def merge():
df_open = ipo_lab.load_data("./ipo_open.csv")
sdf_open = df_open.sort_values(by=['symbol'])
with open("./ipo5.csv") as f:
line = f.readline().strip()
ls = line.split(',')
ls.insert(1, "date")
print ','.join(ls)
line = f.readline().strip()
while line:
ls = line.split(',')
symbol = ls[0]
d = df_open.loc[df_open["symbol"] == symbol].iloc[0, 0]
if d is None:
print "--------- missing", symbol
continue
ls.insert(1, str(d))
print ','.join(ls)
line = f.readline().strip()
print line
exit()
df = ipo_lab.load_data("./ipo.csv")
sdf = df.sort_values(by=["symbol"])
print "symbol,date,y,x1,x2,x3,x4,x5,underwriter,exchange"
j = 0
for i in range(len(sdf_open)):
row_open = sdf_open.iloc[i]
row = sdf.iloc[j]
while row["symbol"] < row_open["symbol"]:
# print "missing in ipo_open.csv", row["symbol"], row_open["symbol"]
j += 1
row = sdf.iloc[j]
if row["symbol"] > row_open["symbol"]:
# print "missing in ipo.csv", row["symbol"], row_open["symbol"]
continue
line = []
line.append(row["symbol"])
line.append(str(row_open["date"]))
line.append("%.3f" % row["y"])
line.append("%.3f" % row["x1"])
line.append(str(row["x2"]))
line.append("%.3f" % row["x3"])
line.append("%.3f" % row["x4"])
line.append(str(row_open["minutes"]))
line.append(str(row["underwriter"]))
line.append(str(row["exchange"]))
print ','.join(line)
j += 1
def detect_overlap():
df = ipo_lab.load_data("./ipo_open.csv")
sdf = df.sort_values(by=['symbol'])
prev_sym = ""
for i, row in sdf.iterrows():
if prev_sym == "":
prev_sym = row["symbol"]
else:
if prev_sym == row["symbol"]:
print row["symbol"], i
prev_sym = row["symbol"]
def plot_y_vs_luw():
df = ipo_lab.load_data("./ipo.csv")
id, num, p = ipo_lab.luw_dist(df, 1.02, 25)
print id, num, p
x = []
sum = 0
for n in num:
x.append(sum + n)
sum += n
plt.step(x, p)
plt.show()
def plot_y_vs_x1_v101_104():
df = ipo_lab.load_data("./ipo.csv")
fig, axs = plt.subplots(2, 2, sharey=True, tight_layout=True)
for i in range(101, 105):
v = i / 100.0
x, y = ipo_lab.histo(df, "x1", 20, v, 1)
l = "vavle = %.2f" % v
m = i - 101
axs[m / 2][m % 2].step(x, y)
axs[m / 2][m % 2].set_title(l)
plt.show()
def print_y_vs_exchange():
df = ipo_lab.load_data("./ipo.csv")
ex_res = ipo_lab.exchange_dist(df, 1.02)
sum = 0
p = 0
exchange = ['NASDAQ', 'NYSE', 'AMEX']
for i in range(len(ex_res)):
r = ex_res[i]
print exchange[i], r[1] * 1.0 / r[0], r[0]
sum += r[0]
p += r[1]
print "TOTAL", p * 1.0 / sum, sum
def plot_y_vs_x1_4_v102():
df = ipo_lab.load_data("./out_bad.csv")
fig, axs = plt.subplots(2, 2, sharey=True, tight_layout=True)
cols = ['x1', 'x2', 'x3', 'x4']
for i in range(4):
x, y = ipo_lab.histo(df, cols[i], 20, 1.01, 1)
print 'distribution of', cols[i]
for l in range(len(x)):
print x[l], y[l]
axs[i / 2][i % 2].step(x, y)
axs[i / 2][i % 2].set_title('y/' + cols[i])
plt.show()
def plot_x1_4_dist():
df = ipo_lab.load_data("./ipo.csv")
fig, axs = plt.subplots(2, 2, sharey=True, tight_layout=True)
cols = ['x1', 'x2', 'x3', 'x4']
ends = [50, 80, 30, 50]
# ylims = [1200, 1200, 300, 300]
for i in range(4):
x, y, step = ipo_lab.distribution(df, cols[i], 100)
axs[i / 2][i % 2].bar(x[:100 - ends[i]], y[:100 - ends[i]], width=step)
axs[i / 2][i % 2].set_title('distribution of ' + cols[i])
# axs[i/2][i%2].set_ylim(0, ylims[i])
plt.show()
def print_min_dist():
dfm = ipo_lab.load_data("./ipo3.csv")
bin = [0] * 11
for i, row in dfm.iterrows():
if 0 <= row["t10"] <= 11:
bin[row["t10"] - 1] += 1
if 10 <= row["t10"] <= 11:
if row["p30"] < row["y"]:
print row["symbol"], row["y"], row["t10"], row["p30"], row[
"t30"]
else:
#print "unexpected minutes", row["symbol"], row["t10"]
continue
def plot_y_dist():
df = ipo_lab.load_data("./ipo5.csv")
x, y, step = ipo_lab.distribution(df, 'p1m', 100)
print len(x)
plt.bar(x[:40], y[:40], width=step)
plt.show()
exchange = ['NASDAQ', 'NYSE', 'AMEX']
for i in range(len(ex_res)):
r = ex_res[i]
print exchange[i], r[1] * 1.0 / r[0], r[0]
sum += r[0]
p += r[1]
print "TOTAL", p * 1.0 / sum, sum
#def dist_t10():
# df = ipo_lab.load_data("./ipo5.csv")
# for i, row in df.iterrows():
#merge()
#exit()
#plot_y_dist()
#plot_x1_4_dist()
#plot_y_vs_x1_4_v102()
#plot_y_vs_x1_v101_104()
#plot_y_vs_luw()
#print_y_vs_exchange()
#exit()
#detect_overlap()
#print_min_dist()
df = ipo_lab.load_data("./ipo5.csv")
#ipo_lab.simple_filter(df, 1.02)
for i in range(102, 115):
ipo_lab.simple_strat_2(df, "pc10", 1.0001, i / 100.0, range(2010, 2019))
#svm(1000, 1.02)