Why hammer candle signal works, but shooting star does not?

I'm testing different trading signals in Python. To avoid overfitting, I use daily time series of stocks, bonds, and commodities.

I found that the hammer candle works great as expected, but the shooting star does not. The hammer candle works best when the price is above the moving average. However, the shooting star does not work under any conditions.

If a shooting star is detected, then it is profitable to take a long position. This seems very strange to me. Can someone explain why is it so?

Code:

def check_shooting_star_candle(
    yesterday_high: float,
    yesterday_low: float,
    yesterday_close: float,
    today_high: float,
    today_low: float,
    today_open: float,
    today_close: float,
) -> bool:
    if (
        np.isnan(yesterday_high)
        or np.isnan(yesterday_low)
        or np.isnan(yesterday_close)
        or np.isnan(today_high)
        or np.isnan(today_low)
        or np.isnan(today_open)
        or np.isnan(today_close)
    ):
        return False
    if today_close > yesterday_close:
        return False
    if today_close > today_open:
        return False
    yesterday_high_low = yesterday_high - yesterday_low
    if today_high < (yesterday_high + (yesterday_high_low * 0.07)):
        return False
    today_high_low = today_high - today_low
    if (today_high - today_open) < (today_high_low * 0.75):
        return False

    # today close near low
    if (today_close - today_low) > (today_high_low * 0.15):
        return False

    # today close not too low to have potential to go lower in subsequent days
    if (yesterday_close - today_close) > (yesterday_high_low * 0.2):
        return False

    return True

 

When the market's psycho turns negative... like in a bear... you will see the opposite.

 

Your code is not the correct definition.you need consider body size and context and the position comparing previous bar. Maybe together volume.

 

You will probably find that your system is very robust in an uptrending market. Why would that be?