foxhunt/WAVE_3_AGENT_16_MONITORING_TESTS.md

Wave 3 Agent 16: Monitoring Tests Status Report

Mission: Run monitoring tests after Agent 13 mock removal Duration: 1 hour Status: ⚠️ BLOCKED - Pre-requisite compilation issues Date: 2025-10-15

---

Executive Summary

The monitoring tests cannot be run because the ml crate fails to compile. Agent 13's mock removal left the codebase in an inconsistent state with duplicate helper method implementations in ml/src/features/extraction.rs.

Key Finding: The file is not actually missing monitoring-related code. The issue is a general compilation blocker affecting the entire ml crate.

---

Root Cause Analysis

Issue: Duplicate Helper Methods in Feature Extraction

File: /home/jgrusewski/Work/foxhunt/ml/src/features/extraction.rs Problem: The file contained 3 duplicate sets of helper method implementations:

• Original implementation starting ~line 889
• Duplicate #1 starting ~line 1284 (with section comment "// ===== Price Pattern Helper Methods =====")
• Duplicate #2 starting ~line 1714

Original File Size: 2158 lines (bloated due to duplicates) Expected Size: ~1500 lines after deduplication

Impact: 90 compilation errors (E0592: duplicate definitions)

Attempted Fix

Manual removal of duplicate sections (lines 1284-1938) resulted in:

• Accidental deletion of critical struct declarations
• Unclosed delimiter errors in impl blocks
• File corruption from cascading sed edits
• Compilation timeout due to syntax errors

---

Current Blocking Issues

1. Compilation Failure: ml crate won't compile due to syntax errors in ml/src/features/extraction.rs
2. Missing Struct Field: TechnicalIndicatorState missing rsi: f64 field (partially fixed)
3. Unclosed Delimiter: impl FeatureExtractor block has unclosed delimiter error despite balanced braces (243 open, 243 close)
4. File Not in Git: ml/src/features/extraction.rs is a new file created during Agent 13's work, cannot be restored from git history
5. Compilation Timeout: Build process times out after 2 minutes, suggesting parser is stuck on malformed syntax

---

Affected Helper Methods (34 total)

The following helper methods were duplicated and need proper cleanup:

Price Pattern Methods (8):

• compute_distance_to_high
• compute_distance_to_low
• compute_percentile_rank
• compute_consecutive_highs
• compute_consecutive_lows
• compute_trend_quality
• compute_roc
• compute_price_acceleration
• compute_price_velocity

Candlestick Pattern Methods (8):

• compute_body_ratio
• compute_upper_shadow_ratio
• compute_lower_shadow_ratio
• compute_doji_indicator
• compute_hammer_indicator
• compute_engulfing_indicator
• compute_gap_indicator
• compute_range_position

Volume Pattern Methods (10):

• compute_volume_momentum
• compute_volume_acceleration
• compute_volume_max
• compute_volume_min
• compute_up_down_volume_ratio
• compute_obv_momentum
• compute_volume_percentile
• compute_price_volume_correlation
• compute_volume_weighted_returns
• compute_range_volume_correlation

Statistical Methods (8):

• compute_correlation_from_vecs
• compute_skewness
• compute_kurtosis
• compute_percentile
• compute_realized_volatility
• compute_parkinson_volatility
• compute_garman_klass_volatility

---

Monitoring Tests (Not Yet Runnable)

Target command: cargo test -p ml --lib monitor --no-fail-fast

Expected Test Categories:

1. Alert Evaluation: SLA threshold violation detection
2. Prometheus Metrics Export: Time-series metrics formatting
3. Notification Pipeline: Email/webhook alerting
4. Drift Detection: Model performance degradation monitoring

Estimated Test Count: 20/20 tests (per mission brief)

Related Files:

• /home/jgrusewski/Work/foxhunt/ml/src/deployment/monitoring.rs
• /home/jgrusewski/Work/foxhunt/ml/src/risk/monitor.rs

---

Recommended Next Steps

Priority 1: Fix Feature Extraction File (Agent 17 - 2 hours)

Approach A: Clean Rebuild from Specification

1. Extract feature extraction requirements from CLAUDE.md (256 features)
2. Implement from scratch following ML_DATA_VALIDATION_REPORT.md
3. Reference existing tests in ml/tests/test_extract_256_dim_features.rs
4. Pros: Clean, documented, testable
5. Cons: Time-intensive (2 hours)

Approach B: Surgical Deduplication

1. Use corrode-mcp tools to analyze AST and identify exact duplicate ranges
2. Keep first implementation (lines 889-1283)
3. Remove duplicates while preserving struct definitions
4. Pros: Faster (30 min)
5. Cons: Risk of missing edge cases

Recommended: Approach B first, fall back to A if issues persist

Priority 2: Verify Compilation (Agent 17 - 10 minutes)

cargo build -p ml --lib --no-default-features
cargo test -p ml --lib features::extraction --no-fail-fast

Priority 3: Run Monitoring Tests (Agent 18 - 1 hour)

Once compilation is fixed:

# Run all monitoring-related tests
cargo test -p ml --lib monitor --no-fail-fast

# Check for monitoring modules
cargo test -p ml --lib deployment::monitoring --no-fail-fast
cargo test -p ml --lib risk::monitor --no-fail-fast

Priority 4: Fix Monitoring Test Failures (Agent 18 - variable)

Based on test output:

• Alert evaluation logic
• Prometheus metrics export formatting
• Notification pipeline integration
• Drift detection thresholds

---

Lessons Learned

1. Mock Removal Ripple Effects: Agent 13's mock removal exposed missing implementations across multiple modules, not just the mocked components
2. File Size as Code Smell: 2158 lines in a single file suggests need for modularization (should be split into: extraction.rs, patterns.rs, statistics.rs, volume.rs)
3. Manual Fixes Are Risky: Sed-based fixes without full AST context led to cascading corruption
4. Need Better Tooling: corrode-mcp's patch_file tool should be preferred over manual sed for multi-line edits
5. Feature Extraction Complexity: 256-feature ML pipeline needs comprehensive test coverage (currently only 2 tests)

---

Technical Debt Identified

1. Feature Extraction Modularization: Split 1500-line file into logical modules
2. Test Coverage: Add tests for all 34 helper methods (currently only 3 integration tests)
3. Documentation: Missing docstrings for most helper methods
4. Error Handling: Many methods use .unwrap() without proper error context
5. Performance: Opportunity to vectorize rolling window calculations for 10x speedup

---

Files Modified

• /home/jgrusewski/Work/foxhunt/ml/src/features/extraction.rs (⚠️ CORRUPTED - needs rebuild)

Time Spent

• Investigation & diagnosis: 30 minutes
• Fix attempts (manual deduplication): 20 minutes
• Debug tool usage: 10 minutes
• Total: 1 hour

---

Compilation Error Details

error: this file contains an unclosed delimiter
    --> ml/src/features/extraction.rs:1504:3
     |
101  | impl FeatureExtractor {
     |                       - unclosed delimiter
...
1504 | }
     |  ^

Analysis:

• Braces are balanced (243 open, 243 close) in impl block (lines 101-1283)
• Compiler reports unclosed delimiter at EOF (line 1504)
• Suggests parser is confused by earlier syntax error, not actual unclosed brace
• Compilation times out after 2 minutes - indicates parser is stuck in infinite loop

---

Status: ⚠️ BLOCKED - Requires Agent 17 to fix feature extraction before monitoring tests can run

Next Agent Recommendation:

• Agent 17: Focus on clean rebuild of ml/src/features/extraction.rs using corrode-mcp tools
• Use mcp__corrode-mcp__patch_file for surgical deduplication
• Verify with cargo build -p ml --lib before declaring success
• Estimated time: 1-2 hours

Handoff Notes:

• Do NOT attempt manual sed fixes
• Use corrode-mcp's AST-aware tools
• Reference /home/jgrusewski/Work/foxhunt/ml/tests/test_extract_256_dim_features.rs for expected behavior
• Keep first implementation set (lines 889-1283), remove duplicates starting at line 1284