0.10 (generates ~130 signals/year)
  • Sell Signal: Score ≥ 0.10

MODERATE Threshold (Quality Focus):

  • Buy Signal: Score ≥ 0.20 (generates ~80 signals/year)
  • Sell Signal: Score ≥ 0.15

STRONG Threshold (High Conviction):

  • Buy Signal: Score ≥ 0.30 (generates ~40 signals/year)
  • Sell Signal: Score ≥ 0.25

5. Signal Strength Classification

Based on the weighted score:

  • Strong: Score ≥ 0.30 (multiple strategies agree)
  • Moderate: Score 0.20-0.29 (good consensus)
  • Weak: Score 0.10-0.19 (minimal consensus but above noise)

Quality Control Features

1. Signal Strength Scoring

  • Uses actual weighted scores (0.0-1.0) instead of binary signals
  • Higher scores indicate stronger multi-strategy consensus
  • Prevents random noise from single-strategy false signals

2. Volume Confirmation

  • Validates signals with trading volume analysis
  • Higher volume increases signal reliability
  • Filters out low-conviction moves

3. Momentum Analysis

  • 5-day price momentum confirmation
  • Ensures signals align with recent price action
  • Reduces counter-trend false signals

4. Trend Alignment

  • SMA slope analysis for buy signals
  • Moving average distance factors
  • Prevents signals at extreme price levels

Trading Application

Entry Process

  1. Signal Detection: Meta score reaches buy threshold
  2. Quality Assessment: Review Quality score (0.0-1.0)
  3. Position Sizing: Based on signal strength and quality
  4. Entry Timing: Execute at next market open

Exit Process

  1. Sell Signal: Meta score reaches sell threshold
  2. Time-based: Average 5-day holding period
  3. Risk Management: Stop losses based on signal quality
  4. Profit Taking: Quick exits maintain swing trading discipline

Position Sizing Recommendations

  • Quality > 0.60: Full position size
  • Quality 0.40-0.60: Reduce size by 25-50%
  • Quality < 0.40: Small positions or paper trading

Performance Characteristics

Frequency vs. Quality Balance

  • 130+ signals per year: Active trading opportunities
  • 53% success rate: Above-random performance
  • 4.98 days average hold: True swing trading timeframe
  • 0.33% average return per trade: Consistent profit capture

Risk Management

  • Fast exits prevent large losses
  • Multiple strategy confirmation reduces false signals
  • Quality scoring enables position sizing optimization
  • Swing timeframe limits overnight/weekend risk

Advantages Over Individual Strategies

1. Frequency

  • RSI Divergence alone: 5.4 signals/year
  • Meta Strategy: 130+ signals/year (24x improvement)

2. Reliability

  • Single strategy false signals filtered out
  • Multi-strategy consensus required
  • Weighted scoring prevents equal-weight noise

3. Adaptability

  • Works across different market conditions
  • Trend and counter-trend strategies balance each other
  • Configurable thresholds for different trading styles

4. Swing Trading Optimized

  • Eliminated long-holding strategies (BB Breakout: 66 days)
  • Eliminated low-frequency strategies (MACD: 4.6/year)
  • Focused on 2-42 day holding periods

Technical Implementation

Data Processing

  1. Fetch 2-year historical data for sufficient indicator periods
  2. Apply all 7 strategies simultaneously with parallel processing
  3. Calculate weighted scores for each trading day
  4. Generate binary signals based on thresholds
  5. Add quality metrics for position sizing guidance

Signal Output

  • buy_signal: Boolean (True/False)
  • sell_signal: Boolean (True/False)
  • signal_strength: Numeric (0.0-1.0)
  • meta_buy_score: Raw weighted buy score
  • meta_sell_score: Raw weighted sell score
  • signal_strength_desc: Human-readable description


Complete and Continue