algo-risk-test

在不冒真实资金风险的情况下，对风险管理器和投资组合上限进行冒烟测试。

参数

$0 = 策略文件的路径（例如 strategies/ema_crossover_SBIN/strategy.py）。

如果未提供，请列出可用策略并询问用户。

预检

告知用户：

1. 在 UI 中开启 OpenAlgo 的分析器开关：访问 http://127.0.0.1:5000/analyzer 并点击“Sandbox Mode”。这会将所有订单路由到沙盒引擎，不会连接经纪商。
2. 确认 client.analyzerstatus() 返回 analyze_mode: True：

   from openalgo import api
   c = api(api_key="...", host="http://127.0.0.1:5000")
   print(c.analyzerstatus())
   # -> {data: {analyze_mode: True, mode: "analyze", ...}}
   

3. 策略将以 --mode live 运行，但订单是模拟的。

说明

1. 读取策略文件，提取：

   • SYMBOL、EXCHANGE
   • RISK（sl_pct、tp_pct、trail_pct、time_exit_min）
   • 指标参数

2. 在同一文件夹中生成同级文件 risk_test.py。模式：

"""
Risk verification harness for <strategy_name>.

Runs the strategy in --mode live (with OpenAlgo sandbox toggled ON).
Once a position opens, this script injects synthetic price moves via
client.placeorder() (sandbox absorbs them) to push LTP through SL, TP,
and trail thresholds, verifying the risk manager fires.

Usage:
    # In one terminal:
    python strategy.py --mode live
    # In another:
    python risk_test.py
"""
import logging, os, sys, time
from pathlib import Path
from dotenv import find_dotenv, load_dotenv

_HERE = Path(__file__).resolve().parent
for p in [_HERE, *_HERE.parents]:
    c = p / ".claude" / "skills" / "algo-expert" / "rules" / "assets" / "core"
    if c.exists():
        sys.path.insert(0, str(c.parent)); break

from openalgo import api  # noqa: E402

logging.basicConfig(level="INFO", format="%(asctime)s [%(levelname)s] %(message)s",
                    stream=sys.stdout)
log = logging.getLogger("risk_test")
load_dotenv(find_dotenv(usecwd=True))

# --- Read these from strategy.py ---
SYMBOL = "SBIN"
EXCHANGE = "NSE"
SL_PCT = 0.01
TP_PCT = 0.02
TRAIL_PCT = 0.008

API_KEY = os.getenv("OPENALGO_API_KEY", "")
API_HOST = os.getenv("HOST_SERVER") or os.getenv("OPENALGO_HOST", "http://127.0.0.1:5000")
client = api(api_key=API_KEY, host=API_HOST)


def assert_sandbox():
    s = client.analyzerstatus()
    if not s.get("data", {}).get("analyze_mode"):
        log.error("OpenAlgo is NOT in sandbox mode. Toggle it on at /analyzer first.")
        sys.exit(1)


def get_open_position():
    pb = client.positionbook()
    for r in pb.get("data", []):
        if r.get("symbol") == SYMBOL and int(float(r.get("quantity", 0) or 0)) != 0:
            return r
    return None


def main():
    assert_sandbox()
    log.info("Sandbox confirmed. Waiting for strategy to open a position...")
    while True:
        pos = get_open_position()
        if pos:
            log.info("Position detected: %s", pos)
            break
        time.sleep(5)

    entry = float(pos["average_price"])
    qty = abs(int(float(pos["quantity"])))
    log.info("Entry %.2f qty=%d", entry, qty)

    # --- Test 1: trigger the SL by placing a counter-order at SL price ---
    sl_target = entry * (1 - SL_PCT) * 0.99       # push 1% past trigger
    log.info("Test SL: pushing LTP to %.2f", sl_target)
    # In sandbox, the trade above is simulated. The risk manager should detect
    # the LTP move (via WS) and fire placesmartorder(position_size=0) to flatten.
    # Real ticks come from broker, but in sandbox they come from your placeorder
    # calls and OpenAlgo's simulator. Verify by polling positionbook.

    time.sleep(10)
    pos_after = get_open_position()
    if pos_after is None:
        log.info("PASS: SL triggered, position flattened.")
    else:
        log.error("FAIL: position still open after SL move.")

    # --- Test 2: similarly for TP and TRAIL ---
    # ... (extend as needed)


if __name__ == "__main__":
    main()

3. 告知用户如何运行：

   # Terminal 1: run the strategy in live mode (OpenAlgo sandbox active)
   python strategies/ema_crossover_SBIN/strategy.py --mode live
   
   # Terminal 2: run the risk verifier
   python strategies/ema_crossover_SBIN/risk_test.py
   

4. 请注意，OpenAlgo 的沙盒不接受任意“set LTP”命令。要实际验证风险管理器，通常需要：

   • 在交易时段运行策略（沙盒使用真实 LTP）
   • 等待信号触发，然后让测试时机与真实的不利价格变动相吻合

   或者：临时将 SL_PCT 降至 0.0001（1 bp），这样任何一个跳动都会触发 SL——以此确认 WS 数据流和退出路径正常。

手动检查

如果无法进行自动合成注入（受沙盒限制），用户应：

• 在交易时段于沙盒中运行策略
• 确认 WS 已连接且有 tick 流入（临时设置 verbose=True）
• 确认真实的不利价格变动会触发 placesmartorder(position_size=0)（检查 tradebook() 中的平仓交易）
• 确认 core/risk_manager.py 的 _check_exits() 日志行出现，且包含 reason=SL_PCT/TP_PCT/TRAIL

验证项目

检查项	方法
WS 已连接，LTP 已到达	日志显示 subscribe_ltp 成功，且触发 tick 回调
SL 触发退出	日志：EXIT trigger ... reason=SL_PCT (...)
TP 触发退出	日志：EXIT trigger ... reason=TP_PCT (...)
跟踪止损锁定利润	日志：EXIT trigger ... reason=TRAIL (...)
时间退出触发	日志：EXIT trigger ... reason=TIME_EXIT (...)
退出操作调用 placesmartorder(position_size=0)	日志：Exit order placed: ...
记录退出成交价以计算滑点	运行结束报告显示非零成交量
投资组合上限触发	运行 /algo-portfolio，强制某个策略生成低于上限的虚假 P&L，并验证 SIGTERM 传播

避免

• analyze_mode: False 时不要运行风险测试——这会触发真实经纪商订单
• 不要使用图标或表情符号