def print_industry_coer(fund_ts, ostream):
"""
@summary prints standard deviation of returns for a fund
@param fund_ts: pandas fund time series
@param years: list of years to print out
@param ostream: stream to print to
"""
industries = [['$DJUSBM', 'Materials'],
['$DJUSNC', 'Goods'],
['$DJUSCY', 'Services'],
['$DJUSFN', 'Financials'],
['$DJUSHC', 'Health'],
['$DJUSIN', 'Industrial'],
['$DJUSEN', 'Oil & Gas'],
['$DJUSTC', 'Technology'],
['$DJUSTL', 'TeleComm'],
['$DJUSUT', 'Utilities']]
for i in range(0, len(industries) ):
if(i%2==0):
ostream.write("\n")
#load data
norObj = de.DataAccess('Yahoo')
ldtTimestamps = du.getNYSEdays( fund_ts.index[0], fund_ts.index[-1], dt.timedelta(hours=16) )
ldfData = norObj.get_data( ldtTimestamps, [industries[i][0]], ['close'] )
#get corelation
ldfData[0]=ldfData[0].fillna(method='pad')
ldfData[0]=ldfData[0].fillna(method='bfill')
a=np.corrcoef(np.ravel(tsu.daily(ldfData[0][industries[i][0]])),np.ravel(tsu.daily(fund_ts.values)))
b=np.ravel(tsu.daily(ldfData[0][industries[i][0]]))
f=np.ravel(tsu.daily(fund_ts))
fBeta, unused = np.polyfit(b,f,1)
ostream.write("%10s(%s):%+6.2f, %+6.2f " % (industries[i][1], industries[i][0], a[0,1], fBeta))
def fundsAnalysisToPNG(funds,output_file):
plt.clf()
if(type(funds)!=type(list())):
print 'fundsmatrix only contains one timeseries, not able to analyze.'
#convert to daily returns
count=list()
dates=list()
sum=list()
for i in range(0,len(funds)):
ret=tsu.daily(funds[i].values)
for j in range(0, len(ret)):
if (funds[i].index[j] in dates):
sum[dates.index(funds[i].index[j])]+=ret[j]
count[dates.index(funds[i].index[j])]+=1
else:
dates.append(funds[i].index[j])
count.append(1)
sum.append(ret[j])
#compute average
tot_ret=deepcopy(sum)
for i in range(0,len(sum)):
tot_ret[i]=sum[i]/count[i]
#compute std
std=zeros(len(sum))
for i in range(0,len(funds)):
temp=tsu.daily(funds[i].values)
for j in range(0,len(temp)):
std[dates.index(funds[i].index[j])]=0
std[dates.index(funds[i].index[j])]+=math.pow(temp[j]-tot_ret[dates.index(funds[i].index[j])],2)
for i in range(1, len(std)):
# std[i]=math.sqrt(std[i]/count[i])+std[i-1]
std[i]=math.sqrt(std[i]/count[i])
#compute total returns
lower=deepcopy(tot_ret)
upper=deepcopy(tot_ret)
tot_ret[0]=funds[0].values[0]
lower[0]=funds[0].values[0]
upper[0]=lower[0]
# for i in range(1,len(tot_ret)):
# tot_ret[i]=tot_ret[i-1]+(tot_ret[i])*tot_ret[i-1]
# lower[i]=tot_ret[i-1]-(std[i])*tot_ret[i-1]
# upper[i]=tot_ret[i-1]+(std[i])*tot_ret[i-1]
for i in range(1,len(tot_ret)):
lower[i]=(tot_ret[i]-std[i]+1)*lower[i-1]
upper[i]=(tot_ret[i]+std[i]+1)*upper[i-1]
tot_ret[i]=(tot_ret[i]+1)*tot_ret[i-1]
plt.clf()
plt.plot(dates,tot_ret)
plt.plot(dates,lower)
plt.plot(dates,upper)
plt.legend(('Tot_Ret','Lower','Upper'),loc='upper left')
plt.ylabel('Fund Total Return')
plt.ylim(ymin=0,ymax=2*tot_ret[0])
plt.draw()
savefig(output_file, format='png')
def print_other_coer(fund_ts, ostream):
"""
@summary prints standard deviation of returns for a fund
@param fund_ts: pandas fund time series
@param years: list of years to print out
@param ostream: stream to print to
"""
industries = [['$SPX', ' S&P Index'],
['$DJI', ' Dow Jones'],
['$DJUSEN', 'Oil & Gas'],
['$DJGSP', ' Metals']]
for i in range(0, len(industries) ):
if(i%2==0):
ostream.write("\n")
#load data
norObj =de.DataAccess('Yahoo')
ldtTimestamps = du.getNYSEdays( fund_ts.index[0], fund_ts.index[-1], dt.timedelta(hours=16) )
ldfData = norObj.get_data( ldtTimestamps, [industries[i][0]], ['close'] )
#get corelation
ldfData[0]=ldfData[0].fillna(method='pad')
ldfData[0]=ldfData[0].fillna(method='bfill')
a=np.corrcoef(np.ravel(tsu.daily(ldfData[0][industries[i][0]])),np.ravel(tsu.daily(fund_ts.values)))
b=np.ravel(tsu.daily(ldfData[0][industries[i][0]]))
f=np.ravel(tsu.daily(fund_ts))
fBeta, unused = np.polyfit(b,f,1)
ostream.write("%10s(%s):%+6.2f, %+6.2f " % (industries[i][1], industries[i][0], a[0,1], fBeta))
def fundsAnalysisToPNG(funds, output_file):
plt.clf()
if (type(funds) != type(list())):
print 'fundsmatrix only contains one timeseries, not able to analyze.'
#convert to daily returns
count = list()
dates = list()
sum = list()
for i in range(0, len(funds)):
ret = tsu.daily(funds[i].values)
for j in range(0, len(ret)):
if (funds[i].index[j] in dates):
sum[dates.index(funds[i].index[j])] += ret[j]
count[dates.index(funds[i].index[j])] += 1
else:
dates.append(funds[i].index[j])
count.append(1)
sum.append(ret[j])
#compute average
tot_ret = deepcopy(sum)
for i in range(0, len(sum)):
tot_ret[i] = sum[i] / count[i]
#compute std
std = zeros(len(sum))
for i in range(0, len(funds)):
temp = tsu.daily(funds[i].values)
for j in range(0, len(temp)):
std[dates.index(funds[i].index[j])] = 0
std[dates.index(funds[i].index[j])] += math.pow(
temp[j] - tot_ret[dates.index(funds[i].index[j])], 2)
for i in range(1, len(std)):
# std[i]=math.sqrt(std[i]/count[i])+std[i-1]
std[i] = math.sqrt(std[i] / count[i])
#compute total returns
lower = deepcopy(tot_ret)
upper = deepcopy(tot_ret)
tot_ret[0] = funds[0].values[0]
lower[0] = funds[0].values[0]
upper[0] = lower[0]
# for i in range(1,len(tot_ret)):
# tot_ret[i]=tot_ret[i-1]+(tot_ret[i])*tot_ret[i-1]
# lower[i]=tot_ret[i-1]-(std[i])*tot_ret[i-1]
# upper[i]=tot_ret[i-1]+(std[i])*tot_ret[i-1]
for i in range(1, len(tot_ret)):
lower[i] = (tot_ret[i] - std[i] + 1) * lower[i - 1]
upper[i] = (tot_ret[i] + std[i] + 1) * upper[i - 1]
tot_ret[i] = (tot_ret[i] + 1) * tot_ret[i - 1]
plt.clf()
plt.plot(dates, tot_ret)
plt.plot(dates, lower)
plt.plot(dates, upper)
plt.legend(('Tot_Ret', 'Lower', 'Upper'), loc='upper left')
plt.ylabel('Fund Total Return')
plt.ylim(ymin=0, ymax=2 * tot_ret[0])
plt.draw()
savefig(output_file, format='png')
def get_winning_days(fund_ts):
"""
@summary Returns percentage of winning days in fund time series
@param fund_ts: pandas time series of daily fund values
@return Percentage of winning days over fund time series
"""
return tsu.get_winning_days(tsu.daily(fund_ts))
def get_sharpe_ratio(fund_ts):
"""
@summary Returns daily computed Sharpe ratio of fund time series
@param fund_ts: pandas time series of daily fund values
@return Sharpe ratio of fund time series
"""
return tsu.get_sharpe_ratio(tsu.daily(fund_ts))
def simulate(startdate, enddate, symbols, allocation, type):
timeofday = dt.timedelta(hours=16)
timestamps = du.getNYSEdays(startdate, enddate, timeofday)
c_dataobj = da.DataAccess('Yahoo')
ls_keys = [type]
ldf_data = c_dataobj.get_data(timestamps, symbols, ls_keys)
d_data = dict(zip(ls_keys, ldf_data))
#print d_data.values
na_price = d_data[type].values
print 'price', na_price[0], na_price[-1]
#print 'na', na_price
#print 'na_price', na_price
normalized_price = na_price / na_price[0, :]
#print 'norm', normalized_price
c = normalized_price * allocation
#print 'n*a', allocation, c
invest = c.sum(axis=1)
#print 'c', c
#print 'invesi', invest
daily_returns = tsu.daily(invest)
mean_daily_return = np.mean(daily_returns)
stdev_daily_return = np.std(daily_returns)
sharpe_ratio = tsu.get_sharpe_ratio(daily_returns)
return stdev_daily_return, mean_daily_return, sharpe_ratio, invest[-1]
def get_sharpe_ratio(fund_ts):
"""
@summary Returns daily computed Sharpe ratio of fund time series
@param fund_ts: pandas time series of daily fund values
@return Sharpe ratio of fund time series
"""
return tsu.get_sharpe_ratio(tsu.daily(fund_ts))
def get_winning_days(fund_ts):
"""
@summary Returns percentage of winning days in fund time series
@param fund_ts: pandas time series of daily fund values
@return Percentage of winning days over fund time series
"""
return tsu.get_winning_days(tsu.daily(fund_ts))
def print_benchmark_coer(fund_ts, benchmark_close, sym, ostream):
"""
@summary prints standard deviation of returns for a fund
@param fund_ts: pandas fund time series
@param years: list of years to print out
@param ostream: stream to print to
"""
fund_ts=fund_ts.fillna(method='pad')
fund_ts=fund_ts.fillna(method='bfill')
benchmark_close=benchmark_close.fillna(method='pad')
benchmark_close=benchmark_close.fillna(method='bfill')
faCorr=np.corrcoef(np.ravel(tsu.daily(fund_ts.values)),np.ravel(tsu.daily(benchmark_close)));
b=np.ravel(tsu.daily(benchmark_close))
f=np.ravel(tsu.daily(fund_ts))
fBeta, unused = np.polyfit(b,f, 1);
print_line(sym+"Correlattion","%+6.2f" % faCorr[0,1],i_spacing=3,ostream=ostream)
print_line(sym+"Beta","%+6.2f" % fBeta,i_spacing=3,ostream=ostream)
def get_std_dev(fund_ts):
"""
@summary gets standard deviation of returns for a fund as a string
@param fund_ts: pandas fund time series
@param years: list of years to print out
@param ostream: stream to print to
"""
fund_ts=fund_ts.fillna(method='pad')
fund_ts=fund_ts.fillna(method='bfill')
ret=np.std(tsu.daily(fund_ts.values))*10000
return ("%+7.2f bps " % ret)
def getWinningDays(funds1,funds2,year): days=[] i=0; win=0 tot=0 f1ret=tsu.daily(funds1) f2ret=tsu.daily(funds2) relf1=[] relf2=[] for date in funds1.index: if(date.year==year): for date2 in funds2.index: if(date==date2): relf1.append(f1ret[i]) relf2.append(f2ret[i]) i+=1 for i in range(0,len(relf1)): if(f1ret[i]>f2ret[i]): win+=1 tot+=1 return float(win)/tot
def simulate(distro, data): prices = data['close'].values normal_prices = normalize(prices) adj_values = alloc_adj(normal_prices.copy(), distro) daily_values = daily_value(adj_values) daily_rets = tsu.daily(daily_values) volatility = np.std(daily_rets) sharpe = tsu.get_sharpe_ratio(daily_rets)[0] cum_ret = np.sum(daily_rets, axis=0)[0] avg_daily_ret = cum_ret / len(daily_rets) return daily_rets, volatility, avg_daily_ret, sharpe, (cum_ret + 1)
def run(values, symbol): #data = pd.read_csv(values, header=False) data = np.loadtxt(values, delimiter=';', skiprows=0) values = data[:,3] #values = data.values[:,3] #invest = values.sum(axis=1) normalized_price = values / values[0] daily_returns = tsu.daily(normalized_price) mean_daily_return = np.mean(daily_returns) stdev_daily_return = np.std(daily_returns) sharpe_ratio = tsu.get_sharpe_ratio(daily_returns) total_return = values[-1] / values[0] print 'returns', sharpe_ratio, total_return, stdev_daily_return, mean_daily_return startdate = dt.datetime(int(data[0][0]), int(data[0][1]), int(data[0][2]), 16) enddate = dt.datetime(int(data[-1][0]), int(data[-1][1]), int(data[-1][2]), 16) allocation = [1.0] print 'SPX', startdate, enddate, symbol, allocation result = sm.simulate(startdate, enddate, [symbol], allocation, 'close') print 'result', result[2], result[3], result[0], result[1]
def run(values, symbol): #data = pd.read_csv(values, header=False) data = np.loadtxt(values, delimiter=';', skiprows=0) #print 'Data', data values = data[:,3] #print 'Values', values #values = data.values[:,3] #invest = values.sum(axis=1) normalized_price = values / values[0] daily_returns = tsu.daily(normalized_price) mean_daily_return = np.mean(daily_returns) stdev_daily_return = np.std(daily_returns) sharpe_ratio = tsu.get_sharpe_ratio(daily_returns) total_return = values[-1] / values[0] print 'Total', values[-1], values[0] # print 'returns', sharpe_ratio, total_return, stdev_daily_return, mean_daily_return startdate = dt.datetime(int(data[0][0]), int(data[0][1]), int(data[0][2]), 16) enddate = dt.datetime(int(data[-1][0]), int(data[-1][1]), int(data[-1][2]), 16) allocation = [1.0] # print 'SPX', startdate, enddate, symbol, allocation result = sm.simulate(startdate, enddate, [symbol], allocation, 'close') # print 'result', result[2], result[3], result[0], result[1] f = data[-1] print 'The final value of the portfolio using the sample file is -- %d-%d-%d -> %.2f ' % (f[0], f[1], f[2], f[3]) #2009,12,28,54824.0 print '' print 'Details of the Performance of the portfolio' print '' print 'Data Range : %d-%d-%d to %d-%d-%d' % (data[0][0], data[0][1], data[0][2], data[-1][0], data[-1][1], data[-1][2]) print '' print 'Sharpe Ratio of Fund : %.12f' % (sharpe_ratio)# 0.527865227084 print 'Sharpe Ratio of $SPX : %.12f'% (result[2])#-0.184202673931 print '' print 'Total Return of Fund %.12f:' % (total_return)# 1.09648 print 'Total Return of $SPX %.12f:' % (result[3])# 0.779305674563' print '' print 'Standard Deviation of Fund : %.12f' % stdev_daily_return# 0.0060854156452 print 'Standard Deviation of $SPX : %.12f' % result[0]#0.022004631521' print '' print 'Average Daily Return of Fund : %.12f' % mean_daily_return# 0.000202354576186 print 'Average Daily Return of $SPX : %.12f' % result[1]#-0.000255334653467'
def print_performance( fund, bench):
print "Details of the Performance of the portfolio :\n"
print "Data Range : ", fund.index[0], " to " ,fund.index[-1],"\n"
print "Sharpe Ratio of Fund : ", tsu.get_sharpe_ratio(tsu.daily(fund))[0]
print "Sharpe Ratio of " + benchmark +" :", tsu.get_sharpe_ratio(tsu.daily(bench))[0],"\n"
print 'Total Return of Fund :', (fund[-1] / fund[0] - 1) + 1
print 'Total Return of ' + benchmark + ' : ', (bench[-1] / bench[0] -1) + 1, "\n"
print 'Standard Deviation of Fund : ' + str(np.std(tsu.daily(fund.values)))
print 'Standard Deviation of ' + benchmark + ' : ', np.std(tsu.daily(bench.values)), "\n"
print 'Average Daily Return of Fund : ', np.mean( tsu.daily(fund.values))
print 'Average Daily Return of ' + benchmark + ' : ', np.mean( tsu.daily(bench.values))
dataobj = da.DataAccess('Yahoo')
close = dataobj.get_data(timestamps, [symbol], "close", verbose=True)
close = close.fillna(method='ffill')
close = close.fillna(method='bfill')
return close[symbol]
ts_fund = _csv_read_fund(sys.argv[1])
benchmark = sys.argv[2]
bench_vals = _read_bench(benchmark, list(ts_fund.index))
# print bench_vals
multiple = ts_fund[0] / bench_vals[0]
bench_vals = bench_vals * multiple
print "Details of the Performance of the portfolio"
print 'Data Range : ', ts_fund.index[0], ' to ', ts_fund.index[-1]
print 'Sharpe Ratio of Fund :', tsu.get_sharpe_ratio(tsu.daily(ts_fund))[0]
print 'Sharpe Ratio of ' + benchmark + ' :', tsu.get_sharpe_ratio(
tsu.daily(bench_vals))[0]
print 'Total Return of Fund : ', (((ts_fund[-1] / ts_fund[0]) - 1) + 1)
print 'Total Return of ' + benchmark + ' :', (((bench_vals[-1]
/ bench_vals[0]) - 1) + 1)
print 'Standard Deviation of Fund : ', np.std(tsu.daily(
ts_fund.values))
print 'Standard Deviation of ' + benchmark + ' :', np.std(
tsu.daily(bench_vals.values))
print 'Average Daily Return of Fund : ', np.mean(tsu.daily(
ts_fund.values))
print 'Average Daily Return of ' + benchmark + ' :', np.mean(
tsu.daily(bench_vals.values))
close = dataobj.get_data(timestamps, [symbol], "close", verbose=True)
close = close.fillna(method='ffill')
close = close.fillna(method='bfill')
return close[symbol]
ts_fund = _csv_read_fund(sys.argv[1])
benchmark = sys.argv[2]
bench_vals = _read_bench(benchmark, list(ts_fund.index))
# print bench_vals
multiple = ts_fund[0] / bench_vals[0]
bench_vals = bench_vals * multiple
print "Details of the Performance of the portfolio"
print 'Data Range : ', ts_fund.index[0], ' to ', ts_fund.index[-1]
print 'Sharpe Ratio of Fund :', tsu.get_sharpe_ratio(tsu.daily(ts_fund))[0]
print 'Sharpe Ratio of ' + benchmark + ' :', tsu.get_sharpe_ratio(
tsu.daily(bench_vals))[0]
print '\n'
print 'Total Return of Fund : ', (((ts_fund[-1] / ts_fund[0]) - 1) + 1)
print 'Total Return of ' + benchmark + ' :', ((
(bench_vals[-1] / bench_vals[0]) - 1) + 1)
print '\n'
print 'Standard Deviation of Fund : ', np.std(tsu.daily(ts_fund.values))
print 'Standard Deviation of ' + benchmark + ' :', np.std(
tsu.daily(bench_vals.values))
print '\n'
print 'Average Daily Return of Fund : ', np.mean(tsu.daily(ts_fund.values))
print 'Average Daily Return of ' + benchmark + ' :', np.mean(
tsu.daily(bench_vals.values))
def run(cash, orders, values, key):
# print 'Thomas'
data = pd.read_csv(
orders,
parse_dates=True,
names=["year", "month", "day", "symbol", "transaction", "nr_shares", "empty"],
header=0,
)
data = data.sort(["year", "month", "day"])
old = data
data.to_csv("regneark.csv")
# print "Data2", data.values
# print 'Gyldig', data[(data.day == 10) & (data.month == 6) ]
# print 'Ugyldug', data[(data.day == 31) & (data.month == 6) ]
array = np.array(data)
# print array
symbols = np.unique(array[:, 3])
symbols = np.array(symbols, dtype="S").tolist()
dt_start = dt.datetime(array[0, 0], array[0, 1], array[0, 2])
dt_end = dt.datetime(array[-1, 0], array[-1, 1], array[-1, 2], 16)
print dt_start, dt_end
"Get stock prices"
timestamps = du.getNYSEdays(dt_start, dt_end, dt.timedelta(hours=16))
# print 'Ti', timestamps[-1]
dataobj = da.DataAccess("Yahoo")
data = dataobj.get_data(timestamps, symbols, key)
price = data.values
p = pd.DataFrame(price, timestamps)
# print 'price', price
dict = {}
for i in range(0, len(array)):
orderRow = array[i]
date = dt.datetime(orderRow[0], orderRow[1], orderRow[2], 16)
datacolumn = symbols.index(orderRow[3])
cell = p.ix[date][datacolumn]
# cell = dict[date][datacolumn]
quantity = orderRow[5]
if orderRow[4] == "Buy":
cash = cash - cell * quantity
# stocks[i][datacolumn] = stockcell + quantity
elif orderRow[4] == "Sell":
cash = cash + cell * quantity
# stocks[i][datacolumn] = stockcell - quantity
dict[date] = cash
# print 'cash', orderRow[4], cash, cell, quantity
# print dict
cash_array = np.zeros(len(timestamps))
for i in range(0, len(timestamps)):
k = timestamps[i]
# print k, k in dict
if k in dict:
cash_array[i] = dict[k]
else:
cash_array[i] = cash_array[i - 1]
# print cash_array[i]
# print 'Cash array', cash_array
stocks = np.zeros((len(timestamps), len(symbols)))
for i in range(0, len(timestamps)):
ts = timestamps[i]
for j in range(0, len(symbols)):
o = old[(old.day == ts.day) & (old.month == ts.month) & (old.year == ts.year) & (old.symbol == symbols[j])]
if i > 0:
stocks[i][j] = stocks[i - 1][j]
val = stocks[i][j]
if len(o) > 0:
gf = o.values.tolist()
for k in range(0, len(gf)):
row = gf[k]
if row[4] == "Buy":
val = val + row[5]
else:
val = val - row[5]
stocks[i][j] = val
# print 'stocks', stocks[:5]
# print 'prices', price[:5]
value = np.zeros((len(timestamps), 3))
# value[:,2]=cash
for i in range(0, len(timestamps)):
rowprice = 0.0
for j in range(0, len(symbols)):
st = stocks[i][j]
pr = price[i][j]
mul = st * pr
if not np.isnan(mul):
rowprice = rowprice + mul
value[i][0] = rowprice
value[i][1] = cash_array[i]
value[i][2] = rowprice + cash_array[i]
# print 'value dr', value
# print 'Vvalue', value[:,2]
# print 'stocks', stocks[5:], len(stocks)
# print 'Data', data[:5]
na_price = value[:, 0]
# print 'na', len(na_price), na_price[0]
# print 'type', type(na_price)
normalized_price = price / price[0, :]
# print 'normal', normalized_price[5:], len(normalized_price)
# invest = normalized_price.sum()
# stocks = [1.0, 1.0, 1.0, 0.0]
# print 'stocks', stocks[:5]
c = normalized_price * stocks
# print 'n*a', allocation, c
# print "c", c[:10]
# print 'normal', normalized_price[:20]
invest = c.sum(axis=1)
# print 'c', c[-5:]
# print 'invest', invest[:20]
# daily_returns = tsu.returnize0(invest)
daily_returns = tsu.daily(invest)
mean_daily_return = np.mean(daily_returns)
stdev_daily_return = np.std(daily_returns)
sharpe_ratio = tsu.get_sharpe_ratio(daily_returns)
# print 'returns', mean_daily_return, stdev_daily_return, sharpe_ratio, invest[-1]
times = np.zeros((len(timestamps), 4))
value = value.astype(np.int64)
for i in range(0, len(timestamps)):
times[i][0] = timestamps[i].year
times[i][1] = timestamps[i].month
times[i][2] = timestamps[i].day
times[i][3] = value[:, 2][i]
# print 'Times', timestamps[i].year, timestamps[i].month, timestamps[i].day, value[:,2][i]
np.savetxt(values, times, fmt="%d", delimiter=";")
def run(cash, orders, values, key):
#print 'Thomas'
data = pd.read_csv(orders, parse_dates=True, names=['year','month','day','symbol','transaction','nr_shares','empty'], header=0)
data = data.sort(['year', 'month', 'day']);
old = data
#print 'Gyldig', data[(data.day == 10) & (data.month == 6) ]
#print 'Ugyldug', data[(data.day == 31) & (data.month == 6) ]
array = np.array(data)
#print array
symbols = np.unique(array[:,3])
symbols = np.array(symbols, dtype='S').tolist()
dt_start = dt.datetime(array[0,0], array[0,1], array[0,2])
dt_end = dt.datetime(array[-1,0], array[-1,1], array[-1,2], 16)
print dt_start, dt_end
'Get stock prices'
timestamps = du.getNYSEdays(dt_start, dt_end, dt.timedelta(hours=16))
#print 'Ti', timestamps[-1]
dataobj = da.DataAccess('Yahoo')
data = dataobj.get_data(timestamps, symbols, key)
price = data.values
p = pd.DataFrame(price, timestamps)
#print 'price', price
dict = {}
for i in range(0, len(array)):
orderRow = array[i]
date = dt.datetime(orderRow[0], orderRow[1], orderRow[2], 16)
datacolumn = symbols.index(orderRow[3])
cell = p.ix[date][datacolumn]
#cell = dict[date][datacolumn]
quantity = orderRow[5]
if (orderRow[4] == 'Buy'):
cash = cash - cell * quantity
#stocks[i][datacolumn] = stockcell + quantity
elif (orderRow[4] == 'Sell'):
cash = cash + cell * quantity
#stocks[i][datacolumn] = stockcell - quantity
dict[date] = cash
# print 'cash', orderRow[4], cash, cell, quantity
#print dict
cash_array = np.zeros(len(timestamps))
for i in range(0, len(timestamps)):
k = timestamps[i]
#print k, k in dict
if (k in dict):
cash_array[i] = dict[k]
else:
cash_array[i] = cash_array[i-1]
#print cash_array[i]
#print cash_array
stocks = np.zeros((len(timestamps), len(symbols)))
for i in range(0, len(timestamps)):
ts = timestamps[i]
for j in range(0, len(symbols)):
o = old[(old.day == ts.day) & (old.month == ts.month) & (old.year == ts.year) & (old.symbol == symbols[j])]
if i > 0:
stocks[i][j] = stocks[i-1][j]
val = stocks[i][j]
if len(o) > 0:
gf = o.values.tolist()
for k in range(0, len(gf)):
row = gf[k]
if row[4] == 'Buy':
val = val + row[5]
else:
val = val - row[5]
stocks[i][j] = val
#print 'stocks', stocks
value = np.zeros((len(timestamps), 3))
for i in range(0,len(timestamps)):
rowprice = 0
for j in range(0, len(symbols)):
rowprice = rowprice + stocks[i][j] * price[i][j]
value[i][0] = rowprice
value[i][1] = cash_array[i]
value[i][2] = rowprice + cash_array[i]
#print 'value', value[:,2]
#print 'stocks', stocks[5:], len(stocks)
#print 'Data', data[:5]
na_price = value[:,0]
#print 'na', len(na_price), na_price[0]
#print 'type', type(na_price)
normalized_price = price / price[0, :]
#print 'normal', normalized_price[5:], len(normalized_price)
#invest = normalized_price.sum()
#stocks = [1.0, 1.0, 1.0, 0.0]
#print 'stocks', stocks[:5]
c = normalized_price * stocks
#print 'n*a', allocation, c
#print "c", c[:10]
#print 'normal', normalized_price[:20]
#invest = c.sum(axis=1)
invest = c.sum(axis=1)
print 'c', c#[-5:]
print 'invest', invest[:20]
#daily_returns = tsu.returnize0(invest)
daily_returns = tsu.daily(invest)
mean_daily_return = np.mean(daily_returns)
stdev_daily_return = np.std(daily_returns)
sharpe_ratio = tsu.get_sharpe_ratio(daily_returns)
print 'returns', mean_daily_return, stdev_daily_return, sharpe_ratio, invest[-1]
times = np.zeros((len(timestamps),4))
for i in range(0, len(timestamps)):
times[i][0] = timestamps[i].year
times[i][1] = timestamps[i].month
times[i][2] = timestamps[i].day
times[i][3] = value[:,2][i]
np.savetxt(values, times, fmt='%d', delimiter=';')
def plot_portfolio(fund, benchmark):
#Read the value from the funds
ldt_timestamps = []
ls_values = []
fund_csv = csv.reader(open(fund, 'rU'), delimiter=',')
for row in fund_csv:
ls_values.append(float(row[3]))
ldt_timestamps.append(dt.datetime(int(row[0]), int(row[1]), int(row[2]), 16))
ts_fund = pd.TimeSeries(ls_values, index = ldt_timestamps)
#Retrieve Data for Benchmark
dt_timeofday = dt.timedelta(hours=16)
ldt_timestamps = du.getNYSEdays(ldt_timestamps[0], ldt_timestamps[-1], dt_timeofday)
c_dataobj = da.DataAccess('Yahoo', cachestalltime=0)
ls_keys = ['open', 'high', 'low', 'close', 'volume', 'actual_close']
bench_data = c_dataobj.get_data(ldt_timestamps, [benchmark], ls_keys)
d_data = dict(zip(ls_keys, bench_data))
benchmark_price = d_data['close'].values
mul = ts_fund[0] / benchmark_price[0]
benchmark_price = mul * benchmark_price
#Evaluate portfolio performance
#Total risk of the portfolio
daily_ret = tsu.daily(ts_fund.values)
vol = daily_ret.std()
#Average Daily Returns of the Portfolio
avg_daily_ret = daily_ret.mean()
#Calcluate sharpe ratio of the Portfolio (k = 252)
sharpe_ratio = (avg_daily_ret/(vol)) * math.sqrt(252)
#Calculate the Cumulative Return of the portfolio
cum_ret = ts_fund[-1]/ts_fund[0]
#Evaluate portfolio performance
#Total risk of the benchmark
daily_ret_bench = tsu.daily(benchmark_price)
vol_bench = daily_ret_bench.std()
#Average Daily Returns of the Benchmark
avg_daily_ret_bench = daily_ret_bench.mean()
#Calcluate sharpe ratio of the benchmark (k = 252)
sharpe_ratio_bench = (avg_daily_ret_bench/(vol_bench)) * math.sqrt(252)
#Calculate the Cumulative Return of the portfolio
cum_ret_bench = benchmark_price[-1]/benchmark_price[0]
#Plot the performance of the fund and the benchmark
plt.clf()
plt.plot(ldt_timestamps, ts_fund.values, ldt_timestamps, benchmark_price)
plt.legend(['Portfolio', benchmark])
plt.ylabel('Value')
plt.xlabel('Date')
plt.savefig('fund_performance.pdf', format='pdf')
plt.xticks(size='xx-small')
plt.yticks(size='xx-small')
print '\nPortfolio Performance'
print '---------------------------------------'
print ldt_timestamps[0].strftime('Start Date: %B %d, %Y')
print ldt_timestamps[-1].strftime('End Date: %B %d, %Y')
print
print 'Sharpe Ratio of the Fund: ' + str(sharpe_ratio)
print 'Sharpe Ratio of ' + benchmark + ': ' + str(sharpe_ratio_bench) + '\n'
print 'Total Return of the Fund: ' + str(cum_ret)
print 'Total Return of ' + benchmark + ': ' + str(cum_ret_bench[0]) + '\n'
print 'Standard Deviation of the Fund: ' + str(vol)
print 'Standard Deviate of ' + benchmark + ': ' + str(vol_bench) + '\n'
print 'Average Daily Return of the Fund: ' + str(avg_daily_ret)
print 'Average Daily Return of ' + benchmark + ': ' + str(avg_daily_ret_bench) + '\n'
#ts_benchmark_values = _read_in_benchmark(input_benchmark, list(ts_fundvalues.index)) #read in the benchmark closing prices for the same timestamps as ts_fundvalues
ts_benchmark_values = _read_bench(input_benchmark, list(ts_fundvalues.index))
##
##dataobj = da.DataAccess('Yahoo') #data from yahoo
##close = dataobj.get_data(list(ts_fundvalues.index), ['SPY'], ['Close']) #pass a list of symbols to da.get_data
##close = close.fillna(method='ffill')
##close = close.fillna(method='bfill')
##ts_benchmark_values = pd.TimeSeries( dict(zip(list(ts_fundvalues.index), close['SPY'])))
####
# calculate charpe ratios
daily_fund_rets = tsu.daily(ts_fundvalues)
daily_benchmark_rets = tsu.daily(ts_benchmark_values)
sharpe_fund = tsu.get_sharpe_ratio(daily_fund_rets)[0] #note the tsu.get_sharpe_raio returns an ARRAY of ONE value (sharpe ratio)
sharpe_benchmark = tsu.get_sharpe_ratio(daily_benchmark_rets)[0]
std_fund = np.std(daily_fund_rets, axis = 0)[0] #numpy.std returns an ARRAY, so take the 0th elt in the array
std_benchmark = np.std(daily_benchmark_rets, axis = 0)[0]
avg_dailyret_fund = np.mean(daily_fund_rets, axis = 0)[0]
avg_dailyret_benchmark = np.mean(daily_benchmark_rets, axis = 0)[0]
print 'The final value of the portfolio using the sample file is: ' , ts_fundvalues.index[-1], ': $' , ts_fundvalues[-1]
print "Details of the performance of the portfolio: "
print "Data range: ", start_date, " to " , end_date
print "Sharpe Ratio of fund: ", sharpe_fund
print "Sharpe Ratio of Benchmark: ", input_benchmark, ": ", sharpe_benchmark
print "Standard Deviation of Fund : ", std_fund
print "Standard Deviation of $SPX : ", std_benchmark
def do_analysis(values, benchmark_sym):
# first, needs figure out begin and end date.
# then needs to take out benchmark values.
# finnaly plot together.
first_value = values[0]
last_value = values[-1]
dt_start = dt.datetime(first_value[0], first_value[1], first_value[2])
dt_end = dt.datetime(last_value[0], last_value[1], last_value[2])
dt_end += dt.timedelta(days=1)
symbols = [benchmark_sym]
dt_timeofday = dt.timedelta(hours=16)
ldt_timestamps = du.getNYSEdays(dt_start, dt_end, dt_timeofday);
dataobj = da.DataAccess('Yahoo')
ls_keys = ['close']#['open', 'high', 'low', 'close', 'volume', 'actual_close']
ldf_data = dataobj.get_data(ldt_timestamps, symbols, ls_keys, verbose=True)
d_data = dict(zip(ls_keys, ldf_data))
# print d_data
for s_key in ls_keys:
d_data[s_key] = d_data[s_key].fillna(method='ffill')
d_data[s_key] = d_data[s_key].fillna(method='bfill')
# d_data[s_key] = d_data[s_key].fillna(1.0)
benchmark = d_data['close'][benchmark_sym];
# print normalized_benchmark[0];
factory = values[0][3] / benchmark[0];
# normalized to same factory to fund value.
normalized_benchmark = benchmark * factory
normalized_values = map(lambda x: x[3], values)
plt.clf()
plt.plot(ldt_timestamps, normalized_benchmark, "r",
ldt_timestamps, normalized_values)
plt.legend([ benchmark_sym, 'fund value'])
plt.ylabel('Fund Value')
plt.xlabel('Date')
plt.savefig('analysis.pdf', format='pdf')
daily_ret_my = daily_ret(normalized_values)
daily_ret_bm = tsu.daily(d_data['close'][benchmark_sym])
sharp_ratio_my = tsu.get_sharpe_ratio(daily_ret_my)
sharp_ratio_bm = tsu.get_sharpe_ratio(daily_ret_bm)[0]
# why -1 and +1 ?
total_ret_my = (float(normalized_values[-1]) / float(normalized_values[0]))
total_ret_bm = float(benchmark[-1] / float(benchmark[0]))
stddev_my = np.std(daily_ret_my)
stddev_bm = np.std(daily_ret_bm)
avg_my = np.average(daily_ret_my)
avg_bm = np.average(daily_ret_bm)
print "Details of the Performance of the portfolio :"
print ""
print "Data Range : %s to %s" % (ldt_timestamps[0], ldt_timestamps[-1])
print ""
print "Sharpe Ratio of Fund : %f" % sharp_ratio_my
print "Sharpe Ratio of $SPX : %fy" % sharp_ratio_bm
print ""
print "Total Return of Fund : %f" % total_ret_my
print "Total Return of $SPX : %f" % total_ret_bm
print ""
print "Standard Deviation of Fund : %f" % stddev_my
print "Standard Deviation of $SPX : %f" % stddev_bm
print ""
print "Average Daily Return of Fund : %f" % avg_my
print "Average Daily Return of %s : %f" % (benchmark_sym, avg_bm)
