summaryrefslogtreecommitdiff
path: root/docs/superpowers/specs/2026-04-01-operations-and-strategy-expansion-design.md
blob: e1aea742058ad6c9ba82712b24260e955bcd0730 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
# Operations Infrastructure & Strategy Expansion — Design Spec

## Overview

기존 Binance 현물 암호화폐 자동매매 플랫폼의 두 가지 영역을 강화한다:

1. **운영 인프라** — DB 마이그레이션, 구조화된 로깅, Telegram 알림, 에러 복구, 메트릭 수집
2. **전략 확장** — 추세 추종/스캘핑 전략 추가, 백테스트 고도화

접근 순서: 운영 인프라 먼저 완성 → 전략 추가. 안정적인 모니터링/알림 기반 위에서 새 전략을 검증할 수 있어야 한다.

---

## Part 1: Operations Infrastructure

### 1.1 DB Layer Migration (asyncpg → SQLAlchemy 2.0 Async + Alembic)

**목표:** raw SQL과 asyncpg 직접 사용을 SQLAlchemy 2.0 async ORM으로 교체하고, Alembic으로 마이그레이션을 관리한다.

**변경 사항:**

- `shared/src/shared/db.py` — AsyncSession 기반으로 재작성
  - `create_async_engine()` + `async_sessionmaker()` 사용
  - asyncpg는 SQLAlchemy의 내부 드라이버로 유지 (`postgresql+asyncpg://`)
  - 기존 raw SQL 함수들을 ORM 쿼리로 전환

- `shared/src/shared/sa_models.py` — SQLAlchemy ORM 모델 (신규)
  - 기존 Pydantic 모델(models.py)과 1:1 매핑되는 SA 테이블 정의
  - `Candle`, `Signal`, `Order`, `Trade`, `Position`, `PortfolioSnapshot` 테이블
  - Pydantic 모델은 이벤트 직렬화/API 전용으로 유지

- `shared/alembic/` — Alembic 마이그레이션 환경 (신규)
  - `alembic.ini` — 설정 파일 (DATABASE_URL 참조)
  - `env.py` — async 엔진 설정, SA 모델 메타데이터 참조
  - `versions/` — 마이그레이션 파일들
  - 초기 마이그레이션: 기존 `db.py`의 CREATE TABLE 로직을 마이그레이션으로 이전

- `Makefile` 타겟 추가:
  - `make migrate` — `alembic upgrade head`
  - `make migrate-down` — `alembic downgrade -1`
  - `make migrate-new MSG="description"` — `alembic revision --autogenerate -m "description"`

- 각 서비스의 DB 접근 코드를 AsyncSession 기반으로 업데이트:
  - `data-collector/storage.py` — bulk insert 쿼리를 SA ORM으로
  - `order-executor/executor.py` — order CRUD를 SA ORM으로
  - `portfolio-manager/portfolio.py` — position/snapshot 쿼리를 SA ORM으로
  - `backtester/engine.py` — candle 조회를 SA ORM으로

**의존성 추가:** `sqlalchemy[asyncio]>=2.0`, `alembic>=1.13`
**의존성 제거:** `asyncpg` (직접 의존 → SQLAlchemy 내부 의존으로 변경)

---

### 1.2 Structured Logging (structlog)

**목표:** 전 서비스에 JSON 구조화 로깅을 적용하고, 에러 로그를 Telegram 알림과 연결한다.

**변경 사항:**

- `shared/src/shared/logging.py` (신규)
  - `setup_logging(service_name: str, log_level: str)` 함수
  - structlog 프로세서 체인: timestamp, log level, service_name 바인딩, JSON 렌더러
  - 개발 환경: 컬러 콘솔 출력 / 프로덕션: JSON stdout
  - `LOG_FORMAT` 환경변수로 전환 (`console` | `json`, 기본값: `json`)

- 각 서비스 `main.py`에서 `setup_logging()` 호출
- 기존 `logging.getLogger()` 호출을 `structlog.get_logger()` 로 교체
- 컨텍스트 바인딩 예시:
  ```python
  log = structlog.get_logger().bind(service="strategy-engine", symbol="BTCUSDT")
  log.info("signal_generated", strategy="rsi", side="BUY", price=68500)
  ```

- ERROR 이상 로그 → Telegram 알림 트리거 (1.3절 TelegramNotifier 연동)
  - structlog 커스텀 프로세서로 구현
  - 알림 전송 실패 시 로그만 남기고 서비스 중단하지 않음

**의존성 추가:** `structlog>=24.0`

---

### 1.3 Telegram Notification Service

**목표:** 주요 이벤트(시그널, 주문, 에러, 일일 요약)를 Telegram으로 전송한다.

**변경 사항:**

- `shared/src/shared/notifier.py` (신규)
  - `TelegramNotifier` 클래스
    - `__init__(bot_token: str, chat_id: str)` — aiohttp 세션 관리
    - `send(message: str, parse_mode: str = "HTML")` — 메시지 전송
    - `send_signal(signal: Signal)` — 시그널 포맷팅 후 전송
    - `send_order(order: Order)` — 주문 체결/실패 알림
    - `send_error(error: str, service: str)` — 에러 알림
    - `send_daily_summary(positions: list, pnl: Decimal)` — 일일 요약
  - Rate limiting: 초당 최대 1건 (asyncio.Semaphore + 큐)
  - 연결 실패 시 최대 3회 재시도, 실패해도 서비스 중단하지 않음

- `shared/src/shared/config.py` — 설정 추가:
  - `TELEGRAM_BOT_TOKEN: str = ""`
  - `TELEGRAM_CHAT_ID: str = ""`
  - `TELEGRAM_ENABLED: bool = False` (토큰 미설정 시 자동 비활성)

- `.env.example` 업데이트:
  ```
  TELEGRAM_BOT_TOKEN=
  TELEGRAM_CHAT_ID=
  TELEGRAM_ENABLED=false
  ```

- 연동 포인트:
  - `strategy-engine/engine.py` — 시그널 생성 시 `send_signal()`
  - `order-executor/executor.py` — 주문 체결/실패 시 `send_order()`
  - `shared/logging.py` — ERROR 로그 시 `send_error()`
  - `portfolio-manager/main.py` — 매일 자정(UTC) `send_daily_summary()`

**의존성:** aiohttp (이미 존재)

---

### 1.4 Error Recovery & Health Checks

**목표:** 서비스 장애 시 자동 복구하고, 헬스체크 엔드포인트로 상태를 모니터링한다.

**변경 사항:**

- `shared/src/shared/resilience.py` (신규)
  - `retry_with_backoff(func, max_retries, base_delay)` — exponential backoff 데코레이터
    - 지터(jitter) 포함: `delay * (1 + random(0, 0.5))`
    - 최대 지연: 60초
  - `CircuitBreaker` 클래스:
    - 상태: CLOSED(정상) → OPEN(차단) → HALF_OPEN(시험)
    - `failure_threshold`: 연속 실패 N회 시 OPEN (기본: 5)
    - `recovery_timeout`: OPEN 후 N초 뒤 HALF_OPEN (기본: 60)
    - OPEN 전환 시 Telegram 알림 전송

- `shared/src/shared/healthcheck.py` (신규)
  - `HealthCheckServer` — aiohttp 기반 경량 HTTP 서버
    - `GET /health` → `{"status": "ok", "service": "...", "uptime": ..., "checks": {...}}`
    - 체크 항목: Redis 연결, PostgreSQL 연결, Binance WS 연결(해당 서비스만)
    - 포트: `HEALTH_PORT` 환경변수 (서비스별 다르게 설정)

- 각 서비스에 적용:
  - `data-collector` — Binance WS 재연결 (backoff), Redis/DB 재연결
  - `strategy-engine` — Redis 소비자 재연결
  - `order-executor` — 거래소 API 호출 재시도 (circuit breaker)
  - `portfolio-manager` — Redis/DB 재연결

- `docker-compose.yml` — healthcheck를 `/health` 엔드포인트로 변경

- `shared/src/shared/config.py` — 설정 추가:
  - `HEALTH_PORT: int = 8080`
  - `CIRCUIT_BREAKER_THRESHOLD: int = 5`
  - `CIRCUIT_BREAKER_TIMEOUT: int = 60`

---

### 1.5 Prometheus Metrics

**목표:** 각 서비스의 주요 지표를 Prometheus 포맷으로 노출한다.

**변경 사항:**

- `shared/src/shared/metrics.py` (신규)
  - `MetricsServer` — prometheus_client 기반
    - `/metrics` 엔드포인트 (healthcheck 서버에 통합)
  - 공통 메트릭:
    - `service_up` (Gauge) — 서비스 상태
    - `errors_total` (Counter) — 에러 횟수 (label: service, error_type)
    - `event_processing_seconds` (Histogram) — 이벤트 처리 시간

- 서비스별 메트릭:
  - **data-collector:**
    - `candles_received_total` (Counter) — 수신 캔들 수
    - `ws_reconnections_total` (Counter) — WS 재연결 횟수
  - **strategy-engine:**
    - `signals_generated_total` (Counter, label: strategy, side)
    - `strategy_execution_seconds` (Histogram, label: strategy)
  - **order-executor:**
    - `orders_total` (Counter, label: status, side)
    - `risk_rejections_total` (Counter, label: reason)
  - **portfolio-manager:**
    - `portfolio_value` (Gauge) — 총 포트폴리오 가치
    - `unrealized_pnl` (Gauge, label: symbol)

- `docker-compose.yml` — Prometheus + Grafana 서비스 추가 (선택적 프로필):
  ```yaml
  prometheus:
    image: prom/prometheus:latest
    profiles: ["monitoring"]
    volumes:
      - ./monitoring/prometheus.yml:/etc/prometheus/prometheus.yml
  grafana:
    image: grafana/grafana:latest
    profiles: ["monitoring"]
    ports:
      - "3000:3000"
  ```

- `monitoring/prometheus.yml` (신규) — 스크래핑 설정
- `monitoring/grafana/` (신규) — 대시보드 프로비저닝 (선택적)

**의존성 추가:** `prometheus-client>=0.20`

---

## Part 2: Strategy Expansion

### 2.1 Trend Following Strategies

**MACD Strategy** (`strategies/macd_strategy.py`)
- MACD line = EMA(12) - EMA(26), Signal line = EMA(9) of MACD
- BUY: MACD가 Signal line 위로 교차 + 히스토그램 양전환
- SELL: MACD가 Signal line 아래로 교차 + 히스토그램 음전환
- 파라미터: `fast_period=12`, `slow_period=26`, `signal_period=9`, `quantity`
- warmup_period: `slow_period + signal_period`

**Bollinger Bands Strategy** (`strategies/bollinger_strategy.py`)
- 중심선 = SMA(20), 상단 = 중심 + 2*std, 하단 = 중심 - 2*std
- BUY: 가격이 하단 밴드 아래로 이탈 후 복귀
- SELL: 가격이 상단 밴드 위로 이탈 후 복귀
- 변동성 필터: 밴드 폭이 임계값 미만이면 시그널 무시 (횡보장 필터)
- 파라미터: `period=20`, `num_std=2.0`, `min_bandwidth=0.02`, `quantity`
- warmup_period: `period`

**EMA Crossover Strategy** (`strategies/ema_crossover_strategy.py`)
- 단기 EMA와 장기 EMA 교차
- BUY: 단기 EMA가 장기 EMA 위로 교차 (Golden Cross)
- SELL: 단기 EMA가 장기 EMA 아래로 교차 (Death Cross)
- 파라미터: `short_period=9`, `long_period=21`, `quantity`
- warmup_period: `long_period`

---

### 2.2 Scalping Strategies

**VWAP Strategy** (`strategies/vwap_strategy.py`)
- VWAP = cumsum(price * volume) / cumsum(volume)
- BUY: 가격이 VWAP 아래에서 VWAP으로 복귀 (평균 회귀)
- SELL: 가격이 VWAP 위에서 VWAP으로 복귀
- 일중 리셋: UTC 00:00에 VWAP 재계산
- 파라미터: `deviation_threshold=0.002`, `quantity`
- warmup_period: 최소 30 캔들

**Volume Profile Strategy** (`strategies/volume_profile_strategy.py`)
- 최근 N개 캔들의 가격대별 거래량 분포 계산
- POC (Point of Control): 가장 거래량이 많은 가격대
- Value Area: 전체 거래량 70%가 집중된 구간
- BUY: 가격이 Value Area 하단 지지선에서 반등
- SELL: 가격이 Value Area 상단 저항선에서 거부
- 파라미터: `lookback_period=100`, `num_bins=50`, `value_area_pct=0.7`, `quantity`
- warmup_period: `lookback_period`

---

### 2.3 Strategy Common Improvements

**BaseStrategy 확장:**
```python
class BaseStrategy(ABC):
    @property
    @abstractmethod
    def warmup_period(self) -> int:
        """지표 계산에 필요한 최소 캔들 수"""
        pass

    @abstractmethod
    def on_candle(self, candle: Candle) -> Signal | None:
        pass

    @abstractmethod
    def configure(self, params: dict) -> None:
        pass

    def reset(self) -> None:
        """전략 상태 초기화 (백테스트 간 재사용)"""
        pass
```

**전략 파라미터 외부화:**
- `strategies/config/` 디렉토리에 YAML 설정 파일
- 파일명: `{strategy_name}.yaml` (예: `rsi_strategy.yaml`)
- 구조:
  ```yaml
  # rsi_strategy.yaml
  period: 14
  oversold: 30
  overbought: 70
  quantity: 0.001
  ```
- `plugin_loader.py`가 전략 로드 시 자동으로 같은 이름의 YAML을 찾아 `configure()` 호출
- CLI에서 `--param key=value`로 런타임 오버라이드 가능

**기존 전략 업데이트:**
- `RsiStrategy`, `GridStrategy`에 `warmup_period` 속성 추가
- `reset()` 메서드 구현

**의존성 추가:** `pyyaml>=6.0`

---

### 2.4 Backtest Enhancement

**DetailedMetrics 데이터클래스** (`backtester/src/backtester/metrics.py`, 신규):
```python
@dataclass
class TradeRecord:
    entry_time: datetime
    exit_time: datetime
    symbol: str
    side: str
    entry_price: Decimal
    exit_price: Decimal
    quantity: Decimal
    pnl: Decimal
    pnl_pct: float
    holding_period: timedelta

@dataclass
class DetailedMetrics:
    # 기본
    total_return: float
    total_trades: int
    winning_trades: int
    losing_trades: int
    win_rate: float
    profit_factor: float

    # 리스크 메트릭
    sharpe_ratio: float
    sortino_ratio: float
    calmar_ratio: float
    max_drawdown: float
    max_drawdown_duration: timedelta

    # 수익률 분석
    monthly_returns: dict[str, float]  # "2025-01": 0.05
    avg_win: float
    avg_loss: float
    largest_win: float
    largest_loss: float
    avg_holding_period: timedelta

    # 개별 거래
    trades: list[TradeRecord]
```

**BacktestEngine 확장:**
- `engine.py`에 `DetailedMetrics` 계산 로직 추가
- `simulator.py`에 `TradeRecord` 생성 로직 추가 (진입/청산 시점 기록)
- Sharpe ratio = `mean(daily_returns) / std(daily_returns) * sqrt(365)` (crypto는 365일)
- Sortino ratio = `mean(daily_returns) / downside_std * sqrt(365)`
- Calmar ratio = `annualized_return / max_drawdown`
- Max drawdown = `max(peak - trough) / peak`

**Reporter 개선:**
- `reporter.py` — rich 라이브러리로 테이블 출력
  - 요약 테이블: 핵심 메트릭
  - 월별 수익률 테이블
  - 최고/최악 거래 Top 5
- CSV/JSON 내보내기: `--output csv` / `--output json` 플래그

**CLI 확장:**
- `trading backtest run` — 기존 출력에 상세 메트릭 추가
- `trading backtest run --output csv --file result.csv` — 결과 내보내기

**의존성 추가:** `rich>=13.0`

---

## Updated Tech Stack

| 용도 | 기존 | 변경 |
|------|------|------|
| DB ORM | asyncpg (raw SQL) | **SQLAlchemy 2.0 async** (asyncpg 드라이버) |
| 마이그레이션 | 없음 | **Alembic** |
| 로깅 | Python logging | **structlog** |
| 알림 | 없음 | **Telegram Bot API** (aiohttp) |
| 메트릭 | 없음 | **prometheus-client** |
| 전략 설정 | 하드코딩 | **YAML** (pyyaml) |
| 리포트 출력 | print | **rich** |

---

## Updated .env.example

```env
# Exchange
BINANCE_API_KEY=
BINANCE_API_SECRET=

# Infrastructure
REDIS_URL=redis://localhost:6379
DATABASE_URL=postgresql+asyncpg://trading:trading@localhost:5432/trading

# Logging
LOG_LEVEL=INFO
LOG_FORMAT=json

# Telegram
TELEGRAM_BOT_TOKEN=
TELEGRAM_CHAT_ID=
TELEGRAM_ENABLED=false

# Risk Management
RISK_MAX_POSITION_SIZE=0.1
RISK_STOP_LOSS_PCT=5
RISK_DAILY_LOSS_LIMIT_PCT=10
DRY_RUN=true

# Health & Metrics
HEALTH_PORT=8080
CIRCUIT_BREAKER_THRESHOLD=5
CIRCUIT_BREAKER_TIMEOUT=60
```

---

## New Files Summary

| 파일 | 용도 |
|------|------|
| `shared/src/shared/sa_models.py` | SQLAlchemy ORM 모델 |
| `shared/src/shared/logging.py` | structlog 설정 |
| `shared/src/shared/notifier.py` | Telegram 알림 |
| `shared/src/shared/resilience.py` | retry, circuit breaker |
| `shared/src/shared/healthcheck.py` | 헬스체크 서버 |
| `shared/src/shared/metrics.py` | Prometheus 메트릭 |
| `shared/alembic/` | DB 마이그레이션 환경 |
| `strategies/config/*.yaml` | 전략 파라미터 설정 |
| `strategies/macd_strategy.py` | MACD 전략 |
| `strategies/bollinger_strategy.py` | Bollinger Bands 전략 |
| `strategies/ema_crossover_strategy.py` | EMA Crossover 전략 |
| `strategies/vwap_strategy.py` | VWAP 전략 |
| `strategies/volume_profile_strategy.py` | Volume Profile 전략 |
| `backtester/src/backtester/metrics.py` | 상세 백테스트 메트릭 |
| `monitoring/prometheus.yml` | Prometheus 설정 |

---

## Scope Boundaries

**포함:**
- SQLAlchemy 2.0 async 전환 + Alembic 마이그레이션
- structlog JSON 로깅
- Telegram 알림 (시그널, 주문, 에러, 일일 요약)
- 에러 복구 (retry, circuit breaker) + 헬스체크
- Prometheus 메트릭 수집
- 5개 신규 전략 (MACD, Bollinger, EMA Crossover, VWAP, Volume Profile)
- BaseStrategy에 warmup_period, reset() 추가
- YAML 기반 전략 파라미터
- 백테스트 상세 메트릭 + rich 리포트

**제외:**
- Grafana 대시보드 프로비저닝 (Prometheus만 설정, 대시보드는 수동)
- 멀티 거래소 지원
- REST API / 웹 대시보드
- 전략 조합 프레임워크 (향후 확장)