def time_guarantee(my_contract, year_ini, year_end, my_df):
if my_contract == 'ES':
my_guarantee = 12000 * 2 # between 4000$ and 30000$
my_multiplier = 50
my_tic = 0.25
if my_contract == 'YM':
my_guarantee = 10000 * 2 # between 3000$ and 25000$
my_multiplier = 5
my_tic = 1
if my_contract == 'NQ':
my_guarantee = 15000 * 2 # between 5000$ and 45000$
my_multiplier = 20
my_tic = 0.25
if my_contract == 'CL':
my_guarantee = 5000 * 2 # between 2000$ and 18000$
my_multiplier = 1000
my_tic = 0.01
if my_contract == 'RTY':
my_guarantee = 7000 * 2 # between 2000$ and 20000$
my_multiplier = 5
my_tic = 1
int_date_0 = year_ini * 10000 + 1 * 100 + 1 # the day d0 = date(2008, 8, 18)
int_date_1 = year_end * 10000 + 12 * 100 + 31 # the day d0 = date(2008, 8, 18)
int_date_01 = mda.locate_1st_day('Y', year_ini, my_df, 'date')
int_date_11 = mda.locate_lst_day('Y', year_end, my_df, 'date')
d1 = my_df['date'][int_date_11]
d0 = my_df['date'][int_date_01]
delta = d1 - d0
delta_sec = delta.total_seconds()
delta_years = delta_sec / (365 * 24 * 60 * 60)
return my_multiplier, my_guarantee, delta_years, my_tic
def short_long_operationLSEX(my_contract, year_ini, year_end, my_df, limit_qty,
target_qty, my_multiplier):
pd.options.mode.chained_assignment = None # default='warn' This is to avoid warnings of chained-assignment
# creation of new columns in datframe
my_df['operation'] = ' '
my_df['target'] = np.zeros(len(my_df)).astype('float')
my_df['limit'] = np.zeros(len(my_df)).astype('float')
my_df['result-pts'] = np.zeros(len(my_df)).astype('float')
my_df['result-accum-pts'] = np.zeros(len(my_df)).astype('float')
my_df['contract-qty'] = np.zeros(len(my_df)).astype('float')
my_df['multiplier'] = np.zeros(len(my_df)).astype('float')
my_df['result-amt'] = np.zeros(len(my_df)).astype('float')
my_df['result-accum-amt'] = np.zeros(len(my_df)).astype('float')
# define variables to use
long = 'L' #
short = 'S' #
operation = long #
operation_ant = long #
limit = 0.0 #
limit_ant = 0.0 #
target = 0.0 #
target_ant = 0.0 #
limit_qty_rng = limit_qty ##########
target_qty_rng = target_qty ############
range_trail = 0.0 #
axis_size = 0.0 #
tic = 0.0 #
contract_qty = 1.0 #
# define variable for operation
long_accum = 0.0 #
short_accum = 0.0 #
result_accum_amt = 0.0 #
result_accum_pts = 0.0 #
limit_nxt = 0 - limit_qty_rng * axis_size #
target_nxt = 0 + target_qty_rng * axis_size #
operation_nxt = long #
target_reached = 0 #
# locate the initial and final position
pt_ini = mda.locate_1st_day('Y', year_ini, my_df,
'date') # locate the initial index
pt_end = mda.locate_lst_day('Y', year_end, my_df,
'date') # locate the final index
# the loop
for ind in range(pt_ini, pt_end + 1):
range_trail = my_df['range-trail'][ind]
long_result = my_df['long-rst'][ind]
short_result = my_df['short-rst'][ind]
axis_size = range_trail
if ind == pt_ini:
limit_nxt = 0 - limit_qty_rng * axis_size
target_nxt = 0 + target_qty_rng * axis_size
operation = operation_nxt
limit = limit_nxt
target = target_nxt
long_accum = long_accum + long_result
short_accum = short_accum + short_result
continue_no_changes = 1
if operation == long:
result_accum_pts = result_accum_pts + long_result - tic
result_accum_amt = result_accum_amt + (
long_result - tic) * contract_qty * my_multiplier
if (long_accum <
limit) & (continue_no_changes
== 1): # overcome the limit and change to shorts
operation_ant = long
operation_nxt = short
limit_ant = limit
limit_nxt = short_accum - (limit_qty_rng * axis_size)
target_ant = target
target_nxt = short_accum + (target_qty_rng * axis_size)
continue_no_changes = 0
if ((target_reached == 1) & (long_result <=
((0 - 1) * range_trail * 0.2)) &
(continue_no_changes
== 1)): # target reached with reverse of 20%
operation_ant = long
operation_nxt = short
limit_ant = limit
limit_nxt = short_accum - (limit_qty_rng * axis_size)
target_ant = target
target_nxt = short_accum + (target_qty_rng * axis_size)
target_reached = 0
continue_no_changes = 0
if ((long_accum > target) & (target_reached == 0) &
(continue_no_changes
== 1)): # target reached but waiting reverse
target_reached = 1
operation_ant = operation
operation_nxt = operation_ant
limit_ant = limit
limit_nxt = limit_ant
target_ant = target
target_nxt = target_ant
continue_no_changes = 0
if operation == short:
result_accum_pts = result_accum_pts + short_result - tic
result_accum_amt = result_accum_amt + short_result * contract_qty * my_multiplier
if (short_accum < limit) & (continue_no_changes
== 1): # overcome limit and change
operation_ant = short
operation_nxt = long
limit_ant = limit
limit_nxt = long_accum - (limit_qty_rng * axis_size)
target_ant = target
target_nxt = long_accum + (target_qty_rng * axis_size)
continue_no_changes = 0
if ((target_reached == 1) & (short_result <=
((0 - 1) * range_trail * 0.2)) &
(continue_no_changes
== 1)): # target reached with reverse of 20% and change
operation_ant = short
operation_nxt = long
limit_ant = limit
limit_nxt = long_accum - (limit_qty_rng * axis_size)
target_ant = target
target_nxt = long_accum + (target_qty_rng * axis_size)
target_reached = 0
continue_no_changes = 0
if ((short_accum > target) & (target_reached == 0) &
(continue_no_changes
== 1)): # target reached but waiting reverse
target_reached = 1
operation_ant = operation
operation_nxt = operation_ant
limit_ant = limit
limit_nxt = limit
target_ant = target
target_nxt = target
continue_no_changes = 0
if continue_no_changes == 1: # nothing happen, no target reached, no limit overcame, continue the same
operation_ant = operation
operation_nxt = operation_ant
limit_ant = limit
limit_nxt = limit_ant
target_ant = target
target_nxt = target_ant
# save the daily operation
my_df['operation'][ind] = operation_ant
my_df['target'][ind] = target_ant
my_df['limit'][ind] = limit_ant
my_df['result-pts'][
ind] = result_accum_pts - my_df['result-accum-pts'][ind - 1]
my_df['result-accum-pts'][ind] = result_accum_pts
my_df['contract-qty'][ind] = contract_qty
my_df['multiplier'][ind] = my_multiplier
my_df['result-amt'][
ind] = result_accum_amt - my_df['result-accum-amt'][ind - 1]
my_df['result-accum-amt'][ind] = result_accum_amt
return my_df
def short_long_operationFLIP(my_contract, year_ini, year_end, my_df,
my_multiplier):
pd.options.mode.chained_assignment = None # default='warn' This is to avoid warnings of chained-assignment
# creation of new columns in datframe
my_df['operation'] = ' '
my_df['target'] = np.zeros(len(my_df)).astype('float')
my_df['limit'] = np.zeros(len(my_df)).astype('float')
my_df['result-pts'] = np.zeros(len(my_df)).astype('float')
my_df['result-accum-pts'] = np.zeros(len(my_df)).astype('float')
my_df['contract-qty'] = np.zeros(len(my_df)).astype('float')
my_df['multiplier'] = np.zeros(len(my_df)).astype('float')
my_df['result-amt'] = np.zeros(len(my_df)).astype('float')
my_df['result-accum-amt'] = np.zeros(len(my_df)).astype('float')
# define variables to use
long = 'L' #
short = 'S' #
operation_ant = long #
tic = 0.0 #
contract_qty = 1.0 #
# define variable for operation
long_accum = 0.0 #
short_accum = 0.0 #
result_accum_amt = 0.0 #
result_accum_pts = 0.0 #
operation_nxt = long #
# locate the initial and final position
pt_ini = mda.locate_1st_day('Y', year_ini, my_df,
'date') # locate the initial index
pt_end = mda.locate_lst_day('Y', year_end, my_df,
'date') # locate the final index
# the loop
for ind in range(pt_ini, pt_end + 1):
range_trail = my_df['range-trail'][ind]
long_result = my_df['long-rst'][ind]
short_result = my_df['short-rst'][ind]
axis_size = range_trail
if ind == pt_ini:
operation_nxt = long
operation = operation_nxt
long_accum = long_accum + long_result
short_accum = short_accum + short_result
continue_no_changes = 1
if operation == long:
result_accum_pts = result_accum_pts + long_result - tic
result_accum_amt = result_accum_amt + (
long_result - tic) * contract_qty * my_multiplier
if (long_result <
0) & (continue_no_changes
== 1): # bad long performance and change to shorts
operation_ant = long
operation_nxt = short
continue_no_changes = 0
if operation == short:
result_accum_pts = result_accum_pts + short_result - tic
result_accum_amt = result_accum_amt + short_result * contract_qty * my_multiplier
if (short_result <
0) & (continue_no_changes
== 1): # bad short performance and change to long
operation_ant = short
operation_nxt = long
continue_no_changes = 0
if continue_no_changes == 1: # nothing happen, continue the same
operation_ant = operation
operation_nxt = operation_ant
# save the daily operation
my_df['operation'][ind] = operation_ant
my_df['target'][ind] = 0
my_df['limit'][ind] = 0
my_df['result-pts'][
ind] = result_accum_pts - my_df['result-accum-pts'][ind - 1]
my_df['result-accum-pts'][ind] = result_accum_pts
my_df['contract-qty'][ind] = contract_qty
my_df['multiplier'][ind] = my_multiplier
my_df['result-amt'][
ind] = result_accum_amt - my_df['result-accum-amt'][ind - 1]
my_df['result-accum-amt'][ind] = result_accum_amt
return my_df
def market_operation(df_int, df_day, timetable_ini, timetable_end):
"""
This function complete a dataframe df_day with open, close-long, close-short for a daily base in
a specific timetable
You can realise that we are going to operate long and short.
On the other hand we will use the order STOP-TRAIL
-Parameters
----------
df_int : dataframe with original data (intraday movements of a future contract)
['date', 'time', 'open', 'max', 'min', 'close', 'volume']
df_day : new dataframe to create with columns ['date', 'timetable', 'max', 'min', 'range', range-avg, range-trail].
We will add the following columns
['open', 'close-long', 'close-short', 'close', long-rst, short-rst, long-acc, short-acc]
timetable_ini : time the operation start everyday. Example='09:00'
timetable_end : time the operation finish everyday. Example='18:00'
-Returns
-------
nothing
The function fill the dataframe df_day with daily
['open', 'long-close', 'short-close', 'close', 'long-rst', 'short-rst', 'long-acc', 'short-acc']
"""
pd.options.mode.chained_assignment = None # default='warn' This is to avoid warnings of chained-assignment
# Create new columns in the dataframe and fill them with zeros
df_day['open'] = np.zeros(len(df_day)).astype('float')
df_day['long-close'] = np.zeros(len(df_day)).astype('float')
df_day['short-close'] = np.zeros(len(df_day)).astype('float')
df_day['close'] = np.zeros(len(df_day)).astype('float')
df_day['long-rst'] = np.zeros(len(df_day)).astype('float')
df_day['short-rst'] = np.zeros(len(df_day)).astype('float')
df_day['long-acc'] = np.zeros(len(df_day)).astype('float')
df_day['short-acc'] = np.zeros(len(df_day)).astype('float')
pt_ini_date = df_day['date'][df_day[df_day['range-trail'] != 0].index[0]].year * 10000 + \
df_day['date'][df_day[df_day['range-trail'] != 0].index[0]].month * 100 + \
df_day['date'][df_day[df_day['range-trail'] != 0].index[0]].day
pt_end_date = df_day['date'][df_day[df_day['range-trail'] != 0].index[-1]].year * 10000 + \
df_day['date'][df_day[df_day['range-trail'] != 0].index[-1]].month * 100 + \
df_day['date'][df_day[df_day['range-trail'] != 0].index[-1]].day
pt_ini = mda.locate_1st_day('D', pt_ini_date, df_int, 'datetime_EU')
pt_end = mda.locate_1st_day('D', pt_end_date, df_int, 'datetime_EU')
# testing dataframe is create only for testing purpose. We will delete this part after good perform confirmation
testing = pd.DataFrame(columns=[
'datetime_EU', 'open', 'max', 'min', 'close', 'long-stop',
'short-stop', 'long-exit', 'short-exit', 'long-operation',
'short-operation', 'max-day', 'min-day', 'range-trail'
])
testing['open'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['max'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['min'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['close'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['long-stop'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['short-stop'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['long-exit'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['short-exit'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['long-operation'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['short-operation'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['max-day'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['min-day'] = np.zeros(pt_end - pt_ini + 1).astype('float')
testing['range-trail'] = np.zeros(pt_end - pt_ini + 1).astype('float')
# Define variables to use
operation = 0 # (1 in market, 0 out market)
entry = 0.0
long_stop = 0.0
short_stop = 0.0
long_exit = 0.0
short_exit = 0.0
long_operation = 0
short_operation = 0
maximum = 0.0
minimum = 0.0
market_exit = 0.0
hour_ini = int(timetable_ini[0:2]) # initial values
minu_ini = int(timetable_ini[3:])
hour_end = int(timetable_end[0:2])
minu_end = int(timetable_end[3:])
tmtb_ini = hour_ini * 100 + minu_ini # integer value
tmtb_end = hour_end * 100 + minu_end # integer value
my_int_date_ant = 0 # define my date ant
my_long_acc = 0.0 # long result accumulator
my_short_acc = 0.0 # short result accumulator
j = 0 # counter for testing
for ind in range(pt_ini, pt_end + 1): # the main loop
my_int_date = df_int['datetime_EU'][ind].year * 10000 + df_int['datetime_EU'][ind].month * 100 + \
df_int['datetime_EU'][ind].day # the day
if my_int_date > my_int_date_ant: # This is a new day
my_int_date_ant = my_int_date
my_loc = mda.locate_1st_day('D', my_int_date, df_day,
'date') # locate the day in the df_day
range_trail = df_day['range-trail'][my_loc] # take the range-trail
my_int_time = df_int['datetime_EU'][ind].hour * 100 + df_int[
'datetime_EU'][ind].minute # the time
# checking the bar time is >= than start time of timetable and < than end time of timetable ==>> operation in market
if (my_int_time >= tmtb_ini) & (my_int_time < tmtb_end):
if (my_int_time == tmtb_ini) & (
operation == 0): # The moment to start operation
entry = df_int['open'][ind]
long_stop = entry - range_trail
short_stop = entry + range_trail
long_exit = df_int['close'][ind]
short_exit = df_int['close'][ind]
long_operation = 1
short_operation = 1
maximum = df_int['max'][ind]
minimum = df_int['min'][ind]
operation = 1
if operation == 1:
if short_operation == 1:
short_exit = df_int['close'][ind]
if long_operation == 1:
long_exit = df_int['close'][ind]
if (df_int['max'][ind] > short_stop) & (
short_operation
== 1): # if the price overcome short_stop (up)
short_exit = short_stop
short_operation = 0
if (df_int['min'][ind] < long_stop) & (
long_operation
== 1): # if the price overcome long_stop (down)
long_exit = long_stop
long_operation = 0
if (long_stop < df_int['max'][ind] - range_trail) & (
long_operation == 1): # up long stop
long_stop = df_int['max'][ind] - range_trail
if (short_stop > df_int['min'][ind] + range_trail) & (
short_operation == 1): # down short stop
short_stop = df_int['min'][ind] + range_trail
if (my_int_time == tmtb_end) & (
long_operation
== 1): # long operation end because finish timetable
long_operation = 0
if (my_int_time == tmtb_end) & (
short_operation
== 1): # short operation end because finish timetable
short_operation = 0
if df_int['max'][
ind] > maximum: # update maximum if it is the case
maximum = df_int['max'][ind]
if df_int['min'][
ind] < minimum: # update minimum if it is the case
minimum = df_int['min'][ind]
market_exit = df_int['close'][ind]
elif (my_int_time == tmtb_end) & (
operation == 1): # timetable finish and save the result
df_day['open'][my_loc] = entry
df_day['long-close'][my_loc] = long_exit
df_day['short-close'][my_loc] = short_exit
df_day['close'][my_loc] = market_exit
df_day['long-rst'][my_loc] = long_exit - entry
df_day['short-rst'][my_loc] = entry - short_exit
my_long_acc = my_long_acc + long_exit - entry
df_day['long-acc'][my_loc] = my_long_acc
my_short_acc = my_short_acc + entry - short_exit
df_day['short-acc'][my_loc] = my_short_acc
operation = 0 # Re-init variables
entry = 0.0
long_stop = 0.0
short_stop = 0.0
long_exit = 0.0
short_exit = 0.0
long_operation = 0
short_operation = 0
maximum = 0.0
minimum = 0.0
market_exit = 0.0
# Save tracking information to check the function works only for debuging and testing
j += 1
testing['datetime_EU'][j] = df_int['datetime_EU'][ind]
testing['open'][j] = df_int['open'][ind]
testing['max'][j] = df_int['max'][ind]
testing['min'][j] = df_int['min'][ind]
testing['close'][j] = df_int['close'][ind]
testing['long-stop'][j] = long_stop
testing['short-stop'][j] = short_stop
testing['long-exit'][j] = long_exit
testing['short-exit'][j] = short_exit
testing['long-operation'][j] = long_operation
testing['short-operation'][j] = short_operation
testing['max-day'][j] = maximum
testing['min-day'][j] = minimum
testing['range-trail'][j] = range_trail
return df_day, testing