Adding markers to buying signals on plot

[andreasklippinge]

Hey all,

Firstly, thanks Daniel for the development of this package - I really like it!

I'm trying to implement markers in my plot for marking my buying/selling signals and did not quite understand/made it work from the documentation given. I have the following code:

ticker = " ^omx "
df = yf.download(ticker, period = "1y", interval = "1d", progress = False, rounding = True)
close = df["Close"].values
df["Row"] = df.reset_index().index
rows = df["Row"]
x = rows
y = close
z = np.polyfit(x, y, 1)
p = np.poly1d(z)
df["Trendline"] = p(x)
df["linearreg"] = talib.LINEARREG(df["Close"],timeperiod=14)
df["stddev"] = talib.STDDEV(df["linearreg"],timeperiod=14)*2
short = df["Trendline"] + df["stddev"]
long = df["Trendline"] - df["stddev"]
df["Go short"] = np.where((df["Close"] > short),1,0)
df["Go long"] = np.where((df["Close"] < long),1,0)
slope = round(z[0],1)

buydates,selldates = [],[]
buyprice,sellprice = [],[]

for index,row in df.iterrows():
       if row["Go short"] == 1:
        buydates.append(index)
        buyprice.append(row["Close"])
        
    if row["Go long"] == 1:
        selldates.append(index)
        sellprice.append(row["Close"])

This code works for plotting the markers on a plt.plot as:

plt.figure(figsize=(15,15))        
plt.plot(df["Close"])
plt.scatter(df.loc[buydates].index,df.loc[buydates].Close,marker="v",color="r")
plt.scatter(df.loc[selldates].index,df.loc[selldates].Close,marker="^",color="g")
plt.show()

But I can't make it work on mpf.plot. I've tried just copying plt.scatter(*kwargs) as

mpf.make_addplot(df.loc[selldates].index,df.loc[selldates].Close,marker="^",color="g")

but since the make_addplot only takes 1 positional argument, I don't know how to proceed.
So how do I create the markers in mpf.make_addplot(*kwargs) ?

Many thanks in advance.

[GeeteshMehto]

Hello sir, please assign this issue to me as i already worked on this kind of problem in past and has a great experience.

[DanielGoldfarb]

@andreasklippinge
Please read carefully and thoroughly through the entire addplot tutorial. The tutorial explains exactly how to plot signals giving and example. Key points to notice are:

1. the data passed into mpf.make_addplot() must be the same length as the data frame passed into mpf.plot(). Any points in the addplot that are not meant to be plotted should contain the value float('nan').
2. kwargs available to mpf.make_addplot() are shown throughout the tutorial by example. One can also call function mpf.kwarg_help() or mpf.kwarg_help('make_addplot').

After you have read carefully and thoroughly through the entire addplot tutorial let me know if you still have questions.

[andreasklippinge]

@DanielGoldfarb
I gave it another try and ended up with this code, which works!!

df = pd.read_csv(r'C:\\Users\Andreas\Aktieanalys\dataset\^gdaxi.csv'
                 ,index_col=0,parse_dates=True)
close = df["Close"].values
df["Row"] = df.reset_index().index
rows = df["Row"]
x = rows
y = close
z = np.polyfit(x, y, 1)
p = np.poly1d(z)
df["Trendline"] = p(x)
df["linearreg"] = talib.LINEARREG(df["Close"],timeperiod=14)
df["stddev"] = talib.STDDEV(df["linearreg"],timeperiod=14)*2
short = df["Trendline"] + df["stddev"]
long = df["Trendline"] - df["stddev"]
df["Go short"] = np.where((df["Close"] > short),1,0)
df["Go long"] = np.where((df["Close"] < long),1,0)
slope = round(z[0],1)
r2 = round(r2_score(df["Close"],p(x)),2)
print("R^2 correlation is",r2)
def signal_above_stddev(Go_short,price):
    signal = []
    for date,value in Go_short.iteritems():
        if value > 0:
            signal.append(df["High"][date]*1.01)
        else:
            signal.append(np.nan)
    return signal          
def signal_below_stddev(Go_long,price):
    signal = []
    for date, value in Go_long.iteritems():
        if value > 0:
            signal.append(df["Low"][date]*0.99)
        else:
            signal.append(np.nan)
    return signal
short_signal = signal_above_stddev(df['Go_short'], df['Close'])
long_signal = signal_below_stddev(df['Go_long'], df['Close'])
apd = [mpf.make_addplot(short_signal,type='scatter',markersize=20,marker='v',color="r"),
        mpf.make_addplot(long_signal,type='scatter',markersize=20,marker='^',color="g"),
        mpf.make_addplot(short,type='line',panel=0, color='green'),
        mpf.make_addplot(long,type='line',panel=0, color='red'),
        mpf.make_addplot(df["Trendline"], type='line',panel=0, color='orange')
        ]
mpf.plot(df, type='candle', volume=True, tight_layout=False, style='yahoo', figsize=(20,12), ylabel='Closing price', returnfig=True, 
         addplot=apd, volume_panel=1, panel_ratios=(6,2))

Now I'm off trying to make sense of the backtesting package to see how well this strategy would pay off.
Cheers!

[DanielGoldfarb]

@andreasklippinge
Glad to hear it. That's a beautiful plot you've made. Thanks for sharing it!