What's New
Latest updates and improvements to TargetHit
January 2026
62 updates
January 28, 2026
P5 Portfolio Launch
New expectancy-optimized portfolio using DPO-gated pairwise indicator combinations.
- 10 edges across BTC, ETH, SOL — 5 LONG / 5 SHORT
- SL/TP exits only (no timeout/horizon)
- Train-first methodology with 80% OOS reproduction requirement
P2 ETH V3 Deployment
Deployed 12 new P2 edges with diversified SOL + BTC coverage.
- Edges selected for strong training AND OOS performance
- New hourly forward test pipeline integration
- Database-driven edge loading replacing hardcoded definitions
January 24, 2026
P3 Portfolio Selection
Selected P3 edges with 84% average OOS win rate.
- 12 edges: 3 ETH edges with 100% OOS WR (8/8 wins each)
- 10 unique indicator groups for signal diversity
- Higher training expectancy threshold (2.5%+)
January 21, 2026
Portfolio 4 Launched: Multi-Coin SL/TP System
New P4 portfolio with 12 edges across BTC and SOL (6 LONG, 6 SHORT)
- BTC edges: 3 LONG, 3 SHORT from gentle discovery
- SOL edges: 3 LONG, 3 SHORT with 75%+ OOS win rates
- SL/TP based exits for precise risk management
- Avg backtest PF: 4.33
Portfolio 2 Upgraded: ETH V3 Edges
P2 now runs 14 ETH edges with SL/TP exits (8 LONG, 6 SHORT)
- All edges use DPO_20_z as primary indicator
- SL/TP based exits - no time horizon
- Diverse secondary indicators for signal variety
- Expected ~4-5 signals per day
January 20, 2026
Live Signals Portfolio Tabs Redesign
Portfolio tabs now show active signal count and exit method
- Badge shows Active (x) count instead of static labels
- Description line shows exit method (2h Horizon or SL/TP Exits) and coins
- Green highlight when portfolio has active signals
- Cleaner visual hierarchy
P4 Portfolio Deployment
Deployed P4 Portfolio with 12 BTC/SOL edges using SL/TP exits
- 6 SOL edges (3 LONG, 3 SHORT)
- 6 BTC edges (3 LONG, 3 SHORT)
- All edges use DPO_20 + secondary indicator
- SL/TP exit only (no horizon timeout)
- Edges: globalAccounts_longPct, PSARr_0.02_0.2, limitOrderCount_delta, anchoredTopTraderPositions, SQUEEZE_SQZ_ON
- Expected ~4-5 signals per day
Live Signals Dashboard Stats Fix
Fixed incorrect expectancy and signals per day calculations on live signals page
- Expectancy now calculated from actual signal performance data
- Fixed percentage display formatting
- Added accurate signals per day calculation
- P3 stats now display correctly
P3 Discovery Improvements
Enhanced edge discovery process for P3 with SL/TP only exits
- Trades now exit only on Stop Loss or Take Profit - no timeouts
- Improved data validation methodology
- Over 3,500 potential edges discovered across SOL and BTC
- Balanced LONG/SHORT distribution
January 19, 2026
P3 Exit Logic Fixed
P3 signals now exit only on Stop Loss or Take Profit - no more horizon timeouts
- P3 scanner no longer sets expiration time
- P3 evaluator only checks SL/TP levels
- P1 and P2 horizon logic unchanged
P3 Portfolio V2 - Complete Edge Replacement
Replaced all 8 P3 edges with new discovery results
- 4 SHORT (UI_14, SQUEEZE_SQZ_ON, topTraderAccounts_lsRatio, PSARr) + 4 LONG (CANDLE_BODY_RATIO, SQUEEZE_SQZ_OFF, BOP, anchoredGlobalAccounts_cumulativeLongPct)
- Avg Training WR: 54.8%, Avg Training ER: 1.70%
- SL/TP only exits (no timeout)
January 18, 2026
Footer + Dashboard Legal Links
Added Footer component to dashboard and fixed social links
- Footer with legal links now appears on logged-in dashboard pages
- Fixed YouTube link to @TargetHit
- Fixed Telegram link to t.me/targethitfree
Legal Pages for Ads Compliance
Created comprehensive legal pages required for Google/Meta ads and organic SEO
- Terms of Service - covers VIP membership, auto-trading, liability
- Privacy Policy - GDPR/CCPA compliant with data collection disclosure
- Cookie Policy - details all tracking cookies and consent options
- Risk Disclaimer - critical for trading/financial services compliance
- Refund Policy - 7-day refund window, Stripe compliance
- New Footer component with legal links across all pages
- Updated sitemap to include all legal pages
File Upload API for Claude
Added /upload endpoint to Claude Runner v3 for reliable file transfers
- Supports text mode for code files
- Supports base64 mode for binary files
- Eliminates JSON escaping issues
- 50MB max upload size
Course Graduate Discord Role
Users who complete all 43 lessons now automatically receive a special Discord role
- Discord role ID: 1462739423035658261
- Role assigned when user marks final lesson complete
- Requires Discord account to be linked
- Added courseGraduateRoleAssigned tracking field
Fixed Course Page 404 Errors
Added missing level index pages to fix 404 errors on /learn/foundation, /learn/data, etc.
- Created /learn/[level]/page.tsx to display lessons for each course level
- URLs like targethit.ai/learn/foundation now work correctly
MEXC AutoTrade: Web Token Required
Discovered MEXC futures API limitation - now requires browser session token instead of API keys
- CCXT library does not support MEXC futures orders via API keys
- MEXC futures API has been in maintenance since 2022
- Workaround: Use web token from browser cookies (u_id)
- Affected users notified via email with setup instructions
- Updated CCXT to version 4.5.33
January 17, 2026
School of Quant Trading - 100% COMPLETE
All 43 lessons are now live at targethit.ai/learn
- FOUNDATION: 6 lessons on mindset and system overview
- DATA: 8 lessons on alternative data sources
- TESTING: 7 lessons on backtesting done right
- DISCOVERY: 6 lessons on finding real edges
- RISK: 5 lessons on position sizing and survival
- AUTOMATION: 6 lessons on building trading bots
- PRODUCTION: 5 lessons on going live and staying alive
- Total reading time: ~17 hours of comprehensive education
RISK Level Complete - 5 New Lessons
Deployed all 5 RISK level lessons to School of Quant Trading
- Position Sizing: The Most Important Decision
- Drawdown Management: Staying in the Game
- Correlation and Portfolio Risk
- Stop Losses: Science vs Art
- Risk Budgeting Across Strategies
- Course now at 32/43 lessons (74% complete)
Bitget AutoTrade Fix
Fixed 100% Bitget trade failure rate
- ccxt Bitget returns balance info as list not dict
- Updated equity extraction to handle list response
- Workers restarted successfully
January 16, 2026
Discord VIP Role Sync
Deployed bidirectional Discord VIP role sync cron job.
- Runs every 15 minutes — grants VIP role on purchase, revokes on expiry
- Bidirectional: syncs both Discord → website and website → Discord
School of Quant Trading - DISCOVERY Level Complete
Added all 6 DISCOVERY level lessons to the free education course
- Lesson 4.1: The Edge Discovery Framework - systematic edge finding
- Lesson 4.2: Z-Score Normalization - making indicators comparable
- Lesson 4.3: Regime Filtering - context-dependent trading
- Lesson 4.4: Combining Indicators - multi-factor approach
- Lesson 4.5: Testing Millions of Combinations - scale without overfitting
- Lesson 4.6: From Discovery to Production - the promotion pipeline
- Course progress: 27/43 lessons (63%) now live at targethit.ai/learn
School of Quant Trading: Foundation Level Complete
Free education course now has 11 lessons live - all 6 Foundation lessons plus 5 Data lessons
- Foundation Level 100% complete (6 lessons)
- Data Level 62% complete (5/8 lessons)
- New lessons: Course Roadmap, Alternative Data Intro, Funding Rates, Liquidation Levels
- Overall course 25% complete
School of Quant Trading Launched
Free educational course now live at targethit.ai/learn
- First 4 posts published covering Foundation and Data levels
- Topics: Why strategies fail, Edge vs Strategy, Beyond price/volume, Open Interest
- Full course will have 43 posts across 7 levels
January 15, 2026
Homemade Bot Protocol
Established standardized protocol for deploying new Polymarket prediction bots.
- Required: backtest, INSTRUCTIONS.md, MANIFEST.json per bot
- Minimum 30-trade OOS, 2.0+ profit factor for deployment
- Integration with copytrade engine for subscriber execution
MEXC Balance Display Fix
Fixed MEXC showing /bin/sh.00 balance on website
- Root cause: test-connection.py was hardcoded to return balance:0 for MEXC
- Now uses adapter get_balance() which handles proxy authentication correctly
BitGet Close Fix - Hedge Mode
Fixed hedge mode side parameter for BitGet close orders
- Root cause: BitGet hedge mode with tradeSide=close requires side=buy to close LONG and side=sell to close SHORT (opposite of normal trading)
- Our code had it backwards - verified fix with actual API test that successfully closed positions
MEXC Futures API Bypass Implemented
Added bypass adapter for MEXC Futures API that uses WEB token instead of API keys
- MEXC official Futures API has been in maintenance since July 2022
- New adapter uses internal web API with WEB token from u_id cookie
- Users need to: Login to MEXC -> F12 -> Application -> Cookies -> Copy u_id value
- Store WEB token in webToken field instead of apiKey
- Based on: github.com/biberhund/MEXC_Future_Order_API_Maintenance_Bypass
Telegram Notification Fix
Fixed issue where VIP users were not receiving Telegram signal notifications
- Root cause: Users connected Telegram via login widget but never started a chat with the bot
- Telegram requires users to initiate conversation before bot can send messages
- Added bot start prompt after Telegram linking
- Sent email notification to 53 affected VIP users with instructions
Portfolio Number in Notifications
Signals now show P1 or P2 in all notification channels
- Discord shows P1/P2 in embed title
- Telegram shows P1/P2 in message header
- Email shows P1/P2 in subject line and body
BYDFI Equity Calculation Fix
Fixed position sizing to use equity instead of balance
- Was using wallet balance causing oversized positions
- Now calculates equity = balance + unrealized PnL
- Prevents 250%+ position sizes when account has losses
P2 Scanner Z-Score Fix
Fixed P2 scanner to calculate z-scores for all indicators
- Scanner was only computing z-scores for 7 hardcoded indicators
- Now computes z-scores for ALL numeric columns dynamically
- P2 scanner now generating signals correctly
Backtest Stats Now Show in Telegram
Fixed notifications showing 0W/0L instead of actual backtest statistics
- Root cause: signal-router.py missing backtest_wins and backtest_events in base_data
- Signals now display correct stats like 55W/6L (90% WR)
P2 Scanner Fixed - All 23 Indicators Active
Fixed Portfolio 2 scanner that was missing most indicator calculations
- Added 4 missing HyBlock endpoints (globalBidAskRatio, openInterest, traderSentimentGap)
- Added 15+ technical indicators (ALMA, EMA, FWMA, HMA, PPO, TRIX, PVOL, etc.)
- Scanner now calculates all 23 required indicators for V8 edges
Discord Changelog Channel
New #changelog channel synced with website changelog page
- Updates posted to Discord and targethit.ai/changelog simultaneously
- 36 historical entries added (Dec 27 - Jan 16)
- Automated posting via TargetHit AI bot
FAQ & Mobile Navigation Updates
Added TP/SL explanation to FAQ and fixed mobile menu navigation
- New FAQ entry explaining time-based vs price-based signal exits
- Mobile hamburger menu now includes FAQ and Changelog links
Discovery Engine V8 - Robustness Scoring
New edge discovery methodology prioritizing future performance over past results
- Robustness scoring: fold consistency 25%, regime consistency 25%, sample credibility 20%
- Stricter thresholds: 88% min walk-forward accuracy, 75% holdout accuracy
- Excludes indicators with lookahead bias (future_return patterns)
- ICS_26 contamination discovered - 785/1747 V8 edges affected
January 14, 2026
Pipeline Health Dashboard
New admin dashboard for monitoring data pipeline and signal generation health
- Real-time status of HyBlock data freshness
- P1 and P2 scanner monitoring
- Edge stats updater tracking
- Accessible at /admin/pipeline
P2 V8 Edge Deployment
Deployed 26 V8 edges to Portfolio 2 with 90% average accuracy
- 11 LONG edges (91% avg WF accuracy)
- 15 SHORT edges (89% avg WF accuracy)
- All use 2h horizon with DPO_20_z primary indicator
- Expected 5-6 signals/day at 90% accuracy
P2 Overfitting Fix
Identified and fixed overfitting in P2 mean reversion edges
- Previous thresholds too strict: 60-114 samples with 98-100% accuracy
- Optimal thresholds changed to {2.0, 1.5} for 2,200+ events at 90%+ accuracy
- Deleted false signals generated by broken scanner
January 13, 2026
Optimal Data V3 — 3-Year Backfill
Completed 3-year historical data backfill across all endpoints.
- 2,182,702 parquet files totaling 16.34 GB
- Date range: 2023-01-09 to present (1,096 days)
- All 121 HyBlock endpoints fully populated
January 11, 2026
Discovery V8 — Deep Discovery
New deep discovery engine scanning all indicator combinations.
- Supports optimal_v3 dataset with 2,200+ columns
- Walk-forward testing with regime analysis
- Per-coin parallel execution for faster discovery
January 10, 2026
Discord Notification Fix - 50x Message Reduction
Fixed critical bug sending 50-60 Discord messages per signal instead of 1
- Signal router was queueing one message per user instead of per signal
- Restructured code to send single channel message per signal
- Added Discord/Telegram account disconnect buttons to settings
HyperLiquid Rate Limiting
Implemented per-exchange rate limiting to prevent API throttling
- HyperLiquid limited to 2 concurrent requests with 1s delays
- Other exchanges unaffected
- Resolved 429 error issues affecting trade execution
Optimal Data V2 Pipeline
New ML-ready data pipeline for edge discovery with 923 columns
- 7-phase build process for gap-free parquet files
- Pre-computed z-scores, percentiles, ROC values
- IC (Information Coefficient) calculation for indicator profiling
- 60 strong edges identified with EXTREME type signals
January 9, 2026
Signal Router Crash Fix
Fixed Redis data type error causing 11.5-hour service outage
- AttributeError on .decode() for already-decoded string
- Fixed health monitoring page undefined value crashes
- Updated URL redirections to /live-signals
January 8, 2026
AutoTrade v3 Deployment
Major auto-trade infrastructure upgrade with distributed workers
- autotrade-worker-v3: 25 workers on port 8765
- autoclose-worker-v3: 20 workers on port 8766
- Circuit breakers and health endpoints added
- V2 workers kept as backup
Free Telegram Channel Launch
Public channel receives one VIP signal per day after 3 AM ET
- Signals sent to @targethitfree channel
- Promotional link added to free channel messages
- Redis tracking for daily signal limit
Data Quality Backfill
Fixed 121 signals missing exit prices and 631 trades missing PnL
- Calculated exit prices from PnL percentages
- Derived missing PnL from signal exit prices
- Auto-Trade Health dashboard enhanced with per-exchange details
Claude Direct Server Access
Enabled Claude AI to execute commands directly on TargetHit servers
- ngrok tunnel for secure remote execution
- Robust HTTP server with multi-threading
- Sync script for automated context gathering
- Eliminates 400+ daily copy-paste operations
Market Cache Fix - CCXT Attributes
Fixed KeyError: 0 affecting Binance and HyperLiquid trades
- Cache was missing currencies, symbols, ids attributes
- Extended cache TTL from 1 hour to 12 hours
- Thread-safe locking prevents parallel market loads
January 7, 2026
Security Middleware Implementation
Protected all sensitive API endpoints after security audit
- NextAuth JWT authentication on /api/edges, /api/signals, /api/performance
- Updated robots.txt to stop advertising sensitive routes
- All endpoints now return 401 for unauthenticated requests
HyBlock API Complete Mapping
Discovered and documented all 73 available HyBlock endpoints
- Previously using 46 indicators, now have access to all 73
- 27 new indicators added to DATA-MONSTER
- Complete parameter requirements documented
Discovery Engine V7.1/V7.2 - Vectorized
Edge discovery optimized from 20+ days to minutes
- Multi-stage funnel with IC pre-screening
- Pre-computed signal masks for vectorized processing
- Walk-forward testing and regime analysis
- DPO_20 identified as primary alpha source
January 5, 2026
Notification System Overhaul
Fixed notifications only reaching auto-trade users
- New get_notification_users() queries ALL VIP users
- 71 VIP users now receive notifications (was only auto-trade subset)
- Changed display from PF to Forward W/L format
January 4, 2026
Comprehensive SEO Implementation
Added unique metadata to 25+ pages for better search visibility
- Created sitemap.xml and robots.txt
- Structured data markup added
- Google Search Console verification
- SOL and ETH trading guides created
January 3, 2026
Bitcoin Checkout Flow Redesign
New step-by-step checkout page for Bitcoin payments
- Intermediate checkout page with payment instructions
- Works with any wallet or exchange
- Custom BTCPay checkout styling
Forward Test Stats Display Fix
Fixed edge cards not showing forward test statistics
- Forward test wins/losses now display on edge cards
- Exit prices properly calculated and stored
- Fixed 195 signals missing exit price data
January 2, 2026
Portfolio #4 Launch - 91% Win Rate
Deployed new high-performance edge portfolio
- 14 new edges across BTC, SOL, and ETH
- 91.1% win rate on 45 forward-tested signals
- +27.4% total P&L in first week
- Retired 46 old SOLO edges
Live Signals Page Enhancements
Edge details now show regime performance and forward test data
- Click any edge for detailed stats
- Regime performance breakdown (Bull/Bear/Sideways)
- Current market regime indicator
- Signal history with entry/exit prices
January 1, 2026
Auto-Trade 15x Speed Improvement
Parallelized autoclose worker for faster position closing
- Now processes 15 trades in parallel
- Reduced closing delays from minutes to seconds
- Fixed sequential processing bottleneck
Exchange Adapter Fixes
Fixed multiple exchange integration issues
- Bybit: Fixed API timeout with proxy latency
- HyperLiquid: Resolved rate limiting (429 errors)
- BYDFI: Fixed position size calculation (was 100x too small)
- Bitget: Fixed hedge mode vs one-way mode detection