def test_2(df):
# One manually picked candlestick pattern
patterns = [['inverted_hammer', 12, 13]]
# 'bu_' infornt of the name paints it green, 'be_' red
# Example: 'bu_inverted_hammer'
# And we add some Support and Resistance lines
lines = [{
'start': [3.5, 0.711],
'stop': [20, 0.711]
}, {
'start': [10, 0.688],
'stop': [20, 0.688]
}]
mfe.plot_candlestick(
data=df,
name='BTC_XRP_5min',
cs_patterns=patterns,
columns=['bband_upper_20', 'bband_Lower_20', 'MA_36', 'EMA_8'],
draw_verticals=False,
lines=lines,
# save='BTC_XRP_5min_candlestick.png'
)
def example_1(data, signals):
# Example 1)
# In this example we create a fig and an axis and
# plot the candlestick chart plus some indicators
# and the rsi on a seperate axis:
# Create fig
fig, _ = plt.subplots(facecolor=mfe.background_color)
# Create first axis
ax0 = plt.subplot2grid((8, 4), (0, 0),
rowspan=4,
colspan=4,
facecolor=mfe.background_color)
# Add content to the first axis.
# 1) Candlestick chart
# Alternative you can use plot_candlestick()
# instead of plot_filled_ohlc().
# The functionality is the same
# 2) Bollinger Bands 20
# 3) Moving Average 36
# 4) Exponential Moving Average 8
_, _ = mfe.plot_candlestick(
fig=fig,
axis=ax0,
data=data,
name='BTC_XRP_5min',
signals=signals,
plot_columns=['bband_upper_20', 'bband_lower_20', 'MA_36', 'EMA_8'],
draw_verticals=False,
signal_evaluation=True,
signal_evaluation_form='rectangle', # 'arrow_1'
disable_x_ticks=True,
)
# Create second axis
ax1 = plt.subplot2grid((8, 4), (4, 0),
rowspan=4,
colspan=4,
sharex=ax0,
facecolor=mfe.background_color)
# Add content to the second axis.
# 1) RSI 14
mfe.plot(data=data,
name='Chart 2',
plot_columns=['RSI_14'],
axis=ax1,
fig=fig,
xhline_red=0.8,
xhline_green=0.2,
gradient_fill=True)
plt.show()
def example_3(data, cs_patterns):
# Example 2)
# Alternative to the first example we dont use
# any given figure and axis.
# If no axis is provided it will plt.show the
# chart automatically. To disable this set show=False.
# To save the chart set save='path_to_png.png'
fig, ax = mfe.plot_candlestick(
data=data,
name='BTC_XRP_5min',
cs_patterns=cs_patterns,
plot_columns=['bband_upper_20', 'bband_lower_20', 'MA_36', 'EMA_8'],
draw_verticals=False,
# save='BTC_XRP_5min_candlestick.png'
)
def main():
# 1) Arrange data -------------------------------------------------
# Load dataset (max 120000 datapoints)
data = pd.read_csv('BTC_XRP_5min.csv', index_col=0).tail(5000)
data = data.drop(['date', 'quoteVolume', 'volume', 'weightedAverage'], 1)
# Calculate indicators
data = relative_strength_index(df=data, n=14)
data = bollinger_bands(df=data, n=20, std=4, add_ave=False)
data = exponential_moving_average(df=data, n=10)
data = moving_average(df=data, n=12)
data = macd(df=data, n_fast=12, n_slow=26)
# Cut DataFrame
data = data.iloc[40::]
# Reset index
data = data.reset_index()
# Delete old index
data = data.drop('index', 1)
# Normalize data
data_n = (data - data.mean()) / (data.max() - data.min())
# 2) RNN ----------------------------------------------------------
# Parameters
batch_size = 3
test_dataset_size = 0.05
num_units = 12
learning_rate = 0.001
epochs = 8
# Which names are avaiable? print(list(data_n))
features = ['MA_12', 'MACD_12_26']
dataset_train_length = len(data_n.index) -\
int(len(data_n.index) * test_dataset_size)
training_data = data_n.iloc[:dataset_train_length]
# Train and test the RNN
predicted_data = network(data_n, features, batch_size,
dataset_train_length, num_units, learning_rate,
epochs)
# Append test close data and the predicted data
test_close = pd.DataFrame(data_n['close'][dataset_train_length::])
df = pd.concat([training_data, predicted_data, test_close])
# 3) Plot ---------------------------------------------------------
fig, _ = plt.subplots(facecolor=mfe.background_color)
ax0 = plt.subplot2grid((10, 8), (0, 0),
rowspan=6,
colspan=8,
facecolor=mfe.background_color)
mfe.plot_candlestick(
fig=fig,
axis=ax0,
data=df,
plot_columns=[
'bband_upper_20', 'bband_lower_20', 'MA_12', 'predicted', 'close'
],
vline=dataset_train_length - 1,
vspan=[dataset_train_length - 1,
len(data_n.index)],
)
tracking_error = np.std(predicted_data['predicted'] -
data_n['close'][dataset_train_length::]) * 100
print('Tracking_error: ' + str(tracking_error))
# Plot RSI
ax1 = plt.subplot2grid((10, 8), (6, 0),
rowspan=2,
colspan=8,
sharex=ax0,
facecolor=mfe.background_color)
mfe.plot(data=data_n,
name='RSI_14',
plot_columns=['RSI_14'],
axis=ax1,
fig=fig,
xhline_red=0.8,
xhline_green=0.2,
vline=dataset_train_length - 1,
vspan=[dataset_train_length - 1,
len(data_n.index)])
# Plot MACD
ax1 = plt.subplot2grid((10, 8), (8, 0),
rowspan=2,
colspan=8,
sharex=ax0,
facecolor=mfe.background_color)
mfe.plot(data=data_n,
name='MACD',
plot_columns=['MACD_12_26', 'MACDsign_12_26', 'MACDdiff_12_26'],
axis=ax1,
fig=fig,
xhline_dashed1=0,
vline=dataset_train_length - 1,
vspan=[dataset_train_length - 1,
len(data_n.index)])
plt.subplots_adjust(left=.07,
bottom=.05,
right=.94,
top=.96,
hspace=0.2,
wspace=0.03)
plt.show()
def test_1(df):
# Example 1)
# In this example we plot the candlestick chart plus
# some indicators and rsi on a seperate axis:
# Structure: [(signal, index, price), ... ].
# Signal can be 'BUY' or 'SELL'
signals = [
('BUY', 12, 0.69),
('SELL', 27, 0.7028),
('BUY', 56, 0.6563),
('SELL', 81, 0.6854),
('BUY', 106, 0.665),
('SELL', 165, 0.640),
('BUY', 183, 0.66),
('SELL', 202, 0.7063),
]
# Create fig
fig, _ = plt.subplots(facecolor=mfe.cb)
# Create first axis
ax0 = plt.subplot2grid((8, 4), (0, 0),
rowspan=4,
colspan=4,
facecolor=mfe.cb)
# Add content to the first axis.
mfe.plot_candlestick(
fig=fig,
axis=ax0,
data=df,
title='BTC_XRP_5min',
signals=signals,
columns=['bband_upper_20', 'bband_Lower_20', 'MA_36', 'EMA_8'],
disable_xticks=False,
)
# Just to show what is possible
# mfe.plot_volume(
# twin_axis=ax0,
# fig=fig,
# data=df
# )
# Create second axis
ax1 = plt.subplot2grid((8, 4), (4, 0),
rowspan=4,
colspan=4,
sharex=ax0,
facecolor=mfe.cb)
# Add content to the second axis.
# 1) RSI 14
mfe.plot(axis=ax1,
fig=fig,
data=df,
name='RSI 2',
columns=['RSI_14'],
hlines=[{
'index': 0.8,
'color': 'red'
}, {
'index': 0.2,
'color': 'green'
}])
plt.show()