Trading System Diagnostics

Purpose

Use this skill when:

• No trading signals are being generated

• Need to verify AI analysis is working

• Validating technical indicator calculations

• Debugging market data issues

Diagnostic Commands

Full Diagnostic (Default)

cd /home/linuxuser/nautilus_AlgVex source venv/bin/activate python3 scripts/diagnose.py

Quick Diagnostic (Skip AI calls)

cd /home/linuxuser/nautilus_AlgVex source venv/bin/activate python3 scripts/diagnose.py --quick

With Update and Restart

python3 scripts/diagnose.py --update --restart

Expected Output

Normal Operation Signs

✅ Configuration loaded successfully ✅ Market data fetched successfully ✅ TechnicalIndicatorManager initialized ✅ Technical data retrieved ✅ Sentiment data retrieved ✅ MultiAgent 层级决策成功 🐂 Bull Agent 分析中... 🐻 Bear Agent 分析中... ⚖️ Judge Agent 判断中... 🛡️ Risk Manager 评估中... 🎯 Judge 最终决策: BUY/SELL/HOLD

Key Checkpoints

Check Normal Value Abnormal Handling

RSI 0-100 Out of range = data error

MACD Any value NaN = insufficient data

Judge Signal BUY/SELL/HOLD ERROR = API failure

Winning Side BULL/BEAR/TIE 显示辩论胜方

信号决策流程 (层级决策架构)

采用 TradingAgents 层级决策架构，Judge 决策即最终决策

决策流程: Phase 1: Bull/Bear Debate (辩论) └→ 🐂 Bull Agent: 寻找做多理由 └→ 🐻 Bear Agent: 寻找做空理由

Phase 2: Judge (Portfolio Manager) Decision └→ ⚖️ 评估辩论结果，做出最终决策

Phase 3: Risk Evaluation └→ 🛡️ 确定仓位大小和止损止盈

注意: 所有策略参数配置在 configs/base.yaml 中管理。

Common Issues

1. No Trading Signals

Possible Causes:

• Judge returns HOLD (Bull/Bear辩论无明显胜者)

• Confidence below min_confidence_to_trade

• Risk Manager 认为风险过高

Check Command:

python3 scripts/diagnose_realtime.py 2>&1 | grep -E "(Judge|Final Signal|Confidence|Winning Side)"

2. DeepSeek API Failure

Check:

grep "DEEPSEEK_API_KEY" ~/.env.algvex

3. Abnormal Technical Indicators

Check:

python3 scripts/diagnose.py 2>&1 | grep -E "(RSI|MACD|SMA)"

Key Files

File Purpose

scripts/diagnose.py

Main diagnostic script

scripts/diagnose_realtime.py

Real-time API diagnostic

scripts/diagnostics/

Modular diagnostic system

scripts/smart_commit_analyzer.py

Regression detection (auto-evolving rules)

strategy/ai_strategy.py

Main strategy logic

configs/base.yaml

Base configuration (all parameters)

configs/production.yaml

Production environment overrides

Order Flow Simulation (v3.18)

模块化诊断系统包含 7 个订单流场景模拟。

7 Scenarios (模拟场景)

场景 描述 测试目标

1. New Position 开新仓 (无现有持仓) Bracket 订单创建

2. Add to Position 加仓 (同方向) SL/TP 数量更新

3. Reduce Position 减仓 SL/TP 数量减少

4. Reversal 反转仓位 (多→空/空→多) 两阶段提交逻辑

5. Close Position 平仓信号 仓位关闭 + 取消 SL/TP

6. Bracket Failure SL/TP 订单失败 CRITICAL 告警 (不回退)

7. SL/TP Modify Failure SL/TP 修改失败 WARNING 告警

Run Order Flow Simulation

cd /home/linuxuser/nautilus_AlgVex source venv/bin/activate python3 scripts/diagnose_realtime.py

诊断输出将包含 "Step 9: Order Flow Simulation (v3.18)" 显示所有 7 个场景的模拟结果。

Expected Output

============================================================ Step 9: Order Flow Simulation (v3.18)

✅ Scenario 1: New Position - PASSED ✅ Scenario 2: Add to Position - PASSED ✅ Scenario 3: Reduce Position - PASSED ✅ Scenario 4: Reversal - PASSED ✅ Scenario 5: Close Position - PASSED ✅ Scenario 6: Bracket Failure - PASSED ✅ Scenario 7: SL/TP Modify Failure - PASSED

📊 Order Flow Simulation Summary: 7/7 scenarios passed

回归检测 (修改代码后必须运行)

智能回归检测 (规则自动从 git 历史生成)

python3 scripts/smart_commit_analyzer.py

预期结果: ✅ 所有规则验证通过