-
Notifications
You must be signed in to change notification settings - Fork 0
/
ts.py
291 lines (243 loc) · 11 KB
/
ts.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
import streamlit as st
import yfinance as yf
import datetime
import pandas as pd
import numpy as np
# import matplotlib.pyplot as plt
import plotly.express as px
import plotly.graph_objs as go
def run():
st.header('Guide to forecast')
stocks = ['^GSPC', 'AAPL', 'WMT', 'XOM', 'JNJ', 'KO', 'AMZN', 'BAC', 'AAL']
start_date = "2015-01-01"
end_date = "2022-05-21"
if st.checkbox('Load data'):
if 'stocks' not in st.session_state:
data = pd.DataFrame()
for stock in stocks:
d = yf.download(stock, start=start_date, end=end_date)
d = d[ ['Close'] ]
d.columns = [stock] # col rename
data = pd.concat([data, d], axis=1)
st.session_state['stocks'] = data
df = st.session_state['stocks']
if 'stocks' not in st.session_state:
st.info('Data not yet loaded')
else:
st.success('Data loaded')
with st.expander('Show data'):
try:
st.write(df)
except:
st.write(pd.DataFrame())
st.info('Data not yet loaded')
###########################
def make_line_chart(df, title=''):
import plotly.express as px
fig = px.line(df)
fig.update_layout(title=title)
return fig
with st.expander('Show line chart'):
fig = make_line_chart(df)
st.plotly_chart(fig, use_container_width=True)
##################
def make_scatter_chart(df, xdata, ydata, title):
import plotly.express as px
fig = px.scatter(df, x=xdata, y=ydata, trendline='ols', hover_name=df.index)
fig.update_layout(title=title)
return fig
def transform_df(df, transform='no'):
choices = ['no', 'diff', 'log', 'logdiff']
if transform not in choices:
raise ValueError("'transform' only accepts ['no', 'log', 'logdiff']")
if transform == 'no':
return df
data = df.copy()
if transform == 'log':
data = np.log(data)
elif transform == 'logdiff':
data = np.log(data).diff()
data.dropna(inplace=True)
return data
with st.expander('Show scatter chart'):
with st.form('Scatter'):
col1, col2 = st.columns([1,1])
with col1:
xd = st.selectbox('Select X data', stocks, index=0)
transform = st.selectbox('Select data transformation', ['no', 'log', 'logdiff'], key='t0')
with col2:
yd = st.selectbox('Select Y data', stocks, index=1)
if st.form_submit_button('Get linear relationship'):
sfig = make_scatter_chart( transform_df(df, transform=transform), xdata=xd, ydata=yd, title=f'{xd}-{yd} ({transform} transform)')
st.plotly_chart(sfig, use_container_width=True)
if 'diff' not in transform:
st.error('This scatter plot is meaningless. You got ruined by autocorrelation')
else:
st.info('Log diff is the only way to draw any meaningful linear relationship between prices.')
with st.expander('Show MA chart'):
with st.form('MA'):
col1, col2 = st.columns([1,1])
with col1:
window = st.number_input('Select Moving Average Interval', 1, 252, 21)
axis_scale = st.selectbox('Y-axis scale', ['no scale', 'standard', 'minmax'])
with col2:
transform1 = st.selectbox('Select data transformation', ['no', 'log', 'logdiff'], key='t1')
if st.form_submit_button("Get moving averages"):
rolling_mean = transform_df( df, transform1 ).rolling(window = window).mean()
rolling_std = transform_df( df, transform1 ).rolling(window = window).std() *100
tdf = transform_df(df, transform1)
tdf_minus = tdf - rolling_mean
tdf_minus.dropna(inplace=True)
def scale_df(df, scaler='standard'):
from sklearn import preprocessing
scaler_choice = ['standard', 'minmax']
if scaler not in scaler_choice:
raise ValueError("Invalid scaler type. Expected one of: %s" % scaler_choice)
if scaler == 'standard':
scaler = preprocessing.StandardScaler()
if scaler == 'minmax':
scaler = preprocessing.MinMaxScaler()
cols= df.columns
idx = df.index
# skl scaler will strip colname and index
scaled_df = scaler.fit_transform(df)
scaled_df = pd.DataFrame(scaled_df, columns=cols)
scaled_df = scaled_df.set_index(idx)
return scaled_df
rt = df['^GSPC'].pct_change().dropna()
import arch
am = arch.univariate.arch_model(
df['^GSPC'].pct_change().dropna(),
x=None, mean='HARX',
lags=0, vol='Garch',
p=1, o=0, q=1,
dist='skewt', hold_back=None, rescale=True
)
volatility_model = am.fit()
const, omega, alpha, beta, eta, lamb = volatility_model.params # Retrieve Model Parameters
garch_vol = volatility_model.conditional_volatility.round(2) * np.sqrt(252) # Retrieve conditional volatility
VL = omega / (1 - alpha - beta ) # long-term variance under GARCH
sigma_L = np.sqrt(VL) * np.sqrt(252) # long-term volatility under GARCH (convert from variance)
sample_sigma = rt.std() *np.sqrt(252) * 100 # sample volatility estimate
VIX = yf.download('^VIX', start= start_date, end= end_date)
vol_df = pd.concat([rolling_std['^GSPC'], VIX['Close'], garch_vol], axis=1)
vol_df.columns=['Actual Vol', 'Implied Vol', 'Conditional Vol']
vol_df = vol_df.dropna()
if 'no' in axis_scale:
pass
else:
vol_df = scale_df(vol_df, scaler=axis_scale)
vol_minus = pd.DataFrame()
vol_minus['Implied minus Actual'] = vol_df['Implied Vol'] - vol_df['Actual Vol']
vol_minus['Implied minus Conditional'] = vol_df['Implied Vol'] - vol_df['Conditional Vol']
vol_minus['Actual minus Conditional'] = vol_df['Actual Vol'] - vol_df['Conditional Vol']
vol_minus = vol_minus.dropna()
st.plotly_chart( make_line_chart(rolling_mean, 'MA mean'), use_container_width=True)
st.plotly_chart( make_line_chart(rolling_std, 'MA std'), use_container_width=True)
st.plotly_chart( make_line_chart(tdf_minus, 'Original - MA mean' ), use_container_width=True)
st.info('If rolling mean and rolling standard deviation is not stable with time, time series is not stationary.')
st.plotly_chart( make_line_chart(vol_df, f'Implied Vol vs Actual Vol (Y-axis={axis_scale})'), use_container_width=True)
st.plotly_chart( make_line_chart(vol_minus, f'Volatility gap (Y-axis={axis_scale})'), use_container_width=True)
# st.plotly_chart( make_scatter_chart(vol_minus, x=vol_minus.index, y=vol_minus.values, f'Volatility gap (Y-axis={axis_scale})'), use_container_width=True)
st.plotly_chart( make_KDE_plot(vol_minus, remove_outliers=False), use_container_width=True )
st.info('If implied minus actual is above 0, it means traders overpaid for insurance. If below 0, traders are under-insured.')
st.info('If actual minus conditional is above 0, it means models are too optimistic. If below 0, risk models are too pessimistic.')
with st.expander('Autocorrelation Function Plot'):
with st.form('ACF'):
col1, col2 = st.columns([1,1])
with col1:
tickersel2 = st.selectbox('Select ticker data', stocks, index=0)
with col2:
transform2 = st.selectbox('Select data transformation', ['no', 'log', 'logdiff'], key='t2')
if st.form_submit_button('Get ACF plots'):
# st.write( transform_df( df[ tickersel2 ] ) )
acfp = make_acf_plot( transform_df( df[ tickersel2 ], transform=transform2 ), plot_pacf=False)
pacfp= make_acf_plot( transform_df( df[ tickersel2 ], transform=transform2 ), plot_pacf=True)
st.plotly_chart(acfp, use_container_width=True)
st.plotly_chart(pacfp, use_container_width=True)
st.info('If ACF plot have downward trending spikes, series is not stationary. After log-diff, ACF spike signals MA(n), PACF spikes signals AR(n)')
with st.expander('Run forecast model'):
st.info('**** forecasting in Streamlit. **** most Python forecasting libraries in general. Difficult to customize and bring up to the front-end')
st.error('Dynamic forecasting wastes so much computational resources. Lack of customizibility means delivering low resolution info')
st.error('Why are you not doing it on Jupyter Notebook?')
def make_KDE_plot(resid, remove_outliers=False):
import plotly.figure_factory as ff
import numpy as np
from scipy import stats
if remove_outliers == True:
resid = resid[(np.abs(stats.zscore(resid)) < 3).all(axis=1)]
title = f'Distribution of Residuals:<br>(Z-score cutoff=3)'
elif remove_outliers != False:
resid = resid[(np.abs(stats.zscore(resid)) < remove_outliers).all(axis=1)]
title = f'Distribution of Residuals:<br>(Z-score cutoff={remove_outliers})'
else:
title = f'Distribution of Residuals: (With Outliers)'
rug_text=[]
for col in resid.columns:
rug_text.append(resid.index)
fig = ff.create_distplot(
[resid[c] for c in resid.columns],
group_labels=resid.columns,
rug_text= rug_text,
show_hist=False,
show_rug=True
)
return fig
# @st.cache()
def make_acf_plot(series, plot_pacf=False):
"""
Produces an ACF/PACF plot for a series. Lags = 42. Areas are shaded at 0.05 alpha.
series: A series (column) of a dataframe.
plot_pacf: Default = False. If true, returns a PACF plot instead of ACF plot.
"""
from statsmodels.tsa.stattools import acf, pacf
import plotly.graph_objects as go
corr_array = pacf(series.dropna(), alpha=0.05) if plot_pacf else acf(series.dropna(), alpha=0.05)
lower_y = corr_array[1][:,0] - corr_array[0]
upper_y = corr_array[1][:,1] - corr_array[0]
fig = go.Figure()
# [ fig.add_scatter(x=(x,x), y=(0,corr_array[0][x]), mode='lines', line_color='#3f3f3f') for x in range(len(corr_array[0])) ]
for x in range(len(corr_array[0])):
fig.add_scatter(x=(x,x), y=(0,corr_array[0][x]), mode='lines', line_color='#3f3f3f')
fig.add_scatter(
x=np.arange(len(corr_array[0])),
y=corr_array[0],
mode='markers',
marker_color='#1f77b4',
marker_size=8
)
fig.add_scatter(
x=np.arange(len(corr_array[0])),
y=upper_y,
mode='lines',
line_color='rgba(255,255,255,0)'
)
fig.add_scatter(
x=np.arange(len(corr_array[0])),
y=lower_y, mode='lines',
fillcolor='rgba(32, 146, 230,0.6)',
fill='tonexty',
line_color='rgba(255,255,255,0)'
)
fig.update_traces(showlegend=False)
fig.update_xaxes(
range=[-1,42],
# rangeslider_visible=True
)
fig.update_yaxes(zerolinecolor='#000000', autorange= True, fixedrange = False)
#fig.update_xaxes(showgrid=True, gridwidth=0.2, gridcolor='darkslategrey')
#fig.update_yaxes(showgrid=True, gridwidth=0.2, gridcolor='darkslategrey')
yaxis_title = 'Partial Autocorrelation' if plot_pacf else 'Autocorrelation'
title=f'Partial Autocorrelation (PACF) for "{series.name}"' if plot_pacf \
else f'Autocorrelation (ACF) for "{series.name}"'
fig.update_layout(
xaxis_title= 'Lag',
yaxis_title= yaxis_title,
title=title,
height= 300
)
return fig
#--------------------
# Run app
#--------------------
run()