pip install "tickflow[all]" --upgrade

from tickflow import TickFlow

# 免費(fèi)服務(wù)（日 K 線夠用）
tf_free = TickFlow.free()

# 完整服務(wù)（需要實(shí)時(shí)行情、財(cái)務(wù)數(shù)據(jù)等）
tf = TickFlow(api_key="your-api-key")

from tickflow import TickFlow

tf = TickFlow.free()

symbol = "600519.SH"

# 比例前復(fù)權(quán)（默認(rèn)，推薦用于量化回測(cè)）
df_forward = tf.klines.get(symbol, period="1d", count=2000, adjust="forward", as_dataframe=True)

# 差值前復(fù)權(quán)（與東方財(cái)富/同花順價(jià)格一致）
df_additive = tf.klines.get(symbol, period="1d", count=2000, adjust="forward_additive", as_dataframe=True)

# 不復(fù)權(quán)（原始成交價(jià)格）
df_none = tf.klines.get(symbol, period="1d", count=2000, adjust="none", as_dataframe=True)

# 比例后復(fù)權(quán)
df_backward = tf.klines.get(symbol, period="1d", count=2000, adjust="backward", as_dataframe=True)

# 差值后復(fù)權(quán)
df_backward_add = tf.klines.get(symbol, period="1d", count=2000, adjust="backward_additive", as_dataframe=True)

import pandas as pd

compare = pd.DataFrame({
    "trade_date": df_none["trade_date"],
    "不復(fù)權(quán)": df_none["close"],
    "比例前復(fù)權(quán)": df_forward["close"],
    "差值前復(fù)權(quán)": df_additive["close"],
    "比例后復(fù)權(quán)": df_backward["close"],
})

print(compare.tail(5))
print(f"\n不復(fù)權(quán)最新價(jià): {df_none['close'].iloc[-1]}")
print(f"前復(fù)權(quán)最新價(jià): {df_forward['close'].iloc[-1]}")
print(f"后復(fù)權(quán)最新價(jià): {df_backward['close'].iloc[-1]}")

tf = TickFlow(api_key="your-api-key")

factors = tf.klines.ex_factors(["600519.SH", "000001.SZ"], as_dataframe=True)

# 按標(biāo)的分組查看
for symbol in ["600519.SH", "000001.SZ"]:
    sym_factors = factors[factors["symbol"] == symbol]
    print(f"\n{symbol} 最近 5 次除權(quán):")
    print(sym_factors[["trade_date", "ex_factor"]].tail(5).to_string(index=False))

import pandas as pd
from tickflow import TickFlow

tf = TickFlow.free()

symbols = ["600519.SH", "000858.SZ", "601318.SH"]
dfs = tf.klines.batch(symbols, period="1d", count=500, adjust="forward", as_dataframe=True)

# 提取所有交易日的并集
all_dates = set()
for df in dfs.values():
    all_dates.update(df["trade_date"].tolist())
all_dates = sorted(all_dates)

# 對(duì)齊到統(tǒng)一日期索引
aligned = pd.DataFrame({"trade_date": all_dates})
for symbol, df in dfs.items():
    close = df[["trade_date", "close"]].rename(columns={"close": symbol})
    aligned = aligned.merge(close, on="trade_date", how="left")

print(f"對(duì)齊后形狀: {aligned.shape}")
print(aligned.tail())

# 取交集（所有股票都有數(shù)據(jù)的日期）
common_dates = None
for df in dfs.values():
    dates = set(df["trade_date"].tolist())
    common_dates = dates if common_dates is None else common_dates & dates

common_dates = sorted(common_dates)

aligned_common = pd.DataFrame({"trade_date": common_dates})
for symbol, df in dfs.items():
    close = df[["trade_date", "close"]].rename(columns={"close": symbol})
    aligned_common = aligned_common.merge(close, on="trade_date", how="inner")

print(f"共有交易日: {len(aligned_common)}")
print(aligned_common.tail())

# 在方案一的基礎(chǔ)上，前向填充
aligned_filled = aligned.copy()
for symbol in symbols:
    aligned_filled[symbol] = aligned_filled[symbol].ffill()

# 檢查填充效果
null_before = aligned[symbols].isnull().sum()
null_after = aligned_filled[symbols].isnull().sum()
print("填充前缺失值:")
print(null_before)
print("\n填充后缺失值:")
print(null_after)

import pandas as pd
from tickflow import TickFlow

tf = TickFlow.free()

symbols = ["600519.SH", "000858.SZ", "601318.SH", "000001.SZ", "600036.SH"]
dfs = tf.klines.batch(symbols, period="1d", count=1000, adjust="forward", as_dataframe=True)

print("各標(biāo)的數(shù)據(jù)量:")
for symbol, df in dfs.items():
    start = df["trade_date"].iloc[0]
    end = df["trade_date"].iloc[-1]
    print(f"  {symbol}: {len(df)} 根 K 線 ({start} ~ {end})")

# 將多只股票收盤價(jià)合并為一個(gè) DataFrame
close_df = pd.DataFrame()
for symbol, df in dfs.items():
    s = df.set_index("trade_date")["close"]
    close_df[symbol] = s

# 方法 1：前向填充（適合停牌）
close_filled = close_df.ffill()

# 方法 2：刪除含有任何缺失值的行（保守但干凈）
close_clean = close_df.dropna()
print(f"原始行數(shù): {len(close_df)}, 清洗后: {len(close_clean)}")

# 方法 3：僅刪除缺失值過(guò)多的行（某天超過(guò)半數(shù)股票無(wú)數(shù)據(jù)才刪）
threshold = len(symbols) // 2
close_filtered = close_df.dropna(thresh=threshold)
print(f"過(guò)濾后: {len(close_filtered)}")

import pandas as pd
from tickflow import TickFlow

tf = TickFlow(api_key="your-api-key")

symbol = "600000.SH"

# 獲取日 K 線
kdf = tf.klines.get(symbol, period="1d", count=2000, as_dataframe=True)

# 獲取股本數(shù)據(jù)
sdf = tf.financials.shares([symbol], as_dataframe=True)

# 轉(zhuǎn)為日期類型
kdf["date"] = pd.to_datetime(kdf["trade_date"])
sdf["date"] = pd.to_datetime(sdf["period_end"])

# 使用 merge_asof 為每根 K 線匹配當(dāng)時(shí)有效的流通股本
merged = pd.merge_asof(
    kdf.sort_values("date"),
    sdf[["date", "float_shares"]].sort_values("date"),
    on="date",
    direction="backward",
)

# 計(jì)算換手率
merged["turnover_rate"] = (merged["volume"] * 100 / merged["float_shares"]).round(4)
merged["turnover_pct"] = merged["turnover_rate"] * 100

print(merged[["trade_date", "close", "volume", "float_shares", "turnover_pct"]].tail(10))

symbols = ["600519.SH", "000858.SZ", "601318.SH"]

# 批量獲取 K 線
kline_dfs = tf.klines.batch(symbols, period="1d", count=60, as_dataframe=True)

# 批量獲取股本
shares_df = tf.financials.shares(symbols, as_dataframe=True)

for symbol in symbols:
    kdf = kline_dfs.get(symbol)
    sdf = shares_df[shares_df["symbol"] == symbol].copy()

    if kdf is None or len(sdf) == 0:
        continue

    kdf["date"] = pd.to_datetime(kdf["trade_date"])
    sdf["date"] = pd.to_datetime(sdf["period_end"])

    merged = pd.merge_asof(
        kdf.sort_values("date"),
        sdf[["date", "float_shares"]].sort_values("date"),
        on="date",
        direction="backward",
    )

    merged["turnover_pct"] = (merged["volume"] * 100 / merged["float_shares"] * 100).round(2)
    avg_turnover = merged["turnover_pct"].mean()
    print(f"{symbol}: 近 60 日平均換手率 {avg_turnover:.2f}%")

import numpy as np
from tickflow import TickFlow

tf = TickFlow.free()

df = tf.klines.get("600519.SH", period="1d", count=250, adjust="forward", as_dataframe=True)

# 簡(jiǎn)單收益率（常用）
df["simple_return"] = df["close"].pct_change()

# 對(duì)數(shù)收益率（適合統(tǒng)計(jì)分析，可加性更好）
df["log_return"] = np.log(df["close"] / df["close"].shift(1))

print(df[["trade_date", "close", "simple_return", "log_return"]].tail(10))

# 簡(jiǎn)單收益率的累計(jì)
df["cum_return"] = (1 + df["simple_return"]).cumprod() - 1

# 或用對(duì)數(shù)收益率
df["cum_return_log"] = np.exp(df["log_return"].cumsum()) - 1

print(f"期間總收益: {df['cum_return'].iloc[-1] * 100:.2f}%")

from tickflow import TickFlow

tf = TickFlow.free()

df = tf.klines.get("600000.SH", period="1d", count=5000, adjust="forward", as_dataframe=True)

# 1. 檢查是否有重復(fù)日期
dup = df[df["trade_date"].duplicated()]
print(f"重復(fù)日期: {len(dup)} 個(gè)")

# 2. 檢查價(jià)格異常（收盤價(jià)為 0 或負(fù)數(shù)）
bad_price = df[df["close"] <= 0]
print(f"異常價(jià)格: {len(bad_price)} 個(gè)")

# 3. 檢查 OHLC 邏輯（high >= low, high >= open, high >= close）
ohlc_err = df[(df["high"] < df["low"]) | (df["high"] < df["open"]) | (df["high"] < df["close"])]
print(f"OHLC 邏輯異常: {len(ohlc_err)} 個(gè)")

# 4. 檢查日期連續(xù)性（是否有異常缺失）
df["date"] = pd.to_datetime(df["trade_date"])
df["gap"] = df["date"].diff().dt.days
big_gaps = df[df["gap"] > 5]  # 超過(guò) 5 天的間隔（排除周末和短假）
print(f"超過(guò) 5 天的間隔: {len(big_gaps)} 個(gè)")
if len(big_gaps) > 0:
    print(big_gaps[["trade_date", "gap"]].head())

import os
import pandas as pd
from tickflow import TickFlow

CACHE_DIR = "./kline_cache"
os.makedirs(CACHE_DIR, exist_ok=True)


def get_klines_cached(tf, symbol, period="1d", count=10000):
    """帶本地緩存的 K 線獲取"""
    cache_file = os.path.join(CACHE_DIR, f"{symbol}_{period}.csv")

    if os.path.exists(cache_file):
        cached = pd.read_csv(cache_file)
        last_date = cached["trade_date"].iloc[-1]
        print(f"{symbol}: 緩存到 {last_date}，增量更新...")

        # 獲取緩存之后的新數(shù)據(jù)
        import datetime
        last_ts = int(datetime.datetime.strptime(last_date, "%Y-%m-%d").timestamp() * 1000)
        new_data = tf.klines.get(
            symbol, period=period, start_time=last_ts + 86400000,
            count=count, as_dataframe=True,
        )

        if len(new_data) > 0:
            # 過(guò)濾掉可能重復(fù)的最后一天
            new_data = new_data[new_data["trade_date"] > last_date]
            combined = pd.concat([cached, new_data], ignore_index=True)
            combined.to_csv(cache_file, index=False)
            print(f"  新增 {len(new_data)} 根 K 線")
            return combined
        return cached
    else:
        print(f"{symbol}: 首次獲取...")
        df = tf.klines.get(symbol, period=period, count=count, as_dataframe=True)
        df.to_csv(cache_file, index=False)
        return df


tf = TickFlow.free()
df = get_klines_cached(tf, "600519.SH")
print(f"總計(jì) {len(df)} 根 K 線")