def yield_rate_plot(data: pd.DataFrame, stocks: []):
"""
收益率图表
:param data: 数据
:param stocks: 股票
:return:
"""
time = data.loc[(data["symbol"] == stocks[0])]['time']
time = timestamp_2_date(time.tolist())
min_date = time[0]
max_date = time[-1]
df = pd.DataFrame(index=time[1:])
for stock in stocks:
stock_data = data.loc[(data['symbol'] == stock)]
logging.info('%s data size %s', stock, len(stock_data))
ln_return_rate = log_yield_rate(list(stock_data['close']))
df[stock] = np.cumsum(ln_return_rate)
print(np.cumsum(ln_return_rate))
g = sns.lineplot(data=df)
# X轴刻度设置
g.format_xdata = dates.AutoDateFormatter(dates.MonthLocator())
plt.legend()
plt.title('ETF累计收益率({0} - {1})'.format(date_2_month(min_date),
date_2_month(max_date)))
plt.show()
def week_line_overlay_plot(data: pd.DataFrame, stocks: []):
time = data.loc[(data["symbol"] == stocks[0])]['time']
time = timestamp_2_date(time)
df = pd.DataFrame(index=time)
for stock in stocks:
stock_data = data.loc[(data["symbol"] == stock)]
df[stock] = normalize(stock_data, 'close')
g = sns.lineplot(data=df)
# X轴刻度设置
g.format_xdata = dates.AutoDateFormatter(dates.WeekdayLocator())
plt.legend()
plt.show()
def linear_regression_plot(data: pd.DataFrame, stocks: [], base_stock: str):
"""
线性回归图
:param data: 数据
:param stocks: 票代码
:return: 基准票代码
"""
spy_data = data.loc[(data['symbol'] == base_stock)]
return_spy = spy_data['close'].pct_change().dropna()
time = timestamp_2_date(spy_data['time'].tolist())
min_date = time[0]
max_date = time[-1]
m, n = subplot_num(len(stocks) - 1)
fig = plt.figure()
idx = 1
for stock in stocks:
if stock == base_stock:
continue
stock_data = data.loc[(data["symbol"] == stock)]
logging.info("%s data size %s", stock, len(stock_data))
return_stock = list(stock_data['close'].pct_change().dropna())
alpha_spy, beta_spy = alpha_beta(return_spy, return_stock)
x2 = np.linspace(return_spy.min(), return_spy.max(), 100)
y_hat = x2 * beta_spy + alpha_spy
ax = fig.add_subplot(m, n, idx)
ax.scatter(list(return_spy), return_stock, alpha=0.3)
ax.plot(x2, y_hat, alpha=0.9)
plt.xlabel('{0} Return'.format(base_stock))
plt.ylabel('{0} Return'.format(stock))
idx += 1
plt.suptitle('以{0}为基准的线性回归({1} - {2})'.format(base_stock,
date_2_month(min_date),
date_2_month(max_date)))
plt.show()
def line_overlay_plot(data: pd.DataFrame, stocks: [],
plotDateType: PlotDateType):
"""
拆线叠加图
:param data:
:param stocks:
:param plotDateType:
:return:
"""
time = data.loc[(data["symbol"] == stocks[0])]['time']
time = timestamp_2_date(time.tolist())
min_date = time[0]
max_date = time[-1]
df = pd.DataFrame(index=time)
for stock in stocks:
stock_data = data.loc[(data["symbol"] == stock)]
# logging.info(timestamp_2_month(stock_data['time']))
stock_data = list(normalize(stock_data, 'close'))
logging.info('%s %s', stock, len(stock_data))
df[stock] = stock_data
g = sns.lineplot(data=df)
# X轴刻度设置
date_locator = None
if plotDateType == PlotDateType.MONTH:
date_locator = dates.AutoDateFormatter(dates.MonthLocator())
elif plotDateType == PlotDateType.WEEK:
date_locator = dates.AutoDateFormatter(dates.WeekdayLocator())
elif plotDateType == PlotDateType.DAY:
date_locator = dates.AutoDateFormatter(dates.DateLocator())
g.format_xdata = date_locator
# 鼠标hover 事件
# plt.gcf().canvas.mpl_connect('motion_notify_event', onpick)
plt.title('ETF价格叠加图({0} - {1})'.format(date_2_month(min_date),
date_2_month(max_date)))
plt.legend()
plt.show()
def correlation_coefficient_plot(data: pd.DataFrame, stocks: []):
"""
计算关联系数
:param data:
:param stocks:
:return:
"""
time = data.loc[(data["symbol"] == stocks[0])]['time']
time = timestamp_2_date(time.tolist())
min_date = time[0]
max_date = time[-1]
df = pd.DataFrame()
for stock in stocks:
stock_data = data.loc[(data["symbol"] == stock)]
df[stock] = stock_data['close'].values
corr = df.corr()
plt.figure(figsize=(12, 10))
foo = sns.heatmap(corr, vmax=0.6, square=True, annot=True)
plt.show()