| name | strategy-optimization |
| description | Trading strategy optimization methodology, an iterative process from diagnosis to hitting the target |
| schema | 1.0 |
| version | 1.0.0 |
| triggers | {"keywords":{"primary":["strategy optimization","optimize strategy","improve win rate","improve returns"],"secondary":["win rate","backtest","parameter tuning","performance improvement"],"context_boost":["trading","quant","return","profit"]},"context_penalty":["design","marketing","frontend"],"priority":"high"} |
| keywords | ["finance","trading","optimization","strategy","backtest"] |
| dependencies | {"domain-skills":["quant-trading"]} |
| author | claude-domain-skills |
| metadata | {"mcpmarket-version":"1.0.0"} |
Trading Strategy Optimization
A systematic strategy optimization process, from diagnosing the problem to reaching the goal
When to Use
- Win rate too low (< 40%)
- Returns below target
- Strategy performs poorly in specific market states
- Need to hit a specific return target (e.g. 15%+)
Core Process
┌─────────────────────────────────────────────────────────────────┐
│ │
│ Strategy Optimization Iteration Loop │
│ │
│ ┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────┐ │
│ │ ANALYZE │ → │ DIAGNOSE│ → │ RESEARCH│ → │IMPLEMENT│ │
│ │ │ │ │ │ │ │ │ │
│ └────┬────┘ └────┬────┘ └────┬────┘ └────┬────┘ │
│ │ │ │ │ │
│ ▼ ▼ ▼ ▼ │
│ gather metrics find root cause search solutions make changes │
│ │
│ ┌─────────────────────────────────────────┐ │
│ │ │ │
│ ▼ │ │
│ ┌─────────┐ ┌─────────┐ │ │
│ │VALIDATE │ → │ ITERATE │ ────────────────────┘ │
│ │ │ │ │ not met → back to ANALYZE │
│ └────┬────┘ └─────────┘ │
│ │ │
│ ▼ │
│ target met → document & deploy │
│ │
└─────────────────────────────────────────────────────────────────┘
Phase 1: ANALYZE
Gather current performance metrics:
| Metric | Description | Healthy Standard |
|---|
| Win rate | Proportion of winning trades | > 40% |
| Return | Total profit/loss | Positive |
| Drawdown | Maximum loss magnitude | < 20% |
| Trade count | Statistical significance | > 100 trades |
| Expectancy | Expected profit per trade | Positive |
Expectancy Formula
Expectancy = (WinRate × AvgWin) - ((1-WinRate) × AvgLoss)
Example:
Win rate: 54.5%, Avg win: $7.42, Avg loss: $8.26
Expectancy: (0.545 × 7.42) - (0.455 × 8.26) = +$0.29/trade
Positive expectancy × high trade count = steady profit
Phase 2: DIAGNOSE
Key: find the root cause, not the symptom
Common Root Causes
| Symptom | Possible Root Cause | Diagnostic Method |
|---|
| Low win rate | Strategy doesn't fit the market state | Check ADX / market regime |
| Low returns | Poor SL/TP ratio | Analyze profit/loss distribution |
| Large drawdown | Leverage too high | Compute risk exposure |
| Few trades | Conditions too strict | Loosen entry conditions |
Market State Analysis
┌─────────────────────────────────────────────────────────────────┐
│ Market State vs Strategy Fit │
│ │
│ ADX < 25 (ranging) → Grid Trading, Mean Reversion │
│ ADX > 25 (trending) → Trend Following, Momentum │
│ High volatility → Scalping, reduce position size │
│ Low volatility → Grid Trading, prepare for breakout │
│ │
│ ⚠️ 90% of the time the market is ranging; trend strategies │
│ will keep losing! │
└─────────────────────────────────────────────────────────────────┘
Diagnostic Example
Symptom: win rate only 25%
↓
Analysis: using an RSI+MACD trend strategy
↓
Check: average ADX < 25 (ranging 90% of the time)
↓
Root cause: trend strategy used in a ranging market
↓
Direction: need a ranging-market strategy (Grid Trading)
Phase 3: RESEARCH
Strategy-to-Market-State Mapping
| Market State | Recommended Strategy | Reason |
|---|
| Ranging (ADX < 25) | Grid Trading | Exploit price oscillation |
| Mean Reversion | Revert to the mean |
| Scalping | Quick in and out |
| Trending (ADX > 25) | Trend Following | Go with the trend |
| Momentum | Ride the momentum |
| Breakout | Follow the breakout |
| High volatility | Scalping | Fast profit |
| Reduce position | Control risk |
Sharp Edge: Validate Research Data
┌─────────────────────────────────────────────────────────────────┐
│ ⚠️ SE-1: Research data must be validated with real testing │
│ │
│ Severity: critical │
│ │
│ Case: │
│ - Research claimed chart patterns had 80-84% success rate │
│ - Actual backtest: made the strategy perform worse │
│ - Reason: stock market data doesn't apply to crypto │
│ │
│ Lesson: │
│ ❌ Implementing as soon as you see research data │
│ ✅ First validate by backtesting with real data │
└─────────────────────────────────────────────────────────────────┘
Phase 4: IMPLEMENT
Parameter Tuning Strategy
| Parameter | Tuning Direction | Impact |
|---|
| SL/TP ratio | Symmetric (1:1) vs asymmetric | Win rate vs risk/reward |
| Leverage | 1x → 10x | Return vs risk |
| Entry threshold | Loose vs strict | Trade count vs quality |
Key Findings
┌─────────────────────────────────────────────────────────────────┐
│ Optimization Findings │
│ │
│ 1. Symmetric SL/TP is better │
│ - 1.8%/1.8% yields ~55% win rate │
│ - Outperforms 1%/3% │
│ │
│ 2. Grid Trading in ranging markets │
│ - Win rate rose from 25% to 55% │
│ - Key: enable when ADX < 25 │
│ │
│ 3. More trades + positive expectancy │
│ - Small profits accumulate into a large return │
│ - 191 trades × $0.29/trade = +$55 │
└─────────────────────────────────────────────────────────────────┘
Phase 5: VALIDATE
Backtest Requirements
| Item | Minimum | Recommended |
|---|
| Backtest period | 60 days | 180 days |
| Trade count | 50 trades | 100+ trades |
| Out-of-sample test | Yes | Required |
| Trading costs | Included | Include slippage |
Validation Checklist
□ Did the win rate improve?
□ Did returns hit the target?
□ Is the drawdown acceptable?
□ Is expectancy positive?
□ How does it perform out of sample?
Phase 6: ITERATE
Target not met → back to ANALYZE
│
├── Analyze the new results
├── Adjust parameters or strategy
├── Re-validate
└── Repeat until target met
Best Practices
- Diagnose before acting - Don't blindly tune parameters
- Change one variable at a time - So you know what worked
- Validate research conclusions - Theoretical data can't be trusted
- Enough trades - For statistical significance
- Document successful experience - Write it up as an ADR
Common Mistakes
| Mistake | Consequence | Fix |
|---|
| ❌ Only tuning parameters, never changing strategy | Can't solve the underlying problem | Diagnose the root cause first |
| ❌ Backtest period too short | Overfitting | 180+ days |
| ❌ Ignoring market state | Strategy-market mismatch | Use regime detection |
| ❌ Chasing high win rate | Sacrifices risk/reward | Focus on expectancy |
Success Case
Before optimization:
- Return: 0.46%
- Win rate: 23.5%
- Strategy: RSI+MACD (trend strategy used in a ranging market)
Diagnosis:
- ADX < 25 90% of the time (ranging)
- Trend strategy is a poor fit
After optimization:
- Return: +19.93%
- Win rate: 54.5%
- Strategy: Grid Trading + Regime Detection
Key changes:
1. Added a Grid Trading strategy
2. Used regime detection to switch dynamically
3. Symmetric SL/TP (1.8%/1.8%)
4. 10x leverage + 191 trades
Risk Warning
⚠️ High returns come with high risk
- The case above had a maximum drawdown of 48.84%
- Consider lowering leverage (5x) for lower risk
- Continuously monitor changes in market state
- Set up a stop-loss mechanism
Reference Configuration
config = {
"leverage": 10.0,
"stop_loss_pct": 0.018,
"take_profit_pct": 0.018,
"top_n_signals": 3,
"min_confidence": 0.5,
"use_regime_selector": True,
}
Related Skills
finance/quant-trading - Quantitative trading fundamentalsfinance/investment-analysis - Investment analysismethodology/knowledge-acquisition-4c - Learning a new domain
