Rewrite system prompt + watcher to focus on profit over trade volume

- System prompt: remove 'redeploy freed capital immediately' from tool protocol
- System prompt: replace 'Overtrading (max 10-50/day)' with quality-focused guidance
- System prompt: de-emphasize trade count goals, emphasize P&L targets
- Watcher: rewrite proactive wake message to prioritize portfolio management
- Watcher: remove trade count target pressure ('X/15 trades — Y more needed')
- Watcher: wake for portfolio checks regardless of trade count
- Watcher: restructured wake priorities (manage positions > review performance > find edge)

Author: Ope Olatunji

package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "@agenticmail/enterprise",
-  "version": "0.5.488",
+  "version": "0.5.489",
   "description": "AgenticMail Enterprise — cloud-hosted AI agent identity, email, auth & compliance for organizations",
   "type": "module",
   "bin": {

package.json

@@ -1230,13 +1230,6 @@ function _startFastLoop(db: any, opts: WatcherEngineOpts) {
             ).catch(() => ({ cnt: 0 }));
             const tradeCount = parseInt(todayTrades?.cnt || '0');
 
-            // Check min daily trades goal (filter by agent_id)
-            const goal = await edb.get(
-              `SELECT target_value FROM poly_goals WHERE agent_id = ? AND type = 'min_trades_daily' AND enabled = 1`,
-              [agentId]
-            ).catch(() => null);
-            const targetTrades = goal?.target_value || 15;
-
             // Check if user explicitly paused proactive wakes (said stop/abort)
             const pauseRow = await edb.get(
               `SELECT paused_at FROM poly_proactive_pause WHERE agent_id = ?`, [agentId]
@@ -1262,8 +1255,8 @@ function _startFastLoop(db: any, opts: WatcherEngineOpts) {
               continue;
             }
 
-            // Wake agent via HTTP API if behind on trades
-            if (tradeCount < targetTrades) {
+            // Wake agent via HTTP API for proactive portfolio check
+            {
               // ── Balance gate — don't wake if wallet can't afford minimum trade ──
               try {
                 const walletRow = await edb.get(`SELECT address FROM poly_wallets WHERE agent_id = ?`, [agentId]).catch(() => null);
@@ -1288,29 +1281,27 @@ function _startFastLoop(db: any, opts: WatcherEngineOpts) {
                 const dep = agentConfig?.deployment;
                 const port = dep?.port || dep?.config?.local?.port || 3101;
                 const secret = process.env.AGENT_RUNTIME_SECRET || '';
-                const wakeMsg = `[PROACTIVE TRADING CHECK] You have ${tradeCount}/${targetTrades} trades today — ${targetTrades - tradeCount} more needed.
+                const wakeMsg = `[PROACTIVE PORTFOLIO CHECK] Today: ${tradeCount} trades placed.
 
-MANDATORY SEQUENCE (do ALL of these IN ORDER before placing any new trades):
+PRIORITY 1 — MANAGE EXISTING POSITIONS:
+1. poly_watcher_events action=check — Check for unread signals. Act on critical ones.
+2. poly_daily_scorecard — Review today's P&L. Are you profitable?
+3. poly_position_heatmap — Check all open positions. CRITICAL/HIGH urgency first.
+4. poly_exit_strategy action=check — Any exit conditions triggered?
+5. poly_drawdown_monitor action=check — Portfolio-level risk. HALT if drawdown > 15%.
 
-1. poly_watcher_events action=check — Check for unread signals first
-2. poly_goals action=check — Review your performance targets
-3. poly_drawdown_monitor action=check — Check portfolio-level risk. HALT if drawdown > 15%
-4. poly_calibration — Review your prediction accuracy. Are you over/under-confident?
-5. poly_pnl_attribution — Which strategies and categories are making/losing money?
-6. poly_strategy_performance — Which strategies are actually profitable? Double down on winners.
-7. poly_get_positions — Review all open positions, check P&L drift
-8. poly_exit_strategy action=check — Check if any exit conditions triggered
+PRIORITY 2 — REVIEW PERFORMANCE:
+6. poly_calibration — Are your predictions accurate?
+7. poly_pnl_attribution — Which strategies/categories are making money?
+8. poly_goals action=check — Review P&L targets (not trade count targets).
 
-ONLY AFTER completing the above analysis:
-9. poly_screen_markets with strategies: high_volume, momentum, contested, closing_soon
-10. poly_search_markets for sports: NBA, MLB, Premier League, Champions League, UFC
-11. For EACH candidate: run poly_quick_analysis + poly_resolution_risk + poly_manipulation_detector
-12. Use poly_kelly_criterion for position sizing — do NOT just buy random $5 positions
-13. For orders > $50: use poly_scale_in (TWAP/VWAP), NOT market orders
-14. For time-sensitive opportunities: use poly_sniper with trailing limits
-15. Consider poly_hedge for correlated positions
+PRIORITY 3 — LOOK FOR OPPORTUNITIES (ONLY if you have edge):
+9. poly_momentum_scanner — Any movers with real edge?
+10. poly_breaking_news — News-driven opportunities?
+11. For candidates: poly_quick_edge → poly_resolution_risk → size with Kelly
 
-QUALITY > QUANTITY. Each trade must have analysis backing it. No blind trades.`;
+Remember: NO good setups = NO new trades. Managing profitable positions IS working.
+A day with $50 profit from 2 trades beats a day with -$10 from 15 trades.`;
 
 
                 // Determine the manager's preferred communication channel
@@ -1366,7 +1357,7 @@ QUALITY > QUANTITY. Each trade must have analysis backing it. No blind trades.`;
                     const dc = _proactiveDailyCount[agentId] || { date: new Date().toISOString().slice(0, 10), count: 0 };
                     dc.count++;
                     _proactiveDailyCount[agentId] = dc;
-                    log(`[poly-watcher] Proactive trading: woke agent ${agentId.slice(0,8)} (${tradeCount}/${targetTrades} trades, check ${dc.count}/${maxDaily} today, interval ${intervalMins}m)`);
+                    log(`[poly-watcher] Proactive check: woke agent ${agentId.slice(0,8)} (${tradeCount} trades today, check ${dc.count}/${maxDaily}, interval ${intervalMins}m)`);
                     _lastSpawnByAgent[agentId] = now;
                     // Sync state to 'running' so UI reflects actual status
                     await edb.run(

src/agent-tools/tools/polymarket-watcher.ts

@@ -90,36 +90,60 @@ Then: handle CRITICAL/HIGH positions first, scan opportunities, record lessons.
 4. \`poly_watcher_config action=set\` with cheap model (xai/grok-3-mini recommended)
 5. \`poly_setup_monitors\` → creates full suite (BTC tracker, news scanner, geo scanner, sentiment, arbitrage, etc.)
 
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+## TRADING PHILOSOPHY — PROFIT OVER ACTIVITY
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+**Your job is to MAKE MONEY, not to place trades.** The best traders spend most of their time watching and waiting. A day with 0 new trades but well-managed positions is better than a day with 20 random $5 bets.
+
+### TIME HORIZONS — Mix these for maximum profit:
+| Horizon | Duration | Strategy | When |
+|---------|----------|----------|------|
+| **Scalp** | Minutes-hours | Momentum, news spikes, mispricing | Breaking news, sudden volume surge |
+| **Swing** | 1-7 days | Trend-following, event anticipation | Clear directional setup, upcoming catalyst |
+| **Position** | 1-4 weeks | Fundamental conviction, value bets | High-confidence edge on underpriced outcomes |
+| **Hold to resolution** | Weeks-months | Deep research, contrarian | Strong thesis, >15% edge, patient capital |
+
+**CRITICAL:** When you BUY, decide your time horizon FIRST. A scalp has tight stops and quick exits. A position trade should NOT be panic-sold on a 2% dip. Tag every trade with its horizon.
+
+### PATIENCE IS A STRATEGY
+- No good setups? **Don't trade.** Monitor positions, review performance, research.
+- Holding profitable positions IS working. You don't need to sell winners just to "do something".
+- 3 well-researched $25 trades beat 15 random $5 trades every time.
+- Sitting in cash during uncertainty is a valid, profitable decision.
+
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 ## THE TRADING LOOP
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
-### 1. SCAN — Find opportunities
-\`poly_momentum_scanner\` (find movers NOW) → \`poly_breaking_news\` → \`poly_search_markets\` → \`poly_calendar_events action=upcoming\`
-💡 **\`poly_momentum_scanner\` first** — markets moving RIGHT NOW have the highest edge.
+### 1. SCAN — Find opportunities (don't force trades)
+\`poly_momentum_scanner\` (find movers NOW) → \`poly_breaking_news\` → \`poly_search_markets\`
+💡 **Only proceed if you find genuine edge.** No edge? Stop scanning and manage existing positions.
 
 ### 2. DECIDE — Quick GO/NO-GO per candidate
 \`poly_quick_edge token_id="..." estimated_prob=0.XX bankroll=YY\` — One-call decision with edge %, Kelly size, GO/NO-GO.
-If \`decision\` is STRONG_BUY or BUY → proceed to execute. If MARGINAL → run deeper analysis. If NO_TRADE → skip.
+If \`decision\` is STRONG_BUY or BUY → proceed. MARGINAL → deeper analysis OR skip. NO_TRADE → **move on immediately**.
 
 ### 2b. DEEP ANALYZE (only for MARGINAL candidates or large positions)
-\`poly_resolution_risk\` → \`poly_manipulation_detector\` → \`poly_regime_detector\` → \`poly_smart_money_index\` → \`poly_recall_lessons\`
+\`poly_resolution_risk\` → \`poly_manipulation_detector\` → \`poly_regime_detector\` → \`poly_recall_lessons\`
 
 ### 3. CHECK RISK — Before every trade
 \`poly_profit_lock current_pnl=X daily_target=Y\` — Returns your trading mode. If LOCKED → stop trading. If CONSERVATIVE → half size.
 \`poly_record_prediction\` (ALWAYS before trading)
 
-### 4. EXECUTE
-- Re-check slippage: \`poly_market_microstructure\`
-- Orders <$500 liquid: \`poly_place_order\`. Orders >$500 or thin: \`poly_scale_in\` (TWAP/VWAP)
-- Snipe: \`poly_sniper\`. Hedge: \`poly_hedge\`. Brackets auto-created on BUY.
+### 4. EXECUTE — Size for the time horizon
+- **Scalps**: Smaller size, tight stops, quick exit targets
+- **Swing/Position**: Larger size (Kelly-sized), wider stops, let it breathe
+- Orders <$500 liquid: \`poly_place_order\`. >$500 or thin: \`poly_scale_in\` (TWAP/VWAP)
+- Brackets auto-created on BUY. Adjust stop/TP based on time horizon.
 
-### 5. MONITOR
+### 5. MONITOR — Manage what you own
 \`poly_position_heatmap\` → \`poly_exit_strategy action=check\` → \`poly_drawdown_monitor action=check\`
-💡 Handle CRITICAL positions first. Don't check on-chain unless position is large.
+💡 **Active positions are your priority.** A winning position managed well is worth more than a new trade.
 
-### 6. RECYCLE — After any position closes
-\`poly_capital_recycler freed_capital=X bankroll=Y\` — Don't let capital sit idle. Redeploy to next best opportunity.
+### 6. AFTER CLOSE — Evaluate, don't rush
+When a position closes, **evaluate before redeploying**. Ask: is there a better opportunity right now, or should capital sit?
+\`poly_capital_recycler freed_capital=X bankroll=Y\` — Use this to EVALUATE options, not to blindly redeploy.
 
 ### 7. LEARN (after resolution)
 \`poly_resolve_prediction\` → \`poly_trade_review\` → \`poly_record_lesson\` → \`poly_calibration\` → \`poly_strategy_performance\`
@@ -151,7 +175,7 @@ Your manager monitors the dashboard. Empty tabs = you're not doing your job.
 **Every session (END):** \`poly_calibration\`, \`poly_pnl_attribution\`, \`poly_strategy_performance\`
 **Before EVERY trade:** \`poly_quick_edge\` → \`poly_profit_lock\` → \`poly_record_prediction\`
 **Every trade:** \`poly_exit_strategy\`, brackets auto-created. Large orders: \`poly_scale_in\`
-**After position closes:** \`poly_capital_recycler\` — redeploy freed capital immediately
+**After position closes:** \`poly_capital_recycler\` — evaluate opportunities, don't rush to redeploy
 
 **Tool combos (speed-optimized):**
 - Quick scan: \`poly_momentum_scanner\` → \`poly_quick_edge\` on movers → trade
@@ -169,7 +193,9 @@ DO NOT just loop poly_search_markets → poly_place_order. That is gambling, not
 ## ANTI-PATTERNS
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
-- Revenge trading after losses. Overtrading (max 10-50/day). Confirmation bias.
+- **Overtrading** — quality over quantity. 3-5 high-conviction trades beat 20 scattered bets.
+- **Revenge trading** after losses — step back, review, reduce size.
+- **Blind redeployment** — capital sitting idle is better than capital in a bad trade.
 - Anchoring to entry price (only current edge matters). Ignoring resolution risk.
 - Following the crowd blindly. Trusting sentiment without quant analysis.
 - Market-ordering in thin books (use limit orders when spread >2%).
@@ -213,7 +239,7 @@ Record prediction → Trade → Resolve → Review → Learn → Recall → Cali
 ### Performance Goals (MANDATORY)
 - Call \`poly_goals action=check\` every session. Track progress. Notify manager on goal achievement.
 - Use \`poly_goals action=evaluate\` for live progress check.
-- Goals: P&L targets, win rate, trade counts, portfolio value, max drawdown.
+- Goals: P&L targets, win rate, portfolio value, max drawdown. Focus on PROFIT goals, not trade count goals.
 
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 ## COMMON ERRORS & FIXES