225 lines
8.0 KiB
Python
225 lines
8.0 KiB
Python
"""市场温度计
|
||
|
||
综合涨跌家数、涨跌停数、连板高度、炸板率、指数位置，
|
||
输出 0-100 的市场温度值。
|
||
"""
|
||
|
||
import logging
|
||
import pandas as pd
|
||
import numpy as np
|
||
|
||
from app.data import tencent_client
|
||
from app.data.market_breadth_client import get_market_breadth
|
||
from app.data.tushare_client import tushare_client
|
||
from app.data.models import MarketTemperature
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
|
||
def calculate_market_temperature(trade_date: str = None) -> MarketTemperature:
|
||
"""计算指定交易日的市场温度
|
||
|
||
盘中 Tushare 当日数据可能不完整（limit_list_d 只有盘后数据），
|
||
因此涨停数据先尝试当日，若为空则用前一日数据作为基线。
|
||
"""
|
||
if not trade_date:
|
||
trade_date = tushare_client.get_latest_trade_date()
|
||
|
||
# 1. 涨跌家数
|
||
daily_df = tushare_client.get_daily_all(trade_date)
|
||
if daily_df.empty:
|
||
original_trade_date = trade_date
|
||
trade_dates = tushare_client.get_trade_dates()
|
||
fallback_dates = [d for d in trade_dates if d < trade_date][-5:]
|
||
for fallback_date in reversed(fallback_dates):
|
||
fallback_df = tushare_client.get_daily_all(fallback_date)
|
||
if not fallback_df.empty:
|
||
daily_df = fallback_df
|
||
trade_date = fallback_date
|
||
logger.warning(
|
||
"市场温度 %s 日线数据为空，回退到最近有效交易日 %s",
|
||
original_trade_date,
|
||
fallback_date,
|
||
)
|
||
break
|
||
|
||
if daily_df.empty:
|
||
logger.warning("市场温度 %s 无有效日线数据，返回占位温度", trade_date)
|
||
return MarketTemperature(trade_date=trade_date, temperature=50)
|
||
|
||
up_count = len(daily_df[daily_df["pct_chg"] > 0])
|
||
down_count = len(daily_df[daily_df["pct_chg"] < 0])
|
||
|
||
# 2. 涨跌停数据 — 先尝试当日，若为空则用前一日
|
||
limit_df = tushare_client.get_limit_list(trade_date)
|
||
|
||
if limit_df.empty:
|
||
# 当日无涨跌停数据（盘中），用前一日作为基线
|
||
prev_dates = tushare_client.get_trade_dates()
|
||
prev_dates = [d for d in prev_dates if d < trade_date]
|
||
if prev_dates:
|
||
prev_date = prev_dates[-1]
|
||
limit_df = tushare_client.get_limit_list(prev_date)
|
||
logger.info(f"当日涨跌停数据为空，使用前一日 {prev_date} 作为基线")
|
||
|
||
limit_up_count = 0
|
||
limit_down_count = 0
|
||
max_streak = 0
|
||
broken_rate = 0.0
|
||
|
||
if not limit_df.empty:
|
||
# limit: U=涨停, D=跌停
|
||
up_df = limit_df[limit_df["limit"] == "U"]
|
||
down_df = limit_df[limit_df["limit"] == "D"]
|
||
|
||
# 排除 ST（名称含 ST）
|
||
up_df = up_df[~up_df["name"].str.contains("ST", na=False)]
|
||
down_df = down_df[~down_df["name"].str.contains("ST", na=False)]
|
||
|
||
# 排除一字板（open_times == 0 表示全天封板未开板，不算自然涨停...
|
||
# 实际上 open_times 是打开次数，0 表示一直封住，属于强势
|
||
# 这里我们统计所有非 ST 涨停数作为情绪指标）
|
||
limit_up_count = len(up_df)
|
||
limit_down_count = len(down_df)
|
||
|
||
# 连板高度：limit_times 字段
|
||
if "limit_times" in up_df.columns and not up_df.empty:
|
||
max_streak = int(up_df["limit_times"].max())
|
||
|
||
# 炸板率：曾经涨停但未封住的 = up_stat 中 "炸板" 的记录
|
||
# open_times > 0 的算曾经打开过涨停板，如果最终未涨停就是炸板
|
||
# 但 limit_list_d 只返回最终涨停的... 需要另一个思路
|
||
# 用全市场日线数据：最高价触及涨停但收盘未涨停
|
||
if not daily_df.empty:
|
||
# 涨停价 = pre_close * 1.1（普通股），容差 0.01
|
||
daily_df["limit_price"] = daily_df["pre_close"] * 1.1
|
||
touched_limit = daily_df[
|
||
(daily_df["high"] >= daily_df["limit_price"] - 0.01) &
|
||
(daily_df["pct_chg"] < 9.9) # 触及涨停但收盘未涨停
|
||
]
|
||
total_touched = len(touched_limit) + limit_up_count
|
||
if total_touched > 0:
|
||
broken_rate = len(touched_limit) / total_touched * 100
|
||
|
||
# 3. 上证指数位置
|
||
index_df = tushare_client.get_index_daily("000001.SH", days=30)
|
||
index_above_ma20 = False
|
||
if not index_df.empty:
|
||
index_df = index_df.sort_values("trade_date")
|
||
if len(index_df) >= 20:
|
||
ma20 = index_df["close"].tail(20).mean()
|
||
latest_close = index_df["close"].iloc[-1]
|
||
index_above_ma20 = latest_close > ma20
|
||
|
||
# ── 综合评分 ──
|
||
score = 0.0
|
||
|
||
# 涨跌家数比 (0-25分)
|
||
ratio = up_count / max(down_count, 1)
|
||
if ratio > 2.0:
|
||
score += 25
|
||
elif ratio > 1.5:
|
||
score += 20
|
||
elif ratio > 1.0:
|
||
score += 15
|
||
elif ratio > 0.7:
|
||
score += 8
|
||
else:
|
||
score += 0
|
||
|
||
# 涨停数 (0-25分)
|
||
if limit_up_count > 80:
|
||
score += 25
|
||
elif limit_up_count > 50:
|
||
score += 20
|
||
elif limit_up_count > 30:
|
||
score += 15
|
||
elif limit_up_count > 15:
|
||
score += 8
|
||
else:
|
||
score += 0
|
||
|
||
# 连板高度 (0-20分)
|
||
if max_streak >= 7:
|
||
score += 20
|
||
elif max_streak >= 5:
|
||
score += 15
|
||
elif max_streak >= 3:
|
||
score += 10
|
||
elif max_streak >= 2:
|
||
score += 5
|
||
|
||
# 炸板率 (0-15分) - 炸板率越低越好
|
||
if broken_rate < 20:
|
||
score += 15
|
||
elif broken_rate < 30:
|
||
score += 12
|
||
elif broken_rate < 50:
|
||
score += 6
|
||
else:
|
||
score += 0
|
||
|
||
# 指数位置 (0-15分)
|
||
if index_above_ma20:
|
||
score += 15
|
||
else:
|
||
score += 3
|
||
|
||
temperature = min(max(score, 0), 100)
|
||
|
||
result = MarketTemperature(
|
||
trade_date=trade_date,
|
||
up_count=up_count,
|
||
down_count=down_count,
|
||
limit_up_count=limit_up_count,
|
||
limit_down_count=limit_down_count,
|
||
max_streak=max_streak,
|
||
broken_rate=round(broken_rate, 1),
|
||
index_above_ma20=index_above_ma20,
|
||
temperature=round(temperature, 1),
|
||
source="tushare_daily",
|
||
data_status="fresh",
|
||
source_detail=f"daily={trade_date}",
|
||
limit_counts_reliable=not limit_df.empty,
|
||
)
|
||
logger.info(f"市场温度 {trade_date}: {temperature:.1f} (涨{up_count}/跌{down_count}, 涨停{limit_up_count}, 连板{max_streak})")
|
||
return result
|
||
|
||
|
||
async def build_realtime_market_temperature(
|
||
baseline: MarketTemperature | None = None,
|
||
) -> tuple[MarketTemperature, bool]:
|
||
"""基于统一市场广度客户端构建实时市场温度。"""
|
||
baseline = baseline or calculate_market_temperature()
|
||
breadth = await get_market_breadth()
|
||
if not breadth.reliable:
|
||
return baseline, False
|
||
|
||
index_data = await tencent_client.get_index_realtime()
|
||
sh_index = index_data.get("000001.SH")
|
||
pct = sh_index.get("pct_chg", 0) if sh_index else 0
|
||
|
||
ratio = breadth.up_count / max(breadth.down_count, 1)
|
||
# 实时口径里，涨跌停池可靠时用专门池；池接口失败时用全市场行情阈值估算。
|
||
# 估算口径权重略低，并在研究报告中标记为降级，避免悄悄污染判断。
|
||
temp_from_ratio = min(ratio / 3.0 * 40, 40)
|
||
temp_from_limit_up = (
|
||
min(breadth.limit_up_count / 3, 10)
|
||
if breadth.limit_counts_reliable
|
||
else min(breadth.limit_up_count / 5, 8)
|
||
)
|
||
temp_from_index = min(max(pct * 10 + 15, 0), 30)
|
||
|
||
baseline.trade_date = breadth.trade_date
|
||
baseline.up_count = breadth.up_count
|
||
baseline.down_count = breadth.down_count
|
||
baseline.limit_up_count = breadth.limit_up_count
|
||
baseline.limit_down_count = breadth.limit_down_count
|
||
baseline.temperature = round(min(max(temp_from_ratio + temp_from_limit_up + temp_from_index + 20, 0), 100), 1)
|
||
baseline.index_above_ma20 = pct > 0 if sh_index else baseline.index_above_ma20
|
||
baseline.source = breadth.source
|
||
baseline.source_detail = breadth.source
|
||
baseline.limit_counts_reliable = breadth.limit_counts_reliable
|
||
baseline.data_status = "fresh" if breadth.limit_counts_reliable else "estimated"
|
||
return baseline, True
|