diff --git a/czsc/__init__.py b/czsc/__init__.py
index fbe340ad8..da84e6324 100644
--- a/czsc/__init__.py
+++ b/czsc/__init__.py
@@ -15,6 +15,7 @@
 from czsc.strategies import CzscStrategyBase, CzscJsonStrategy
 from czsc.sensors import holds_concepts_effect, CTAResearch, EventMatchSensor
 from czsc.sensors.feature import FixedNumberSelector
+from czsc.utils import ta
 from czsc.traders import (
     CzscTrader,
     CzscSignals,
diff --git a/czsc/utils/ta.py b/czsc/utils/ta.py
index 4430eb1de..075eac933 100644
--- a/czsc/utils/ta.py
+++ b/czsc/utils/ta.py
@@ -50,12 +50,12 @@ def EMA(close: np.array, timeperiod=5):
     return np.array(res, dtype=np.double).round(4)
 
 
-def MACD(close: np.array, fastperiod=12, slowperiod=26, signalperiod=9):
+def MACD(real: np.array, fastperiod=12, slowperiod=26, signalperiod=9):
     """MACD 异同移动平均线
     https://baike.baidu.com/item/MACD%E6%8C%87%E6%A0%87/6271283
 
-    :param close: np.array
-        收盘价序列
+    :param real: np.array
+        价格序列
     :param fastperiod: int
         快周期，默认值 12
     :param slowperiod: int
@@ -65,8 +65,8 @@ def MACD(close: np.array, fastperiod=12, slowperiod=26, signalperiod=9):
     :return: (np.array, np.array, np.array)
         diff, dea, macd
     """
-    ema12 = EMA(close, timeperiod=fastperiod)
-    ema26 = EMA(close, timeperiod=slowperiod)
+    ema12 = EMA(real, timeperiod=fastperiod)
+    ema26 = EMA(real, timeperiod=slowperiod)
     diff = ema12 - ema26
     dea = EMA(diff, timeperiod=signalperiod)
     macd = (diff - dea) * 2
@@ -146,7 +146,7 @@ def RSQ(close: [np.array, list]) -> float:
     return round(rsq, 4)
 
 
-def plus_di(high, low, close, timeperiod=14):
+def PLUS_DI(high, low, close, timeperiod=14):
     """
     Calculate Plus Directional Indicator (PLUS_DI) manually.
 
@@ -179,7 +179,7 @@ def plus_di(high, low, close, timeperiod=14):
     return plus_di_
 
 
-def minus_di(high, low, close, timeperiod=14):
+def MINUS_DI(high, low, close, timeperiod=14):
     """
     Calculate Minus Directional Indicator (MINUS_DI) manually.
 
@@ -213,7 +213,7 @@ def minus_di(high, low, close, timeperiod=14):
     return minus_di_
 
 
-def atr(high, low, close, timeperiod=14):
+def ATR(high, low, close, timeperiod=14):
     """
     Calculate Average True Range (ATR).
 
@@ -234,7 +234,6 @@ def atr(high, low, close, timeperiod=14):
 
     # Calculate ATR
     atr_ = tr.rolling(window=timeperiod).mean()
-
     return atr_
 
 
@@ -338,3 +337,88 @@ def LINEARREG_ANGLE(real, timeperiod=14):
         angles[today] = np.arctan(m) * (180.0 / np.pi)
 
     return angles
+
+
+def DOUBLE_SMA_LS(series: pd.Series, n=5, m=20, **kwargs):
+    """双均线多空
+
+    :param series: str, 数据源字段
+    :param n: int, 短周期
+    :param m: int, 长周期
+    """
+    assert n < m, "短周期必须小于长周期"
+    return np.sign(series.rolling(window=n).mean() - series.rolling(window=m).mean()).fillna(0)
+
+
+def BOLL_LS(series: pd.Series, n=5, s=0.1, **kwargs):
+    """布林线多空
+
+    series 大于 n 周期均线 + s * n周期标准差，做多；小于 n 周期均线 - s * n周期标准差，做空
+
+    :param series: str, 数据源字段
+    :param n: int, 短周期
+    :param s: int, 波动率的倍数，默认为 0.1
+    """
+    sm = series.rolling(window=n).mean()
+    sd = series.rolling(window=n).std()
+    return np.where(series > sm + s * sd, 1, np.where(series < sm - s * sd, -1, 0))
+
+
+def SMA_MIN_MAX_SCALE(series: pd.Series, timeperiod=5, window=5, **kwargs):
+    """均线的最大最小值归一化
+
+    :param series: str, 数据源字段
+    :param timeperiod: int, 均线周期
+    :param window: int, 窗口大小
+    """
+    sm = series.rolling(window=timeperiod).mean()
+    sm_min = sm.rolling(window=window).min()
+    sm_max = sm.rolling(window=window).max()
+    res = (sm - sm_min) / (sm_max - sm_min)
+    res = res.fillna(0) * 2 - 1
+    return res
+
+
+def RS_VOLATILITY(df: pd.DataFrame, timeperiod=30, **kwargs):
+    """RS 波动率，值越大，波动越大
+
+    :param df: str, 标准K线数据
+    :param timeperiod: int, 周期
+    """
+    log_h_c = np.log(df["high"] / df["close"])
+    log_h_o = np.log(df["high"] / df["open"])
+    log_l_c = np.log(df["low"] / df["close"])
+    log_l_o = np.log(df["low"] / df["open"])
+
+    x = log_h_c * log_h_o + log_l_c * log_l_o
+    res = np.sqrt(x.rolling(window=timeperiod).mean())
+    return res
+
+
+def PK_VOLATILITY(df: pd.DataFrame, timeperiod=30, **kwargs):
+    """PK 波动率，值越大，波动越大
+
+    :param df: str, 标准K线数据
+    :param timeperiod: int, 周期
+    """
+    log_h_l = np.log(df["high"] / df["low"]).pow(2)
+    log_hl_mean = log_h_l.rolling(window=timeperiod).sum() / (4 * timeperiod * np.log(2))
+    res = np.sqrt(log_hl_mean)
+    return res
+
+
+try:
+    import talib as ta
+
+    SMA = ta.SMA
+    EMA = ta.EMA
+    MACD = ta.MACD
+    LINEARREG_ANGLE = ta.LINEARREG_ANGLE
+except ImportError:
+    SMA = SMA
+    EMA = EMA
+    MACD = MACD
+    print(
+        f"ta-lib 没有正确安装，将使用自定义分析函数。建议安装 ta-lib，可以大幅提升计算速度。"
+        f"请参考安装教程 https://blog.csdn.net/qaz2134560/article/details/98484091"
+    )
diff --git a/czsc/utils/ta1.py b/czsc/utils/ta1.py
deleted file mode 100644
index 63bd47542..000000000
--- a/czsc/utils/ta1.py
+++ /dev/null
@@ -1,86 +0,0 @@
-# coding: utf-8
-"""
-
-常用技术分析指标：MA, MACD, BOLL
-
-使用 ta-lib 可以大幅提升计算性能，10倍左右
-"""
-import numpy as np
-import talib as ta
-
-SMA = ta.SMA
-EMA = ta.EMA
-MACD = ta.MACD
-
-
-def KDJ(close: np.array, high: np.array, low: np.array):
-    """
-
-    :param close: 收盘价序列
-    :param high: 最高价序列
-    :param low: 最低价序列
-    :return:
-    """
-    n = 9
-    hv = []
-    lv = []
-    for i in range(len(close)):
-        if i < n:
-            h_ = high[0: i+1]
-            l_ = low[0: i+1]
-        else:
-            h_ = high[i - n + 1: i + 1]
-            l_ = low[i - n + 1: i + 1]
-        hv.append(max(h_))
-        lv.append(min(l_))
-
-    hv = np.around(hv, decimals=2)
-    lv = np.around(lv, decimals=2)
-    rsv = np.where(hv == lv, 0, (close - lv) / (hv - lv) * 100)
-
-    k = []
-    d = []
-    j = []
-    for i in range(len(rsv)):
-        if i < n:
-            k_ = rsv[i]
-            d_ = k_
-        else:
-            k_ = (2 / 3) * k[i-1] + (1 / 3) * rsv[i]
-            d_ = (2 / 3) * d[i-1] + (1 / 3) * k_
-
-        k.append(k_)
-        d.append(d_)
-        j.append(3 * k_ - 2 * d_)
-
-    k = np.array(k, dtype=np.double)
-    d = np.array(d, dtype=np.double)
-    j = np.array(j, dtype=np.double)
-    return k, d, j
-
-
-def RSQ(close: [np.array, list]) -> float:
-    """拟合优度 R SQuare
-
-    :param close: 收盘价序列
-    :return:
-    """
-    x = list(range(len(close)))
-    y = np.array(close)
-    x_squred_sum = sum([x1 * x1 for x1 in x])
-    xy_product_sum = sum([x[i] * y[i] for i in range(len(x))])
-    num = len(x)
-    x_sum = sum(x)
-    y_sum = sum(y)
-    delta = float(num * x_squred_sum - x_sum * x_sum)
-    if delta == 0:
-        return 0
-    y_intercept = (1 / delta) * (x_squred_sum * y_sum - x_sum * xy_product_sum)
-    slope = (1 / delta) * (num * xy_product_sum - x_sum * y_sum)
-
-    y_mean = np.mean(y)
-    ss_tot = sum([(y1 - y_mean) * (y1 - y_mean) for y1 in y]) + 0.00001
-    ss_err = sum([(y[i] - slope * x[i] - y_intercept) * (y[i] - slope * x[i] - y_intercept) for i in range(len(x))])
-    rsq = 1 - ss_err / ss_tot
-
-    return round(rsq, 4)
diff --git "a/examples/develop/\345\257\271\346\257\224numpy\344\270\216talib\347\232\204\351\200\237\345\272\246.py" "b/examples/develop/\345\257\271\346\257\224numpy\344\270\216talib\347\232\204\351\200\237\345\272\246.py"
index 75a47430b..1f2700062 100644
--- "a/examples/develop/\345\257\271\346\257\224numpy\344\270\216talib\347\232\204\351\200\237\345\272\246.py"
+++ "b/examples/develop/\345\257\271\346\257\224numpy\344\270\216talib\347\232\204\351\200\237\345\272\246.py"
@@ -1,4 +1,5 @@
 import sys
+import time
 
 sys.path.insert(0, r"A:\ZB\git_repo\waditu\czsc")
 import talib
@@ -6,14 +7,83 @@
 from czsc.connectors import cooperation as coo
 
 
-df = coo.get_raw_bars(symbol="SFIC9001", freq="30分钟", fq="后复权", sdt="20100101", edt="20210301", raw_bars=False)
+df = coo.get_raw_bars(symbol="SFIC9001", freq="日线", fq="后复权", sdt="20100101", edt="20210301", raw_bars=False)
 
 
-def test_with_numpy():
+def test_vs_volatility():
     df1 = df.copy()
-    df1["x"] = ta.LINEARREG_ANGLE(df["close"].values, 10)
+    df1["STD"] = df["close"].pct_change().rolling(30).std()
+    df1["RS"] = ta.RS_VOLATILITY(df1, timeperiod=30)
+    df1["PK"] = ta.PK_VOLATILITY(df1, timeperiod=30)
+    df1[["STD", "RS", "PK"]].corr()
 
 
-def test_with_talib():
+def test_compare_LINEARREG_ANGLE():
     df1 = df.copy()
-    df1["x"] = talib.LINEARREG_ANGLE(df["close"].values, 10)
+    s1 = time.time()
+    df1["x1"] = ta.LINEARREG_ANGLE(df["close"].values, 10)
+    e1 = time.time() - s1
+
+    s2 = time.time()
+    df1["x2"] = talib.LINEARREG_ANGLE(df["close"].values, 10)
+    e2 = time.time() - s2
+
+    print(f"计算时间差异，ta: {e1}, talib: {e2}；相差：{e1 - e2}")
+    df1["diff"] = df1["x1"] - df1["x2"]
+    assert df1["diff"].abs().max() < 1e-6
+    print(df1["diff"].abs().max())
+
+
+def test_compare_CCI():
+    # 数据对不上，需要调整
+    df1 = df.copy()
+    s1 = time.time()
+    df1["x1"] = ta.CCI(df1["high"].values, df1["low"].values, df1["close"].values, timeperiod=14)
+    e1 = time.time() - s1
+
+    s2 = time.time()
+    df1["x2"] = talib.CCI(df1["high"].values, df1["low"].values, df1["close"].values, timeperiod=14)
+    e2 = time.time() - s2
+
+    print(f"计算时间差异，ta: {e1}, talib: {e2}；相差：{e1 - e2}")
+    df1["diff"] = df1["x1"] - df1["x2"]
+    print(df1["diff"].abs().describe())
+
+    assert df1["diff"].abs().max() < 1e-6
+
+
+def test_compare_MFI():
+    # 数据对不上，需要调整
+    df1 = df.copy()
+    s1 = time.time()
+    df1["x1"] = ta.MFI(df1["high"], df1["low"], df1["close"], df1["vol"], timeperiod=14)
+    e1 = time.time() - s1
+
+    s2 = time.time()
+    df1["x2"] = talib.MFI(df1["high"].values, df1["low"].values, df1["close"].values, df1["vol"].values, timeperiod=14)
+    e2 = time.time() - s2
+
+    print(f"计算时间差异，ta: {e1}, talib: {e2}；相差：{e1 - e2}")
+    df1["diff"] = df1["x1"] - df1["x2"]
+    print(df1["diff"].abs().describe())
+
+    assert df1["diff"].abs().max() < 1e-6
+
+
+def test_compare_PLUS_DI():
+    # 数据对不上，需要调整
+    df1 = coo.get_raw_bars(symbol="SFIC9001", freq="日线", fq="后复权", sdt="20100101", edt="20210301", raw_bars=False)
+
+    s1 = time.time()
+    df1["x1"] = ta.PLUS_DI(df1["high"], df1["low"], df1["close"], timeperiod=14)
+    e1 = time.time() - s1
+
+    s2 = time.time()
+    df1["x2"] = talib.PLUS_DI(df1["high"].values, df1["low"].values, df1["close"].values, timeperiod=14)
+    e2 = time.time() - s2
+
+    print(f"计算时间差异，ta: {e1}, talib: {e2}；相差：{e1 - e2}")
+    df1["diff"] = df1["x1"] - df1["x2"]
+    print(df1["diff"].abs().describe())
+
+    assert df1["diff"].abs().max() < 1e-6