低波筛选

• ✓价格围绕均值往返时低买高卖，吃区间波动
• ⚠单边趋势中逆势加仓，回撤风险大

合成多标的池横截面选股模拟（8 个标的）——非未来收益承诺。

"""低波筛选（Low-volatility selection）—— 前置声明式筛选器 + 每标的趋势跟随。

思路：先用筛选器从标的池里选出实现波动率最低的 top_n 个（低波因子/风险平价的选股侧），再对每个选中的
标的各自做 EMA 快慢线趋势跟随。演示"把选股作为策略前置"的用法：子类 ScreenedStrategy，框架每
rebalance_every 步重排、退出跌出榜的标的，on_symbol 只管单标的逻辑。

适用性：instrument="perp"、venues=["HYPERLIQUID"]。回测需多标的 universe feed（source="universe"）。
"""
from __future__ import annotations

from ..base import StrategyBase
from ..context import StrategyContext
from ..indicators import ema
from ..registry import register
from ..screener import Screener, ScreenedStrategy


@register("低波筛选")
class LowVolSelect(ScreenedStrategy, StrategyBase):
    description = "低波筛选（前置）：选实现波动最低的 top_n 标的，各自做 EMA 趋势跟随。"
    params = {"lookback": 30, "top_n": 2, "rebalance": 20, "fast": 10, "slow": 30, "slice": 8000.0}
    venues = ["HYPERLIQUID"]
    symbols: list = []
    instrument = "perp"

    def __init__(self, lookback: int = 30, top_n: int = 2, rebalance: int = 20,
                 fast: int = 10, slow: int = 30, slice: float = 8000.0) -> None:
        self.screener = Screener("volatility", lookback=int(lookback), top_n=int(top_n),
                                 direction="bottom")
        self.rebalance_every = int(rebalance)
        self.fast = int(fast)
        self.slow = int(slow)
        self.slice = float(slice)

    async def on_symbol(self, ctx: StrategyContext, sym: str) -> None:
        h = await ctx.history(sym, self.slow + 2)
        f = ema(h, self.fast)
        s = ema(h, self.slow)
        if f is None or s is None:
            return
        px = await ctx.price(sym)
        if px <= 0:
            return
        qty = self.slice / px
        await ctx.target(sym, qty if f > s else 0.0, band=qty * 0.1)  # 快线上方持有、否则离场

完整 SDK 与全部策略以 MIT 协议开源，可直接回测、paper 试跑或二次开发。