Major Feature Update: Comprehensive Dashboard Overhaul

.gitignore

@@ -44,6 +44,7 @@ next-env.d.ts
 data/*.db
 data/*.db-journal
 data/*.db-wal
+data/*.db-shm
 
 # user configuration
 config.user*
@@ -54,3 +55,11 @@ data/optimizer-jobs.json
 # claude code local settings and agents
 .claude/settings.local.json
 .claude/agents/
+
+# local development files (not for commit)
+[WEB]
+ecosystem.config.js
+scripts/aster-notifier.cjs
+*.swp
+.*.swp
+core

CLAUDE.md

@@ -144,6 +144,20 @@ npm run test:tranche:all          # Run all tranche tests
 - **pnlService.ts**: Real-time P&L tracking and session metrics
 - **thresholdMonitor.ts**: 60-second rolling volume threshold tracking
 - **trancheManager.ts**: Multi-tranche position tracking and lifecycle management
+- **tradeQualityService.ts**: Trade quality scoring (spike/volume/regime analysis, 0-3 score)
+- **cascadeDetector.ts**: Cascade protection — detects rapid consecutive entries and can LOG_ONLY, REDUCE, or BLOCK
+- **accountHealthMonitor.ts**: Account health monitoring — tracks drawdown %, pauses trading, emergency close-all
+
+### Key UI Components
+
+- **TradeQualityPanel.tsx**: Signal feed showing trade opportunities with S/V/R quality scores, TAKEN/SKIPPED filters, expandable details with info tooltips
+- **ReducePositionModal.tsx**: Partial position close modal (25/50/75/100% presets + custom %)
+- **PositionTable.tsx**: Position table with Scale Out, Add, Reduce, and Close actions
+
+### API Endpoints
+
+- **`/api/positions/[symbol]/[side]/reduce`**: POST — partial position close at market (accepts `{ percent }`)
+- **`/api/cascade/status`**: GET — cascade protection status
 
 ### API Layer (`src/lib/api/`)
 
@@ -688,3 +702,92 @@ The custom process manager (`scripts/process-manager.js`) handles:
 9. **Paper Mode**: Always recommend starting in paper mode when testing new features or strategies.
 
 10. **Documentation**: Refer to `docs/STRATEGY.md` for trading strategy details and `docs/aster-finance-futures-api.md` for API documentation.
+
+## Recent Session Changelog (Feb 2026)
+
+### Safety & Risk Management
+- **Cascade Protection** (`cascadeDetector.ts`): Detects rapid consecutive entries into the same symbol. Three modes: `LOG_ONLY` (monitor), `REDUCE` (enter with reduced size via `reducedPositionMultiplier`), `BLOCK` (prevent entry). Changed default to LOG_ONLY after analysis showed cascades spread over days/weeks, not minutes.
+- **Account Health Monitor** (`accountHealthMonitor.ts`): Tracks account drawdown %. Can pause new entries at configurable drawdown threshold, resume when recovered, and emergency close-all at critical levels. Broadcasts `account_health_update` events to dashboard.
+- **Dashboard visuals**: Orange pulsing badge for active drawdown, mode-aware cascade badge (DETECTED/REDUCED/PAUSED).
+- **Config UI**: Added cascade `mode` dropdown (LOG_ONLY/REDUCE/BLOCK), `reducedPositionMultiplier` input, and full Account Health config card with 5 fields.
+
+### Position Management
+- **Reduce Position** (end-to-end): New `ReducePositionModal` with 25%/50%/75%/100% preset buttons + custom %. API endpoint at `/api/positions/[symbol]/[side]/reduce` places MARKET reduce-only orders. Handles HEDGE mode, precision formatting, paper mode. Wired into `PositionTable` (orange ✂ Reduce button in both mobile & desktop).
+
+### Trade Quality Analysis
+- **Signal Feed redesign** (`TradeQualityPanel.tsx`): Complete rewrite from 778→340 lines. Removed: 3-tab layout, SVG circular gauges, VWAP cross dot indicator, mini bar charts. Added: collapsed-by-default compact feed, click-to-expand signal details, ALL/TAKEN/SKIPPED filters, 50 signal history.
+- **Spike detection fix** (`tradeQualityService.ts`): Fixed `detectSpike()` which always showed 0s or 119s. Was measuring from oldest price in 2-min window; now scans backward to find where the rapid move actually started, giving meaningful durations like "0.5% in 8s".
+- **Metric info tooltips**: Added (i) icons with detailed explanations to all metrics (Move, Spike, Vol, VWAP) and the S/V/R score triplet. Each tooltip explains what the metric measures, scoring thresholds, and what good/bad values look like.
+
+### Local Trade History Database (Feb 13, 2026)
+- **`src/lib/db/tradeHistoryDb.ts`** (NEW): Local SQLite DB (`data/trade_history.db`) persisting all trades/orders for deep history. Three tables: `trade_history` (order fills, UPSERT by symbol+orderId+updateTime), `income_history` (PnL/commission/funding), `sync_metadata` (backfill progress tracking). WAL mode, indexed on symbol/time/status/orderId.
+- **Real-time persistence**: `positionManager.ts` `handleOrderUpdate()` now calls `tradeHistoryDb.upsertTrade()` for every ORDER_TRADE_UPDATE WebSocket event (non-blocking try/catch).
+- **Startup backfill**: `scripts/backfill-trades.ts` runs on bot startup (background, non-blocking via `bot/index.ts` `startTradeHistoryBackfill()`). Fetches allOrders + userTrades + income from exchange API going back 30 days. Uses `sync_metadata` to avoid refetching. First run imported 1,873 orders, 546 trades, 1,871 income records in 36.8s.
+- **API endpoints**: `/api/trades/history` (GET — query with symbol/status/time filters, `format=orders|markers|raw`), `/api/trades/stats` (GET — aggregate PnL/commission/funding stats + sync status).
+- **Enhanced `/api/orders/all`**: Now merges local DB history with exchange API data. Falls back to local DB if exchange API fails. Adds deeper history orders not in the API response.
+- **TradingView chart**: `TradingViewChart.tsx` order overlay now fetches from `/api/trades/history?format=orders` for 90-day deep history (was limited to ~50 from exchange API), merges with real-time orderStore updates.
+- **Recent Orders pagination**: `RecentOrdersTable.tsx` changed from infinite-expand to proper pagination. Shows first 15 rows, "Show More" expands to 50-per-page with prev/next page controls and a collapse button.
+
+### Trade Quality Scoring Reference (S/V/R)
+Each trade opportunity scores 0-3 based on three criteria:
+- **S**pike (0/1): Was there a fast price move into the level? Scores 1 if velocity >0.1%/s OR total move ≥0.5%
+- **V**olume (0/1): Is liquidation volume decreasing? Scores 1 if recent/older volume ratio ≤1.1×
+- **R**egime (0/1): Is the market choppy? Scores 1 if ≥3 VWAP crosses/hour (range-bound = good for reversals)
+- **3/3 STRONG** → 1.5× position size | **2/3 NORMAL** → 1× | **1/3 WEAK** → 0.5× | **0/3 SKIP** → blocked
+
+## Active Discussion: Account Health Settings & Global Risk Mode (Feb 13)
+
+### Account Health Monitor — NOT YET CONFIGURED (using defaults)
+The health monitor exists in code but `config.user.json` has no `accountHealth` section.
+Current defaults: 25% drawdown pause, 20% unrealized loss pause, 15% resume, 60s checks, no emergency close-all.
+
+**Recommended settings based on trade data analysis:**
+- `maxDrawdownPercent: 5` (~$18) — normal daily P&L is $1–5, so $18 drawdown = many days of profit gone
+- `maxUnrealizedLossPercent: 3` (~$11) — with $1 trades at 8x, unrealized > $11 means multiple underwater positions
+- `resumeAtDrawdownPercent: 3` — 2% hysteresis band
+- `checkIntervalSeconds: 30` — faster detection, positions are small
+- `closeAllAtDrawdownPercent: 10` (~$36) — month of profit, hard safety limit
+
+### Trade Performance Data (30 days: Jan 14 – Feb 13, 2026)
+- **Account balance**: ~$360
+- **All trades**: $1 trade size, 8–10x leverage, no SL/TP configured
+- **Symbols**: ETHUSDT, ASTERUSDT, HYPEUSDT, ZECUSDT, SOLUSDT, FARTCOINUSDT
+- **maxOpenPositions**: 3, **useTradeQualityScoring**: false, **cascadeProtection**: LOG_ONLY
+- **Normal performance** (excluding Jan 30-31): 97W/0L, 16 consecutive green days, +$259 cumulative
+- **Catastrophic event** (Jan 30-31): -$1,441 in 2 days (400% of account). Caused by enormous positions ($1,900–$2,000 notional vs normal $1 trades) — likely manual positions or extreme DCA. All closed simultaneously on Jan 31 at 18:43:59.
+- **Post-recovery** (Feb 1–13): Back to $1 trades, 100% win rate, +$9.15 in 13 days, slowly recovering.
+- **Per-symbol**: ZECUSDT best (+$54, 12W/0L), HYPEUSDT (+$43, 20W/1L), SOLUSDT worst (-$820, 3 losses each >$200)
+- **Worst individual trades**: SOLUSDT -$603 (notional $1,928), ETHUSDT -$441 (notional $1,249)
+
+### Global Risk Mode — DESIGN DISCUSSION (not yet implemented)
+Concept: A single risk profile selector that scales all symbol configs uniformly.
+
+**Proposed presets:**
+| Mode | Size Multiplier | Max Positions | Description |
+|---|---|---|---|
+| Conservative | 0.5x | 2 | Choppy/uncertain market |
+| Normal | 1.0x | 3 | Business as usual |
+| Aggressive | 2.0x | 4 | Trending/high confidence |
+| Max | 3.0x | 5 | Very high confidence |
+
+**What it would affect:**
+- Trade size: multiply all symbol `tradeSize` by mode multiplier
+- Max open positions: per mode preset
+- Entry thresholds: aggressive = lower liq volume needed to trigger trade
+- Health monitor thresholds: scale proportionally with trade size (but NOT closeAll)
+- SL/TP widths: possibly wider SL in aggressive mode
+
+**Design options discussed:**
+- **Option A** (recommended): Named presets with hardcoded multipliers — simple to flip
+- **Option B**: Named presets + customizable overrides per field
+- **Option C**: Single `globalMultiplier` slider, everything scales from one number
+
+**Key principle**: Risk mode is a MULTIPLIER on top of per-symbol config, not a replacement. ETHUSDT tradeSize=1 in Aggressive(2x) → actual $2.
+
+**Open questions for user:**
+1. Which levers matter most? (trade size? position count? thresholds?)
+2. Named presets or a slider?
+3. Should it change which symbols are traded, or just how much?
+
+**Storage**: Would be a `riskMode` field in `global` config.
+**UI**: Prominent selector at top of dashboard or config page, color-coded by risk level.

CONVO

@@ -0,0 +1,66 @@
+Account Health Monitor — Recommended Settings
+Your data tells a very clear story:
+
+The good: 16 consecutive green days (Jan 14–29), +$259 cumulative profit, 97/107 wins on normal days. With $1 trade sizes you're extremely consistent.
+
+The catastrophe: Jan 30–31 wiped $1,441 — that's 400% of your current balance. The losses came from positions with $1,900–$2,000 notional value (vs your normal $1 trade size). These were clearly manually opened or DCA'd positions that went massively underwater. The Jan 31 batch at 18:43:59 all closed simultaneously — looks like a manual close-all or a liquidation.
+
+Recommended Health Monitor Settings
+Given a ~$360 account running $1 trades:
+
+Setting	Recommendation	Why
+enabled	true	Obviously
+maxDrawdownPercent	5% (~$18)	Your normal daily P&L is $1–$5. A $18 drawdown from peak is 4–18 days of profit — something is very wrong at that point
+maxUnrealizedLossPercent	3% (~$11)	With $1 trades at 8x, max loss per position is ~$1. If unrealized loss hits $11, you have multiple underwater positions building up
+resumeAtDrawdownPercent	3%	Resume once recovered — gives 2% hysteresis band
+checkIntervalSeconds	30	Faster detection. Your positions are small so API cost is worth it
+closeAllAtDrawdownPercent	10% (~$36)	Emergency nuclear option. 10% of $360 is $36 — that's a month of profit. If you're down that much, close everything
+These are tight thresholds because your trade sizes are tiny. You should never be down 5% from $1 trades — if you are, something external happened (manual position, DCA spiral, etc).
+
+Global Risk Mode — Design Discussion
+Now for the risk-on/risk-off idea. Here's how I'd think about it:
+
+The Concept
+A single risk profile selector that acts as a multiplier/modifier across all symbols and behaviors. Something like:
+
+Mode	Trade Size Multiplier	Max Positions	Threshold Sensitivity	SL Width	Description
+Conservative	0.5x	2	Tighter	Tighter	Low confidence / choppy market
+Normal	1.0x	3	Default	Default	Business as usual
+Aggressive	2.0x	4	Looser	Wider	High confidence / trending market
+Max	3.0x	5	Loosest	Widest	You're very confident
+Key Design Questions
+1. What should it actually affect?
+
+Trade size: Most obvious — multiply all symbol tradeSize by the mode's multiplier
+Max open positions: More aggressive = allow more simultaneous positions
+Entry thresholds: Aggressive = lower the liquidation volume needed to trigger a trade (take more signals)
+Health monitor thresholds: Aggressive = widen the drawdown tolerance proportionally
+SL/TP: Maybe wider SL in aggressive mode (give trades more room), tighter in conservative
+2. How should it be stored?
+
+Option A: A single riskMode string in global config ("conservative" | "normal" | "aggressive" | "max"), with the multipliers hardcoded. Clean and simple.
+
+Option B: A riskMode string + customizable multiplier overrides. e.g., you pick "aggressive" but override the trade size multiplier to 1.5x instead of 2x. More flexible.
+
+Option C: Just a single globalMultiplier number (0.5 to 3.0). Everything scales off that one number with sensible formulas. Simplest, but less granular.
+
+3. How does it interact with per-symbol config?
+
+The risk mode would be a multiplier on top of existing per-symbol settings, not a replacement. So if ETHUSDT has tradeSize: 1 and you're in Aggressive mode (2x), the actual trade size becomes $2. Your per-symbol config is the "base" and the risk mode scales it.
+
+4. Should it affect the health monitor?
+
+I think yes — in Aggressive mode, the health monitor thresholds should scale proportionally. If you're taking 2x trades, a 5% drawdown is expected to happen faster, so maybe the threshold becomes 8%. But closeAllAtDrawdownPercent should probably stay fixed (it's a hard safety limit).
+
+5. UI: where does it live?
+
+A prominent toggle/selector at the top of the dashboard or in the config page. Maybe even color-coded — green for conservative, yellow for normal, orange for aggressive, red for max. Something you can flip quickly.
+
+My Recommendation
+I'd go with Option A + light customization: a named risk mode with sensible presets, where the main lever is the trade size multiplier. Keep it dead simple to flip between modes. The multiplier values and position limits would be defined as presets but overridable in config.
+
+What's your thinking? Specifically:
+
+Which of those levers matter most to you? (trade size? position count? thresholds?)
+Do you want it to feel like 3–4 named presets, or more like a slider?
+Should it change anything about which symbols are traded, or just how much?