# DRAFT · WEEKLY-PROFIT-KPI-DASHBOARD

Status: planned (P2)
Quelle: Operator-Backlog-Bullet „13 Operator-Metriken als Dashboard-Plan"
Scope: Plan-Entwurf, kein Code.

---

## 1. Kontext

Operator braucht einen Wochen-Blick auf Performance / Risk / Stability.
Heute existieren nur Einzel-Resources (TradeLog-Liste, PositionSnapshot)
und das tägliche `daily_performance.csv`. Eine konsolidierte
Wochen-Sicht fehlt.

PLAN_T1_ROADMAP_2_WEEKS legt das Profit-Ziel auf realistisch 2-4%/Woche
mit Aspiration 10%. Ohne dashboard ist Operator gezwungen, Zahlen
manuell zu joinen.

---

## 2. Die 13 KPIs (Vorschlag, Operator-Confirm-Pflicht)

| # | KPI | Quelle | Formel | Target |
|---|---|---|---|---|
| 1 | Wochen-PnL absolut (USDT) | trade_logs | SUM(pnl) WHERE closed_at in [w_start, w_end] | > 0 |
| 2 | Wochen-PnL relativ (%) | (1) / starting_capital_w_start | | 2-4% |
| 3 | Anzahl Trades | trade_logs | COUNT | observability |
| 4 | Win-Rate (%) | trade_logs | COUNT(pnl>0) / COUNT | > 50% |
| 5 | Avg Win / Avg Loss (USDT) | trade_logs | AVG(pnl WHERE pnl>0) / AVG(pnl WHERE pnl<0) | > 1.5 |
| 6 | Max Drawdown (USDT) | live_portfolio history + closed_trades | high-water-mark Equity − Trough | < 5% |
| 7 | Trades mit DCA-Rescue (count + %) | dca_log historical + trade_logs.metadata | COUNT(WHERE dca_count>0) | < 30% |
| 8 | SL-Exit-Rate (%) | trade_logs | COUNT(reason LIKE '%stop%') / COUNT | < 50% |
| 9 | TP-Exit-Rate (%) | trade_logs | COUNT(reason LIKE '%profit%') | observability |
| 10 | Sell-Failure-Rate (insufficient_funds) | bot_stdout.log derived + (zukünftig) bot_statuses | COUNT failed-sells / COUNT sell-attempts | < 1% (P0-A) |
| 11 | Avg Position-Age bis Close (h) | trade_logs.opened_at − closed_at | AVG | informational |
| 12 | Bot-Uptime (%) | bot_watchdog logs / bot_statuses | 1 − (downtime / 168h) | > 99% |
| 13 | T2 / T3 Aktivität | trade_logs.strategy_group | COUNT WHERE NOT 't1_core' | 0 (T2/T3 disabled) |

---

## 3. Page-Skizze

```
+---------------------------------------------------------------+
| Wochen-KPIs · KW 20 (2026-05-12 → 2026-05-18)                |
|                                                               |
| Mode-Banner: EXCHANGE_TESTNET                                 |
|                                                               |
| +------------+ +------------+ +------------+ +------------+   |
| | PnL %      | | Win-Rate   | | DD max     | | Sell-Fail  |   |
| | -1.2%      | | 33%        | | -3.4%      | | 12.5% ⚠️    |   |
| | Target 2-4%| | Target 50% | | Target <5% | | Target <1% |   |
| +------------+ +------------+ +------------+ +------------+   |
|                                                               |
| Detail-Tabelle: alle 13 KPIs mit week-over-week-Delta         |
|                                                               |
| Bottom: Liste der Trades dieser Woche                         |
+---------------------------------------------------------------+
```

---

## 4. Boundaries

- **0× neue DB-Tabellen.** Alles aus existierenden Quellen
  (trade_logs, position_snapshots, bot_statuses, live_portfolio.json).
- **0× Backfill-Migration.** KPIs sind Time-Window-Aggregates auf
  bestehenden Daten.
- **0× Bot-Touch.**
- **0× Mainnet.**
- KPI #10 (Sell-Failure-Rate) ist die Operator-eigene Frühwarnung für
  P0-A-Wiederholung. Implementierungsweg:
  - kurzfristig: aus bot_statuses (sobald `_track_sell_failure` einen
    Status emittiert) — siehe P0-B
  - alternative: Log-Parser-Cron mit Output in eigene Tabelle
    (eigener Plan, eigener Cutover)

---

## 5. Phasen-Vorschlag

| Phase | Inhalt | Operator-GO |
|---|---|---|
| W-A | Filament-Page Skelett + KPI 1-5 (alle aus trade_logs) | nein, GUI-only |
| W-B | KPI 6 Max Drawdown (live_portfolio history) | nein |
| W-C | KPI 7-9 Reason-Buckets, KPI 11 Age | nein |
| W-D | KPI 10 Sell-Failure-Rate (depends on bot_statuses emit) | ja, hängt an P0-B |
| W-E | KPI 12-13, Operator-Verify | nein |

---

## 6. STOP

Kein Code vor Operator-GO.