Migration Tracker

name: migration-tracker description: Context for ongoing migration from old Portfolio Buddy app. Use when: fixing bugs, adding migrated features, checking feature parity, or understanding why certain code exists. Contains list of 40 features being migrated and known issues.

Portfolio Buddy 2 - Migration Tracker

Migration Status: 40 Features

✅ Completed (36/40 - 90%)

Core features migrated and working:

• CSV upload and parsing with PapaParse

• Supabase storage integration

• Basic metrics calculation (Sharpe, Max DD, CAGR, Win Rate, etc.)

• Sortino Ratio (completed commits 258ba3a, 9f25040)

• Risk-free rate input (completed commit 258ba3a)

• Asset correlation matrix (Spearman & Pearson)

• Portfolio comparison charts (Chart.js)

• Equity curve visualization

• Responsive UI with Tailwind CSS

• shadcn/ui color system integration

• Date range filtering (completed commit 258ba3a)

• Contract multipliers for futures (useContractMultipliers hook)

• Advanced multi-column sorting (useSorting hook)

• Error handling and validation

• File upload progress tracking

• Multiple file management

• Database Integration (Nov 16, 2025) - PRODUCTION READY ✓

  • Python script uploads trades automatically ✓
  • New database schema (portfolios, strategies, trades) ✓
  • Frontend database fetch ✓ (commits c4fa57c through ee7cec8)
  • Dual CSV/Database support ✓
  • Format auto-detection (1-row vs 2-row) ✓
  • User tested and verified ✓
  • Merged to main and deployed ✓
  • See: dev-docs/supabase-migration-plan.md

🚧 In Progress (3/40)

1. Advanced filtering - Partial implementation
   • Date filtering complete ✓
   • Symbol filtering needed
   • Strategy filtering needed
2. Export functionality - CSV export only
   • Excel export pending
   • PDF reports pending
3. Historical comparison - Backend ready, UI pending
   • Need UI for comparing multiple time periods

❌ Not Started (2/40)

1. Multi-period analysis - Complex, low priority
   • Compare performance across different time windows
   • Requires significant UI work
2. Risk scenario modeling - Requires new backend logic
   • Monte Carlo simulations
   • Stress testing

Recent Completed Features

Database Fetch Implementation (Nov 16, 2025) ✅ COMPLETED

Status: Production ready - Tested and deployed Commits: c4fa57c, a5ce0ec, 676de06, eba4c8d, ee7cec8, ae9202d Merged to main: d56497a (PR #1)

Implementation Journey (6 commits):

1. Initial Implementation (c4fa57c)

   • Added calculateMetricsFromDatabase() and buildFilenameFromMetadata() to dataUtils.ts
   • Rewrote fetchFromSupabase() in App.tsx
   • Changed from old csv_files table to new strategies + trades schema
   • Added TypeScript interfaces: DatabaseTrade, StrategyMetadata

2. Fix Query Syntax (a5ce0ec)

   • Fixed Supabase order clause syntax error
   • Changed order('trades.trade_date') to order('trade_date') with foreignTable parameter
   • Error: "failed to parse order (trades.trade_date.asc)"

3. Fix Trade Count Limit (676de06)

   • Discovered Supabase embedded resource limit (~60 rows)
   • Separated queries: fetch strategies first, then fetch trades separately
   • Added explicit .limit(10000) to get all trades
   • Fixed: 59 trades → 119 trades ✅

4. Fix TypeScript Build Errors (eba4c8d)

   • Added StrategyFromDB interface with optional trades? property
   • Added DatabaseTrade interface to App.tsx
   • Fixed: "Property 'trades' does not exist" errors

5. Fix Metrics Calculation (ee7cec8) ⭐ CRITICAL FIX

   • Auto-detect format: 2-row (Entry/Exit) vs 1-row (database)
   • Modified calculateMetrics() to check for "Entry/Exit" column
   • If present → loop by 2 (old CSV format)
   • If absent → loop by 1 (new database format)
   • Fixed: Metrics now calculated correctly for all 119 trades ✅

6. Update Documentation (ae9202d)

   • Updated migration-tracker skill with implementation details
   • Documented all changes and line numbers

Final Results:

• ✅ 119 trades loaded from database (not 59)
• ✅ All metrics calculated correctly (win rate, profit factor, etc.)
• ✅ CSV upload backward compatibility preserved
• ✅ Dual-mode support: both CSV and database work simultaneously
• ✅ Format auto-detection works seamlessly
• ✅ User tested and verified working
• ✅ Deployed to production

Files Modified:

• src/utils/dataUtils.ts: +235 lines (functions, interfaces, auto-detection)
• src/App.tsx: +145 lines (database fetch, TypeScript types)
• .claude/skills/migration-tracker/SKILL.md: Documentation updates

How It Works:

1. User clicks "Load Data" button
2. App fetches strategies from Supabase
3. For each strategy, fetches ALL trades separately (no 60-row limit)
4. Builds filename from metadata (e.g., SI_Long_Test_TestStrategy1.csv)
5. Transforms to cleanedData format with 3 columns (no Entry/Exit column)
6. calculateMetrics() auto-detects format and processes correctly
7. Pre-populates contract multipliers from database
8. Auto-selects strategies and displays metrics/charts

Backward Compatibility:

• ✅ CSV upload with 4 columns (includes Entry/Exit) → 2-row processing
• ✅ Database with 3 columns (no Entry/Exit) → 1-row processing
• ✅ Both formats work simultaneously
• ✅ All existing components, hooks, charts unchanged

Database Integration Planning (Nov 16, 2025)

Status: Planning complete, ready for implementation What Changed:

• Created comprehensive migration plan (dev-docs/supabase-migration-plan.md)
• Analyzed new Supabase database schema (portfolios, strategies, trades tables)
• Designed dual-mode support (CSV upload + database fetch)
• Planned data transformation strategy (single-row trades vs entry/exit pairs)

New Database Schema:

• portfolios: Portfolio definitions with is_master flag
• strategies: Strategy metadata (market, direction, contract_multiplier, etc.)
• trades: Individual trade records (trade_date, trade_time, profit)
• portfolio_strategies: Links portfolios to strategies

Implementation Plan:

1. Add calculateMetricsFromDatabase() function in dataUtils.ts (~80 lines)
2. Update fetchFromSupabase() query in App.tsx (~60 lines changed)
3. Transform database data to match cleanedData format
4. Pre-populate contract multipliers from database
5. Test with 119 existing trades

Current State:

• Python script on Windows VPS uploads trades automatically ✓
• Database contains 1 strategy with 119 trades ✓
• Frontend still queries old csv_files table (needs update)

Next Steps:

• Implement Phase 1: New calculation function
• Implement Phase 2: Update Supabase query
• Test dual CSV/Database support
• Deploy to production

Strategy Delete Feature (Nov 19, 2025) ✅

Commit: c372ab7a92d267eda3e540b298872484ef09e38d Files: App.tsx (+47), MetricsTable.tsx (+18), PortfolioSection.tsx (+8)

What it does:

• Delete database strategies permanently (red trash icon with confirmation)
• Remove CSV strategies from view (gray trash icon, immediate)
• strategyIdMap tracks DB vs CSV (App.tsx line 67)
• handleDeleteStrategy with Supabase deletion (App.tsx lines 423-458)
• Trash2 icon in Actions column (MetricsTable.tsx lines 1, 239-246)

Git Forensic Recovery (Dec 2, 2025) ✅

Problem: Delete feature was "lost" (local repo behind origin/main)

Solution:

git fetch origin
git merge origin/main  # Fast-forward to c372ab7

Key lesson: Always check git log origin/main when work seems missing

Sortino Ratio (Oct 2025)

Commits: 258ba3a, 9f25040 What Changed:

• Added risk-free rate input field in PortfolioSection (line 131: useState<number>(0))
• Implemented inline Sortino calculation in PortfolioSection (lines 133-158)
• Fixed downside deviation calculation (now properly annualized using sqrt(365))
• Corrected variance calculation (divides by total returns, not just negative returns)
• Displays in portfolio stats section (line 535)

Files Modified:

• PortfolioSection.tsx: Added riskFreeRate state, downside deviation calculation, and display

Implementation Details:

• NOT in dataUtils.ts - Sortino is calculated inline in PortfolioSection using useMemo
• NOT in MetricsTable - Only displayed in portfolio stats area
• Kept in component due to portfolio-level context requirements:
  • Needs user input (risk-free rate)
  • Operates on portfolio daily returns (not trade-level metrics)
  • Different calculation scope than win rate, profit factor, etc.

Date Range Filtering (Oct 2025)

Commit: 258ba3a What Changed:

• usePortfolio hook now accepts date range params
• Filters trades by start/end date
• Recalculates metrics for filtered period only
• UI controls in PortfolioSection

Files Modified:

• usePortfolio.ts: Added date filtering logic
• PortfolioSection.tsx: Added date picker controls

Enhanced Error Handling (Sept 2025)

Commit: 9fb7fdb What Changed:

• Better Supabase error messages
• Client validation before upload
• Error list component shows all errors
• Toast notifications for user feedback

Files Modified:

• UploadSection.tsx: Enhanced error handling
• ErrorList.tsx: New component for error display
• usePortfolio.ts: Better error propagation

Current Tech Debt

High Priority

1. PortfolioSection.tsx is 591 lines (3x the 200-line limit)

   • Needs refactoring into:
     • EquityChartSection.tsx
     • PortfolioStats.tsx
     • ContractControls.tsx
   • Estimated effort: 4-6 hours

2. Remove unused Recharts dependency (11.5KB waste)

   • Currently using Chart.js
   • Recharts never imported anywhere
   • Run: npm uninstall recharts

3. Fix 15 TypeScript any violations

   • usePortfolio.ts: 11 instances
   • useMetrics.ts: 4 instances
   • dataUtils.ts: 1 instance
   • Need proper interfaces for Trade and Metric types

Medium Priority

4. App.tsx is 351 lines (175% of limit)

   • Extract sections into components
   • Estimated effort: 2-3 hours

5. MetricsTable.tsx is 242 lines (121% of limit)

   • Improved from 350 lines
   • Still over limit, could extract more

6. No error boundaries implemented

   • Should wrap risky components
   • Prevents full app crashes

Low Priority

7. No testing setup

   • Should test critical calculations
   • Vitest recommended for Vite projects

8. No CI/CD pipeline

   • Manual Cloudflare deployments
   • Could automate with GitHub Actions

Known Issues

Issue 1: Supabase 500 Errors

Status: Partially fixed (commit 9fb7fdb) Root cause: Row limit exceeded on free tier Current workaround:

• Enhanced error handling shows user-friendly messages
• Batch uploads in smaller chunks recommended Long-term fix: Implement data aggregation before storage

Issue 2: Large Component Files

Status: Documented but not fixed Problem: Components grew during migration Affected:

• PortfolioSection.tsx: 591 lines (was 280, now worse!)
• App.tsx: 351 lines
• MetricsTable.tsx: 242 lines (improved from 350) Fix needed: Systematic refactoring into smaller components

Issue 3: No Select All Button

Status: Feature doesn't exist Note: Previous skill version documented a "Select All bug" but this feature was either removed or never implemented. MetricsTable has individual selection but no "Select All" functionality.

Issue 4: Lost/Missing Code

Status: Recoverable via git forensics Problem: Can't find recently added features or code seems to have disappeared Solution:

1. Check remote: git log origin/main --oneline -10
2. Search commits: git log --all --grep="keyword"
3. Fast-forward if behind: git merge origin/main
4. Use reflog to see recent HEAD positions: git reflog Example: On Dec 2, 2025, the delete feature was recovered by fast-forwarding from f4e752a to c372ab7

Migration Priorities (ICE Scored)

Top Priorities by ICE Score:

1. Error boundaries (50.4) - High impact, easy to implement
2. Export to Excel (44.8) - User-requested feature
3. Remove Recharts (30) - Quick win, technical cleanup
4. Fix TypeScript violations (27) - Code quality
5. Testing setup (24) - Long-term maintainability

What Changed from Old App

Tech Stack Evolution

State Management Migration

• Old: Redux with actions, reducers, middleware (complex)
• New: Plain React hooks (useState, useMemo, useCallback)
• Result: 70% less boilerplate, easier to understand

Note: Skills previously claimed migration to TanStack Query + Zustand, but actual implementation uses plain React hooks only.

Why No Global State Library?

Portfolio Buddy 2 is simple enough to use React's built-in state:

• Small component tree (14 components)
• State rarely shared across distant components
• Custom hooks encapsulate shared logic effectively
• No complex async state management needed

Migration Lessons Learned

What Went Well

1. Vite adoption - Build times dropped from 30s to 2s
2. TypeScript migration - Caught many bugs early
3. Chart.js over Recharts - Better performance with large datasets
4. Simplified state - No Redux complexity
5. Supabase integration - Fast backend setup

What Could Be Better

1. Component size discipline - Let components grow too large
2. TypeScript strictness - Too many any escapes
3. Testing from start - No tests written yet (tech debt)
4. Code reviews - Need refactoring before more features
5. Documentation - Should have updated skills continuously

Migration Velocity

• Weeks 1-4: Core features (upload, parsing, basic metrics)
• Weeks 5-8: Charts, correlation, UI polish
• Weeks 9-12: Advanced features (Sortino, date filtering, sorting)
• Current: Maintenance, refactoring, optimization

Next Steps

Immediate (This Sprint)

1. Remove Recharts dependency
2. Add error boundaries to risky components
3. Fix highest-impact TypeScript any violations

Short Term (Next 2 Sprints)

1. Refactor PortfolioSection into smaller components
2. Implement Excel export
3. Complete symbol/strategy filtering

Long Term (Next Quarter)

1. Set up Vitest testing framework
2. Add CI/CD with GitHub Actions
3. Multi-period analysis UI
4. Risk scenario modeling

Feature Parity Checklist

Comparing to old Portfolio Buddy v1:

Parity Status: 85% (11/13 core features complete)