1

AGENTS.md

@@ -92,11 +92,11 @@ AlphaX 是一个以 `Python + FastAPI + PostgreSQL + Docker + 静态 HTML` 组
 4. `app/services/event_driven_screener.py`
    负责事件/舆情驱动的快速触发检查，是多策略发现层的补充入口。
 5. `app/services/price_streamer.py`
-   负责实时价格缓存，不等同于完整推荐状态跟踪。
+   负责实时价格缓存和策略交易实时执行触发。挂单触价、已开仓 TP/SL、移动止盈、持仓保护这类依赖最新价的动作，应优先由 `price-streamer` 的 websocket tick 驱动；`paper-trader` 调度只能作为补偿任务，不能作为主要成交时钟。
 6. `app/services/price_tracker.py`
    负责可执行推荐的价格跟踪、状态迁移和动态风险提示。
 7. `app/services/paper_trader.py`
-   负责策略交易账本同步和 paper 执行适配。TP/SL、移动止盈、仓位健康、仓位 sizing、账户级风控等可复用交易能力不应长期绑定在 paper trading 层；新增能力优先沉到 `app/core/*` 或独立 execution/risk 模块，再由 paper/live 适配调用。
+   负责策略交易账本补偿同步和 paper 执行适配。它用于处理调度兜底、服务重启后的漏 tick 补账、从可执行推荐创建新持仓/挂单、以及同步 live protection；不要依赖它的 180 秒轮询来完成挂单成交。TP/SL、移动止盈、仓位健康、仓位 sizing、账户级风控等可复用交易能力不应长期绑定在 paper trading 层；新增能力优先沉到 `app/core/*` 或独立 execution/risk 模块，再由 paper/live 适配调用。
 8. `app/db/live_trading.py` / `app/web/routes_live_trading.py`
    负责实盘控制台：多交易所/多 API 账户配置、账号级风控、交易所接口验收和执行审计事件。页面不再使用“订单意图”作为产品概念，也不区分 Demo/正式环境，实际环境由 endpoint/API key 配置决定。
    实盘控制台页面默认只读取 PostgreSQL 中的账户快照，不应在首屏加载时直接阻塞调用交易所 API。`live-trading-sync` 调度任务负责定时同步余额、持仓、挂单、订单历史到 `live_account_snapshots`，并按配置把策略交易 open 仓位同步到实盘账户；手动“立即同步”只是强制刷新同一份 DB 快照。

app/db/scheduler_db.py

@@ -48,7 +48,7 @@ DEFAULT_JOBS = [
         "every_seconds": 180,
         "initial_delay": 30,
         "lock_group": "paper_trading_write",
-        "description": "策略交易账本同步",
+        "description": "策略交易补偿同步（实时成交由 price-streamer 驱动）",
         "sort_order": 25,
     },
     {

app/services/price_streamer.py

@@ -109,6 +109,42 @@ def _load_pending_paper_order_recs() -> list[dict]:
         conn.close()
 
 
+def _load_pending_paper_order_rec(symbol: str) -> dict | None:
+    symbol = str(symbol or "").strip().upper()
+    if not symbol:
+        return None
+    conn = get_conn()
+    try:
+        row = conn.execute(
+            """
+            SELECT
+                po.recommendation_id AS id,
+                po.symbol,
+                po.side,
+                po.target_price AS entry_price,
+                po.stop_loss,
+                po.tp1,
+                po.tp2,
+                po.source_status AS execution_status,
+                po.source_action AS action_status,
+                po.strategy_version,
+                po.strategy_code,
+                po.strategy_signal_id,
+                po.strategy_snapshot_json,
+                po.factor_roles_json,
+                po.entry_plan_snapshot_json AS entry_plan_json
+            FROM paper_orders po
+            WHERE po.status='pending' AND po.symbol=%s
+            ORDER BY po.created_at ASC, po.id ASC
+            LIMIT 1
+            """,
+            (symbol,),
+        ).fetchone()
+        return dict(row) if row else None
+    finally:
+        conn.close()
+
+
 def load_stream_targets(limit: int | None = None, cfg: dict | None = None) -> dict[str, dict]:
     """Return symbol -> recommendation-like payload for websocket updates."""
     cfg = cfg or price_streamer_config()
@@ -146,7 +182,10 @@ def handle_price_tick(symbol: str, price: float, targets: dict[str, dict], event
         update_latest_price_cache(symbol, price, updated_at=event_time, source="price_streamer")
     if not cfg.get("sync_paper_trading", True):
         return {"updated_price": True, "paper_trading": {"skipped": True, "reason": "disabled_by_streamer"}}
-    rec = targets.get(symbol)
+    # Pending limit orders are executable state. Prefer the order snapshot over
+    # any stale in-memory recommendation target so a websocket tick can fill the
+    # order immediately when the price touches.
+    rec = _load_pending_paper_order_rec(symbol) or targets.get(symbol)
     if not rec:
         return {"updated_price": True, "paper_trading": {"skipped": True, "reason": "no_target"}}
     result = sync_recommendation(rec, price, event_time=event_time)

tests/test_price_streamer.py

@@ -175,6 +175,44 @@ def test_price_streamer_fills_pending_paper_order():
     assert list_paper_trades()["items"][0]["entry_price"] == pytest.approx(95)
 
 
+def test_price_streamer_tick_fills_pending_order_even_when_targets_are_stale():
+    set_config("system", "paper_trading", _paper_config())
+    set_config("system", "price_streamer", {
+        "enabled": True,
+        "update_latest_price_cache": True,
+        "sync_paper_trading": True,
+        "include_actionable_recommendations": True,
+        "include_open_paper_trades": True,
+    })
+    altcoin_db.init_db()
+    rec_id = altcoin_db.create_recommendation(
+        symbol="STALEPBO/USDT",
+        rec_state="蓄力",
+        rec_score=22,
+        entry_price=95,
+        stop_loss=90,
+        tp1=105,
+        signals=["等待回踩"],
+        entry_plan={"entry_action": "等回踩", "entry_price": 95, "stop_loss": 90, "tp1": 105, "risk_reward_ok": True, "rr1": 2.0},
+    )
+    with altcoin_db.get_conn() as conn:
+        conn.execute(
+            "UPDATE recommendation SET execution_status='wait_pullback', action_status='等回踩', display_bucket='watch_pool' WHERE id=%s",
+            (rec_id,),
+        )
+        conn.commit()
+    rec = {"id": rec_id, "symbol": "STALEPBO/USDT", "execution_status": "wait_pullback", "action_status": "等回踩", "entry_price": 95, "stop_loss": 90, "tp1": 105, "entry_plan": {"entry_action": "等回踩", "entry_price": 95, "stop_loss": 90, "tp1": 105, "risk_reward_ok": True, "rr1": 2.0}}
+
+    created = price_streamer.handle_price_tick("STALEPBO/USDT", 100, {"STALEPBO/USDT": rec}, event_time="2026-05-16T10:00:00")
+    filled = price_streamer.handle_price_tick("STALEPBO/USDT", 94.9, {}, event_time="2026-05-16T10:05:00")
+
+    assert created["paper_trading"]["reason"] == "paper_order_created"
+    assert filled["paper_trading"]["opened"] is True
+    assert filled["paper_trading"]["paper_order"]["filled"] is True
+    assert list_paper_orders(status="filled")["total"] == 1
+    assert list_paper_trades()["items"][0]["symbol"] == "STALEPBO/USDT"
+
+
 def test_price_streamer_prioritizes_pending_order_over_same_symbol_recommendation():
     set_config("system", "paper_trading", _paper_config())
     set_config("system", "price_streamer", {