I am trying to calculate the connors rsi in python using pandas and numpy. I want to calculate it with the default values of ConnorsRSI(3,2,100).

我正在试着用熊猫和NumPy来计算蟒蛇中的Connors RSI。我想用ConnorsRSI的默认值(3，2,100)来计算它。

The formula for the connors RSI is : [ RSI(Close,3) + RSI(Streak,2) + PercentRank(100) ] / 3

Connors RSI的公式为：[RSI(Close，3)+RSI(Streak，2)+PercentRank(100)]/3

I start with the rsi(3):

我从RSI(3)开始：

# Returns RSI values
def rsi(close, periods):
    
    close_delta = close.diff()

    # Make two series: one for lower closes and one for higher closes
    up = close_delta.clip(lower=0)
    down = -1 * close_delta.clip(upper=0)
    
    ma_up = up.ewm(com = periods - 1, adjust=True, min_periods = periods).mean()
    ma_down = down.ewm(com = periods - 1, adjust=True, min_periods = periods).mean()

    rsi = ma_up / ma_down
    rsi = 100 - (100/(1 + rsi))
    return rsi

df['rsi(3)'] = rsi(df['close'],3)
df.tail()

Then I calculate the streak

然后我计算出连胜线

def get_streaks(closing_prices):
    # logic tables
    series = pd.DataFrame(closing_prices)
    geq = series >= series.shift(1)  # True if rising
    eq = series == series.shift(1)  # True if equal
    logic_table = pd.concat([geq, eq], axis=1)

    streaks = [0]  # holds the streak duration, starts with 0

    for row in logic_table.iloc[1:].itertuples():  # iterate through logic table
        if row[2]:  # same value as before
            streaks.append(0)
            continue
        last_value = streaks[-1]
        if row[1]:  # higher value than before
            streaks.append(last_value + 1 if last_value >= 0 else 1)  # increase or reset to +1
        else:  # lower value than before
            streaks.append(last_value - 1 if last_value < 0 else -1)  # decrease or reset to -1

    return np.array(streaks, dtype=float)
df['streaks_numpy'] = get_streaks(df['close'] )
df.tail()

then i calculate a 2 period RSI with the streak

然后我用条纹计算出2个周期的RSI

df['streak_rsi'] = rsi(df['streaks_numpy'],2)

I then calculate the Percent Rank

然后我计算百分比排名

# Define the lookback period
lookback = 100

# Calculate the one-day return
one_day_return = df['close'].pct_change()

# Calculate the number of values in the lookback period that are less than the current return
less_than_current = one_day_return.rolling(window=lookback).apply(
    lambda x: np.sum(x < x[-1]), raw=True)

# Calculate the total number of values in the lookback period
total_values = one_day_return.rolling(window=lookback).apply(
    lambda x: np.size(x), raw=True)

# Calculate the Percent Rank value
percent_rank = less_than_current / total_values
df['percent_rank'] = percent_rank

and to get thr connors rsi i add them and divide by 3

为了得到康纳斯的RSI，我把它们相加，然后除以3

# calculate the CRSI
df['CRSI'] = (df['rsi(3)'] + df['streak_rsi'] + df['percent_rank']) / 3

So the code runs but the results are wrong, the rsi calculates perfectly i manually checked the streaks output , which looked like it suppose to, leaving me with the percent rank calculation that might be wrong, so, im looking for some extra eyes to help me see where i go wrong in my logic. Thanks

所以代码运行，但结果是错误的，RSI计算完美我手动检查条纹输出，这看起来像它假设，让我与百分比排名计算可能是错误的，所以，我寻找一些额外的眼睛，以帮助我看到我在我的逻辑错误。谢谢

maybe something like this will work

也许这样的事情会奏效

a = data['Close'].to_numpy()
n = 10
percent_rank = np.array([sum(a[i + 1 - n:i + 1] < a[i]) / n for i in range(len(a))])
percent_rank[:n - 1] = np.nan