"""计算3只原油ETF跟踪WTI原油价格的准确率（使用Flask API数据源）"""
import os, sys
sys.path.insert(0, '/app')
from dotenv import load_dotenv
load_dotenv('/app/.env')

import pandas as pd
import numpy as np
import requests
import time

# 绕过系统代理避免SSL EOF
_session = requests.Session()
_session.trust_env = False

BASE_URL = os.getenv('FLASK_API_URL', 'https://k3s.tokenpluse.xyz')

def fetch_ohlcv(code, start='2020-01-01', end='2026-06-20'):
    """通过Flask API获取OHLCV数据"""
    url = f"{BASE_URL}/api/v1/ohlcv"
    params = {'code': code, 'start': start, 'end': end}
    for attempt in range(3):
        try:
            resp = _session.get(url, params=params, timeout=120)
            if resp.status_code == 200:
                data = resp.json()
                if 'error' in data or not data.get('data'):
                    print(f"  {code}: 无数据 - {data.get('error', 'empty')}")
                    return None
                df = pd.DataFrame(data['data'])
                df['date'] = pd.to_datetime(df['date'])
                df = df.set_index('date').sort_index()
                df['ret'] = df['close'].pct_change()
                cnt = data.get('count', len(df))
                dr = data.get('date_range', {})
                print(f"  {code}: {cnt} 条 ({dr.get('start','?')} ~ {dr.get('end','?')})")
                return df
            else:
                print(f"  {code}: HTTP {resp.status_code}")
                return None
        except Exception as e:
            if attempt < 2:
                time.sleep(2)
                continue
            print(f"  {code}: 失败 - {e}")
            return None
    return None

# 1. 获取WTI原油价格
print("=" * 90)
print("获取WTI原油价格 (CL=F)")
print("=" * 90)
cl_df = fetch_ohlcv('CL=F')

if cl_df is None or len(cl_df) < 10:
    print("WTI数据获取失败，退出")
    sys.exit(1)

# 2. 获取3只原油ETF价格
etf_codes = {
    '160723.SZ': '嘉实原油(WTI*100%)',
    '161129.SZ': '易方达原油(标普高盛原油*100%)',
    '501018.SH': '南方原油(WTI*60%+BRENT*40%)',
}

print("\n" + "=" * 90)
print("获取原油ETF价格")
print("=" * 90)
etf_dfs = {}
for code, name in etf_codes.items():
    df = fetch_ohlcv(code)
    if df is not None:
        etf_dfs[code] = name

# 3. 计算跟踪准确率
print("\n" + "=" * 90)
print("跟踪准确率（基于ETF收盘价收益率 vs WTI收盘价日收益率）")
print("=" * 90)

for code, name in etf_dfs.items():
    etf = etf_dfs[code][['ret']].copy()
    etf.columns = ['etf_ret']
    etf.index = etf.index.normalize()
    
    cl = cl_df[['ret']].copy()
    cl.columns = ['cl_ret']
    cl.index = cl.index.normalize()
    
    m = pd.merge(etf.reset_index(), cl.reset_index(), on='date', how='inner', suffixes=('', '_cl'))
    if 'ret' in m.columns and 'ret_cl' in m.columns:
        m = m.rename(columns={'ret': 'etf_ret', 'ret_cl': 'cl_ret'})
    m = m[['date', 'etf_ret', 'cl_ret']].dropna()
    
    if len(m) < 10:
        print(f"\n{code} {name}: 数据不足 ({len(m)} 天)")
        continue
    
    corr = m['etf_ret'].corr(m['cl_ret'])
    r2 = corr ** 2
    diff = m['etf_ret'] - m['cl_ret']
    te_annual = diff.std() * np.sqrt(252)
    cum_etf = (1 + m['etf_ret']).prod() - 1
    cum_cl = (1 + m['cl_ret']).prod() - 1
    bias = diff.mean()
    
    # 分段: 2024至今
    recent = m[m['date'] >= '2024-01-01']
    if len(recent) > 20:
        r2_recent = recent['etf_ret'].corr(recent['cl_ret']) ** 2
    else:
        r2_recent = np.nan
    
    print(f"\n{code} {name}")
    print(f"  重叠交易日: {len(m)} 天")
    print(f"  全区间 R²:   {r2:.4f} ({r2*100:.1f}%)")
    if not np.isnan(r2_recent):
        print(f"  2024至今 R²: {r2_recent:.4f} ({r2_recent*100:.1f}%)")
    else:
        print(f"  2024至今 R²: 数据不足")
    print(f"  年化跟踪误差: {te_annual*100:.2f}%")
    print(f"  日均偏差:     {bias*100:.4f}%")
    print(f"  ETF累计收益:  {cum_etf*100:.1f}%")
    print(f"  WTI累计收益:  {cum_cl*100:.1f}%")
    print(f"  累计收益差:    {(cum_etf-cum_cl)*100:.1f}%")

print("\n" + "=" * 90)
print("注: 原油ETF为QDII-LOF，净值披露有T+1~T+2延迟，")
print("    且需通过期货合约展期，与WTI现货价格存在结构性偏差。")
print("=" * 90)