MoFin/scripts/chip_factors.py

#!/usr/bin/env python3
"""chip_factors.py — 筹码因子计算模块

基于中信建投《筹码分布因子系统构建》研报，实现四大类因子。
用日线数据就够了，分钟数据用于当日穿透率增强。

用法:
  from chip_factors import ChipFactors
  cf = ChipFactors()
  
  # 计算单只股票的筹码乖离率
  result = cf.calc_all("600519")
  print(result["bias"], result["ptr"], result["ptr_today"])
  
  # 批量计算所有持仓/自选
  results = cf.batch_calc(["600519", "00700", "000700"])
"""

import json, os, sqlite3, time, urllib.request
from datetime import datetime, timedelta
from pathlib import Path

DB_PATH = Path("/home/hmo/MoFin/data/mofin.db")
MOFIN_ROOT = Path("/home/hmo/MoFin")
CACHE_DIR = MOFIN_ROOT / "data" / "chip_cache"

# ── 分钟数据限流 ──
_last_minute_call = 0

def _fetch_quote(code):
    """拉腾讯实时价（清代理）"""
    for k in list(os.environ.keys()):
        if 'proxy' in k.lower():
            os.environ.pop(k)
    prefix = "sh" if code.startswith(('60','68','51','56','50')) else "sz" if code.startswith(('00','30','15')) else "hk"
    try:
        url = f"http://qt.gtimg.cn/q={prefix}{code}"
        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
        resp = urllib.request.urlopen(req, timeout=5).read().decode('gbk')
        fields = resp.split('=')[1].strip().strip('"').strip(';').split('~')
        return float(fields[3]) if len(fields) > 3 else 0
    except:
        return 0


def _fetch_minute_kline(code, count=60):
    """拉1分钟K线（带限流）"""
    global _last_minute_call
    now = time.time()
    if now - _last_minute_call < 1.0:
        time.sleep(1.0 - (now - _last_minute_call))
    secid = f"1.{code}" if code.startswith(('6','5')) else f"0.{code}"
    url = (f"https://push2.eastmoney.com/api/qt/stock/kline/get"
           f"?secid={secid}&fields1=f1,f2,f3&fields2=f51,f52,f53,f54,f55,f56,f57"
           f"&klt=1&fqt=1&end=20500101&lmt={min(count, 240)}")
    try:
        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
        resp = urllib.request.urlopen(req, timeout=8)
        data = json.loads(resp.read())["data"]["klines"]
        _last_minute_call = time.time()
        return [line.split(",") for line in data]
    except:
        return None


class ChipFactors:
    """筹码因子计算器"""
    
    def __init__(self):
        CACHE_DIR.mkdir(parents=True, exist_ok=True)
        self._cache = {}  # code → {last_chip, winner, bias}
        self._load_cache()
    
    def _load_cache(self):
        """加载缓存的筹码状态"""
        for f in CACHE_DIR.glob("*.json"):
            code = f.stem
            try:
                with open(f) as fp:
                    self._cache[code] = json.load(fp)
            except:
                pass
    
    def _save_cache(self, code):
        """保存筹码状态"""
        if code in self._cache:
            path = CACHE_DIR / f"{code}.json"
            with open(path, "w") as fp:
                json.dump(self._cache[code], fp, ensure_ascii=False)
    
    # ── 筹码分布估算（用日线OHLCV） ──
    def _build_chip_distribution(self, code):
        """从日线K线估算筹码分布。
        
        原理：假设每日成交量在OHLC区间内均匀分布，
        每根K线的成交量按价格区间分配，累积成筹码分布。
        """
        for k in list(os.environ.keys()):
            if 'proxy' in k.lower():
                os.environ.pop(k)
        # 从腾讯API取60日K线
        prefix = "sh" if code.startswith(('60','68','51')) else "sz" if code.startswith(('00','30','15')) else "hk"
        url = f"http://ifzq.gtimg.cn/appstock/app/fqkline/get?param={prefix}{code},day,,,640,qfq"
        try:
            opener = urllib.request.build_opener(urllib.request.ProxyHandler({}))
            req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
            resp = opener.open(req, timeout=8).read().decode('utf-8')
            data = json.loads(resp)
            day_key = 'qfqday' if prefix != 'hk' else 'day'
            bars = data.get('data', {}).get(f'{prefix}{code}', {}).get(day_key, [])
        except:
            return {}
        
        # 估算筹码分布：价格区间 → 累积量
        chip_dist = {}  # price_level → accumulated_volume
        decay = 0.97    # 每日衰减因子（老筹码逐步换手）
        
        for bar in bars:
            try:
                if len(bar) < 6:
                    continue
                high = float(bar[3])  # index 3 = high
                low = float(bar[4])   # index 4 = low
                volume = float(bar[5]) if len(bar) > 5 else 0  # index 5 = volume
                if high <= low or volume <= 0:
                    continue
                # 在OHLC区间均匀分配成交量
                step = max(round((high - low) / 5, 2), 0.01)
                level = round(low, 2)
                vol_per_level = volume / max(int((high - low) / step) + 1, 1)
                while level <= high:
                    chip_dist[level] = chip_dist.get(level, 0) + vol_per_level
                    level = round(level + step, 2)
            except:
                continue
        
        # 衰减老筹码
        total = sum(chip_dist.values())
        if total > 0:
            for k in chip_dist:
                chip_dist[k] *= decay
        
        return chip_dist
    
    # ── 三大因子计算 ──
    def calc_all(self, code, name="", price=None):
        """计算全部筹码因子，返回dict"""
        result = {"code": code, "name": name, "price": price}
        
        # 获取当前价（如果没传）
        if not price:
            price = _fetch_quote(code)
            if not price:
                return result
        result["price"] = price
        
        # 构建筹码分布
        chip_dist = self._build_chip_distribution(code)
        if not chip_dist or price <= 0:
            return result
        
        # 计算盈利/亏损筹码占比
        total_vol = sum(chip_dist.values())
        if total_vol <= 0:
            return result
        
        winner_vol = sum(v for k, v in chip_dist.items() if k <= price)  # 盈利筹码(cost≤现价)
        loser_vol = total_vol - winner_vol  # 亏损筹码(cost>现价)
        
        winner_pct = winner_vol / total_vol
        loser_pct = loser_vol / total_vol
        
        # 获取前日状态
        prev = self._cache.get(code, {})
        prev_winner = prev.get("winner_pct", winner_pct)
        prev_bias = prev.get("bias", 0)
        
        # 估算换手率（近10日均量/总流通股）
        turnover = 0.02  # 默认2%
        try:
            prefix2 = "sh" if code.startswith(('60','68','51','56','50')) else "sz" if code.startswith(('00','30','15')) else "hk"
            url2 = f"http://ifzq.gtimg.cn/appstock/app/fkline/get?param={prefix2}{code},day,,,10,qfq"
            req2 = urllib.request.Request(url2, headers={"User-Agent": "Mozilla/5.0"})
            resp2 = urllib.request.urlopen(req2, timeout=5).read().decode('utf-8')
            data2 = json.loads(resp2)
            dk = 'qfqday' if prefix2 != 'hk' else 'day'
            bars2 = data2.get('data', {}).get(f'{prefix2}{code}', {}).get(dk, [])
            if len(bars2) > 5:
                avg_vol = sum(float(b[5]) for b in bars2[-10:] if len(b)>5) / min(len(bars2), 10)
                # 用近60日最高量估算总流通股
                max_vol = avg_vol * 50  # 估算值
                turnover = min(avg_vol / max(max_vol, 1), 0.3)
        except:
            pass
        
        # 1. 筹码穿透率 PTR = (winner_pct - prev_winner) / turnover
        ptr = (winner_pct - prev_winner) / max(turnover, 0.001)
        
        # 2. 当日筹码穿透率(简化版) = 今日量 / 总筹码 / turnover
        ptr_today = 0
        minute_data = _fetch_minute_kline(code, count=30)
        if minute_data:
            today_vol = sum(float(m[5]) for m in minute_data if len(m) > 5)
            ptr_today = today_vol / max(total_vol, 1) / max(turnover, 0.001)
        
        # 3. 筹码乖离率(亏损版本—按文章发现，亏损筹码版反而最强)
        # bias = loser_pct * turnover + prev_bias * (1 - turnover)
        bias = loser_pct * turnover + prev_bias * (1 - turnover)
        
        # 更新缓存
        self._cache[code] = {
            "winner_pct": winner_pct,
            "loser_pct": loser_pct,
            "bias": bias,
            "updated_at": datetime.now().isoformat()
        }
        self._save_cache(code)
        
        return {
            "code": code,
            "name": name,
            "price": price,
            "winner_pct": round(winner_pct, 4),
            "loser_pct": round(loser_pct, 4),
            "ptr": round(ptr, 4),
            "ptr_today": round(ptr_today, 4),
            "bias": round(bias, 4),
            "turnover": round(turnover, 4),
        }
    
    def batch_calc(self, stocks):
        """批量计算多只股票"""
        results = []
        for i, (code, name) in enumerate(stocks):
            if i > 0:
                time.sleep(1.5)  # 限流
            result = self.calc_all(code, name)
            results.append(result)
        return results


# ── 主入口 ──
if __name__ == "__main__":
    import sys
    cf = ChipFactors()
    
    # 从decisions.json获取持仓+自选
    dec = json.loads(open(str(MOFIN_ROOT / "data" / "decisions.json")).read())
    stocks = [(s["code"], s.get("name","")) for s in dec.get("decisions", []) if s.get("status") != "closed"]
    
    results = cf.batch_calc(stocks)
    
    # 按bias排序显示（亏损筹码占比最高的排前面）
    results.sort(key=lambda r: r.get("bias", 0), reverse=True)
    print(f"{'股票':16} {'亏损筹码%':>10} {'PTR':>8} {'乖离率':>8} {'换手率':>8}")
    print("-" * 60)
    for r in results:
        if r.get("price"):
            print(f"{r['name']:8}({r['code']:6}) {r['loser_pct']*100:>8.1f}% {r['ptr']:>8.4f} {r['bias']:>8.4f} {r['turnover']*100:>6.1f}%")
    
    print(f"\n共计算{len(results)}只股票")
    print(f"筹码缓存目录: {CACHE_DIR}")