Machine Learning Mean Reversion Signal

Entry Logic

• Long Entry: A machine learning model (e.g., a neural network, support vector machine, or random forest) generates a "buy" signal.
• Short Entry: The model generates a "sell" signal.
• Confirmation: The model's confidence score for the signal is high.
• Timeframe: Varies, depending on the model.
• Location: Determined by the model.
• Market Condition: Determined by the model.

Exit Logic

• Profit Target: Determined by the model.
• Scaling Out: Determined by the model.
• Trailing Stop: Determined by the model.
• Signal Failure: The model generates an "exit" signal.
• Opposite Signal: The model generates an opposite signal.
• Time Expiration: Determined by the model.
• Momentum Loss: Determined by the model.

Stop Loss Structure

• Hard Stop: Determined by the model.
• Soft Stop: Determined by the model.
• Max Dollar Loss: Varies.
• Max Percent Loss: 2%.
• Structural Stop: Determined by the model.

Risk Management Framework

• Risk Per Trade: 1%.
• Daily Limit: Varies.
• Weekly Limit: 5%.
• Max Drawdown: 20%.
• R:R Requirement: Varies.

Position Sizing Model

• Sizing Approach: Based on the model's confidence score.
• Volatility Adjustment: The model should account for volatility.
• Conviction Sizing: Yes, based on the confidence score.
• Scaling In: Determined by the model.
• Scaling Out: Determined by the model.

Trade Filtering

• Market Conditions: The model is the filter.
• Setups: The model identifies the setups.
• Instruments: Any.
• Time Restrictions: Any.
• Chop/News Avoidance: The model should be trained on data that includes news events.

Context Framework

• Trend Direction: The model determines the context.
• VWAP Relationship: The model may use VWAP as a feature.
• MA Relationship: The model may use MAs as features.
• Range Location: The model may use range location as a feature.
• Higher TF Alignment: The model may use multi-timeframe analysis.

Trade Management Rules

• Breakeven: Determined by the model.
• Scale Out: Determined by the model.
• Add Size: Determined by the model.
• Fast vs Slow Moves: Determined by the model.

Time Rules

• Optimal Window: Determined by the model.
• Times to Avoid: Determined by the model.
• Session Notes: The ultimate black box strategy.

Setup Classification

• A+ Setup: High confidence signal.
• A Setup: Medium confidence signal.
• B Setup: Low confidence signal.
• C Setup: No signal.

Market Selection Criteria

• Instruments: Any.
• Volume: High.
• Volatility: Any.

Statistical Edge Metrics

• Win Rate: Varies.
• Avg Win: Varies.
• Avg Loss: Varies.
• Profit Factor: Varies.
• Expectancy: Varies.

Failure Conditions

• Market Conditions: The model is overfit to the training data and fails to adapt to new market conditions.
• Specific Scenarios: A change in market regime that is not represented in the training data.

Psychological Rules

• Discipline: Must have complete faith in the model and be able to execute its signals without question. Also requires the discipline to constantly monitor and retrain the model.

Advanced Components

• Regime Detection: The model can be designed to be a regime detection model itself.
• Filters: The model is the filter.
• Correlation: The model can use correlation as a feature.
• MTF Alignment: The model can use MTF data as input.

Location

• Strongest: In the market conditions that it was trained on.
• Weakest: In new and unseen market conditions.