def research_db(sectors_list, sectors_filter_out, countries_list,
countries_filter_out, evr_range, pe_range, pm_range,
ev_millions_range, research_mode_max_ev, csv_db_path,
db_filename, read_united_states_input_symbols, scan_mode,
generate_result_folders, appearance_counter_min,
appearance_counter_max, favor_sectors, favor_sectors_by,
newer_path, older_path, db_exists_in_both_folders,
diff_only_result, movement_threshold, res_length):
if scan_mode == SCAN_MODE_TASE:
tase_mode = 1
else:
tase_mode = 0
if research_mode_max_ev:
ev_millions_range = list(
reversed(ev_millions_range)
) # Flip order to have stocks with higher EV first (as limit shall be Max and not Min)
appearance_counter_dict_sss = {}
prepare_appearance_counters_dictionaries(csv_db_path, db_filename,
appearance_counter_dict_sss)
ev_millions_range_len = len(ev_millions_range)
pe_range_len = len(pe_range)
evr_range_len = len(evr_range)
pm_range_len = len(pm_range)
research_rows_sss = np.zeros(
(ev_millions_range_len, pe_range_len, evr_range_len, pm_range_len),
dtype=int)
for ev_millions_index, ev_millions_limit in enumerate(ev_millions_range):
for pe_index, price_to_earnings_limit in enumerate(pe_range):
for evr_index, enterprise_value_to_revenue_limit in enumerate(
evr_range):
for pm_index, profit_margin_limit in enumerate(
pm_range
): # TODO: ASAFR: Below 1. Ambiguity of parameters - narrow down. 2. Some magic numbers on ev_to_cfo_ration etc 100.0 and 1000.0 - make order and defines/constants/multi_dim here
num_results_for_ev_pe_evr_and_pm = sss.sss_run(
reference_run=[],
sectors_list=sectors_list,
sectors_filter_out=sectors_filter_out,
countries_list=countries_list,
countries_filter_out=countries_filter_out,
build_csv_db_only=0,
build_csv_db=0,
csv_db_path=csv_db_path,
db_filename=db_filename,
read_united_states_input_symbols=
read_united_states_input_symbols,
tase_mode=tase_mode,
num_threads=1,
market_cap_included=1,
research_mode=1,
profit_margin_limit=float(profit_margin_limit) / 100.0,
enterprise_value_millions_usd_limit=ev_millions_limit,
research_mode_max_ev=research_mode_max_ev,
ev_to_cfo_ratio_limit=10e9,
debt_to_equity_limit=10e9,
price_to_earnings_limit=price_to_earnings_limit,
enterprise_value_to_revenue_limit=
enterprise_value_to_revenue_limit,
favor_sectors=favor_sectors,
favor_sectors_by=favor_sectors_by,
generate_result_folders=generate_result_folders,
appearance_counter_dict_sss=appearance_counter_dict_sss,
appearance_counter_min=appearance_counter_min,
appearance_counter_max=appearance_counter_max)
if num_results_for_ev_pe_evr_and_pm < appearance_counter_min:
break # already lower than appearance_counter_min results. With higher profit margin limit there will always be less results -> save running time by breaking
research_rows_sss[ev_millions_index][pe_index][evr_index][
pm_index] = int(num_results_for_ev_pe_evr_and_pm)
print(
'ev_millions_limit {:6} | price_to_earnings_limit {:8} | row {:3} -> (enterprise_value_to_revenue_limit {:8}) | col {:3} -> (profit_margin_limit {:7}%): num_results_for_ev_pe_evr_and_pm = {}'
.format(ev_millions_limit, price_to_earnings_limit,
evr_index, enterprise_value_to_revenue_limit,
pm_index, profit_margin_limit,
num_results_for_ev_pe_evr_and_pm))
results_filename = 'results_without_labels.csv'
mesh_combined = combine_multi_dim_to_table_4d(multi_dim=research_rows_sss,
dim4=ev_millions_range,
dim3=pe_range,
rows=evr_range,
cols=pm_range)
np.savetxt(csv_db_path + '/' + results_filename,
mesh_combined,
fmt='%f',
delimiter=',')
title_row = pm_range # column 3 and onwards
title_row.insert(0, 'evr / pm') # column 2
title_row.insert(0, 'pe') # column 1
title_row.insert(0, 'ev') # column 0
ev_pe_evr_rows_pm_cols_filenames_list = [
csv_db_path + '/' + results_filename
]
# Read Results, and add row and col axis:
for filename in ev_pe_evr_rows_pm_cols_filenames_list:
ev_pe_evr_rows_pm_cols = [title_row]
with open(filename, mode='r', newline='') as engine:
reader = csv.reader(engine, delimiter=',')
row_index = 0 # title + len(ev_millions_range)*len(evr_range)
for row in reader:
ev_pe_evr_rows_pm_cols.append(row)
row.insert(2, evr_range[
int(row_index) %
len(evr_range)]) # column 2 is evr repetitively cyclic
row_index += 1
for index in range(len(ev_pe_evr_rows_pm_cols_filenames_list)):
row_col_csv_filename = ev_pe_evr_rows_pm_cols_filenames_list[
index].replace('.csv', '_with_labels.csv')
os.makedirs(os.path.dirname(row_col_csv_filename), exist_ok=True)
with open(row_col_csv_filename, mode='w', newline='') as engine:
writer = csv.writer(engine)
writer.writerows(ev_pe_evr_rows_pm_cols)
sorted_appearance_counter_dict_sss = {
k: v
for k, v in sorted(appearance_counter_dict_sss.items(),
key=lambda item: item[1],
reverse=True)
}
result_sorted_appearance_counter_dict_sss = {
k: v
for k, v in sorted_appearance_counter_dict_sss.items() if v > 0.0
}
result_list_filename_sss = csv_db_path + '/results_sss.csv'
with open(result_list_filename_sss, 'w') as f:
f.write("Symbol,Name,Sector,sss_value,close,appearance_counter\n")
for key in result_sorted_appearance_counter_dict_sss.keys():
# Symbol, Name, Sector S Close appearance_counter
f.write(
"%s,%s,%s,%s,%s,%s\n" %
(key[0], str(key[1]).replace(',', ' '), key[2], key[3], key[4],
round(result_sorted_appearance_counter_dict_sss[key], 4)))
if older_path is not None:
diff_lists = sss_diff.run(
newer_path=newer_path,
older_path=older_path,
db_exists_in_both_folders=db_exists_in_both_folders,
diff_only_result=diff_only_result,
movement_threshold=movement_threshold,
res_length=res_length,
consider_as_new_from=PDF_NUM_ENTRIES_IN_REPORT)
# 0:15 is date and time
pdf_generator.csv_to_pdf(
csv_filename=result_list_filename_sss,
csv_db_path=csv_db_path,
data_time_str=result_list_filename_sss.replace(
'Results', '').replace('Tase', '').replace('Nsr', '').replace(
'All', '').replace('Custom', '').replace('/', '')[0:15],
title=TITLES[scan_mode].replace('_', ' '),
limit_num_rows=PDF_NUM_ENTRIES_IN_REPORT,
diff_list=diff_lists[0],
tase_mode=tase_mode)
if not research_mode: # Run Build DB Only:
if run_custom_tase:
sss.sss_run(reference_run=reference_run_tase,
sectors_list=[],
sectors_filter_out=0,
countries_list=[],
countries_filter_out=0,
build_csv_db_only=1,
build_csv_db=1,
csv_db_path='None',
db_filename='None',
read_united_states_input_symbols=0,
tase_mode=1,
num_threads=1,
market_cap_included=1,
research_mode=0,
profit_margin_limit=0.0001,
ev_to_cfo_ratio_limit=10e9,
debt_to_equity_limit=10e9,
enterprise_value_millions_usd_limit=5,
research_mode_max_ev=False,
price_to_earnings_limit=10e9,
enterprise_value_to_revenue_limit=10e9,
favor_sectors=[],
favor_sectors_by=[],
generate_result_folders=1,
custom_portfolio=sss_config.custom_portfolio_tase)
if run_custom:
sss.sss_run(reference_run=reference_run_all,
sectors_list=[],
# ==============================================
# sss.sss_run(sectors_list=[], sectors_filter_out=0, countries_list=['United States'], countries_filter_out=0, build_csv_db_only=1, build_csv_db=1, csv_db_path='None', read_united_states_input_symbols=0, tase_mode=0, num_threads=20, market_cap_included=1, use_investpy=0, research_mode=0, profit_margin_limit=0.01, ev_to_cfo_ratio_limit=200.0, min_enterprise_value_millions_usd=100, best_n_select=2, enterprise_value_to_revenue_limit=200, favor_sectors=['Technology', 'Financial Services'], favor_sectors_by=[4.5, 0.33333], generate_result_folders=1)
# Run Build DB Only: All/Others
# =============================
sss.sss_run(sectors_list=[],
sectors_filter_out=0,
countries_list=['United States'],
countries_filter_out=0,
build_csv_db_only=1,
build_csv_db=1,
csv_db_path='None',
read_united_states_input_symbols=1,
tase_mode=0,
num_threads=20,
market_cap_included=1,
use_investpy=0,
research_mode=0,
profit_margin_limit=0.01,
ev_to_cfo_ratio_limit=20000.0,
min_enterprise_value_millions_usd=5,
best_n_select=3,
enterprise_value_to_revenue_limit=1000,
favor_sectors=['Technology', 'Financial Services'],
favor_sectors_by=[4.5, 1],
generate_result_folders=1)
# Research Mode:
# ==============
def research_db(sectors_list, sectors_filter_out, countries_list,
countries_filter_out, evr_range, pm_range, ev_millions_range,
csv_db_path, read_united_states_input_symbols, scan_mode,
generate_result_folders, appearance_counter_min,
appearance_counter_max, favor_sectors, favor_sectors_by,
newer_path, older_path, db_exists_in_both_folders,
diff_only_recommendation, movement_threshold, newer_rec_ranges,
older_rec_ranges, rec_length):
if scan_mode == SCAN_MODE_TASE:
tase_mode = 1
else:
tase_mode = 0
appearance_counter_dict_sss = {}
appearance_counter_dict_ssss = {}
appearance_counter_dict_sssss = {}
prepare_appearance_counters_dictionaries(csv_db_path,
appearance_counter_dict_sss,
appearance_counter_dict_ssss,
appearance_counter_dict_sssss)
ev_millions_range_len = len(ev_millions_range)
evr_range_len = len(evr_range)
pm_range_len = len(pm_range)
research_rows_sss = np.zeros(
(ev_millions_range_len, evr_range_len, pm_range_len), dtype=int)
research_rows_ssss = np.zeros(
(ev_millions_range_len, evr_range_len, pm_range_len), dtype=int)
research_rows_sssss = np.zeros(
(ev_millions_range_len, evr_range_len, pm_range_len), dtype=int)
for ev_millions_index, ev_millions_limit in enumerate(ev_millions_range):
for evr_index, enterprise_value_to_revenue_limit in enumerate(
evr_range):
for pm_index, profit_margin_limit in enumerate(pm_range):
num_results_for_ev_evr_and_pm = sss.sss_run(
sectors_list=sectors_list,
sectors_filter_out=sectors_filter_out,
countries_list=countries_list,
countries_filter_out=countries_filter_out,
build_csv_db_only=0,
build_csv_db=0,
csv_db_path=csv_db_path,
read_united_states_input_symbols=
read_united_states_input_symbols,
tase_mode=tase_mode,
num_threads=1,
market_cap_included=1,
use_investpy=0,
research_mode=1,
profit_margin_limit=float(profit_margin_limit) / 100.0,
min_enterprise_value_millions_usd=ev_millions_limit,
ev_to_cfo_ratio_limit=100.0,
best_n_select=3,
enterprise_value_to_revenue_limit=
enterprise_value_to_revenue_limit,
favor_sectors=favor_sectors,
favor_sectors_by=favor_sectors_by,
generate_result_folders=generate_result_folders,
appearance_counter_dict_sss=appearance_counter_dict_sss,
appearance_counter_dict_ssss=appearance_counter_dict_ssss,
appearance_counter_dict_sssss=appearance_counter_dict_sssss,
appearance_counter_min=appearance_counter_min,
appearance_counter_max=appearance_counter_max)
if num_results_for_ev_evr_and_pm < appearance_counter_min:
break # already lower than appearance_counter_min results. With higher profit margin limit there will always be less results -> save running time by breaking
research_rows_sss[ev_millions_index][evr_index][
pm_index] = int(num_results_for_ev_evr_and_pm)
research_rows_ssss[ev_millions_index][evr_index][
pm_index] = int(num_results_for_ev_evr_and_pm)
research_rows_sssss[ev_millions_index][evr_index][
pm_index] = int(num_results_for_ev_evr_and_pm)
print(
'ev_millions_limit {:6} | row {:3} -> (enterprise_value_to_revenue_limit {:8}) | col {:3} -> (profit_margin_limit {:7}%): num_results_for_ev_evr_and_pm = {}'
.format(ev_millions_limit, evr_index,
enterprise_value_to_revenue_limit, pm_index,
profit_margin_limit,
num_results_for_ev_evr_and_pm))
results_filename = 'results_evm{}-{}_evr{}-{}_pm{}-{}.csv'.format(
ev_millions_range[0], ev_millions_range[-1], evr_range[0],
evr_range[-1], pm_range[0], pm_range[-1])
mesh_combined = combine_multi_dim_to_table(multi_dim=research_rows_sss,
dim3=ev_millions_range,
rows=evr_range,
cols=pm_range)
np.savetxt(csv_db_path + '/' + results_filename,
mesh_combined.astype(int),
fmt='%d',
delimiter=',')
title_row = pm_range # column 2 and onwards
title_row.insert(0, 'evr / pm') # column 1
title_row.insert(0, 'ev') # column 0
ev_evr_rows_pm_cols_filenames_list = [csv_db_path + '/' + results_filename]
# Read Results, and add row and col axis:
for filename in ev_evr_rows_pm_cols_filenames_list:
ev_evr_rows_pm_cols = [title_row]
with open(filename, mode='r', newline='') as engine:
reader = csv.reader(engine, delimiter=',')
row_index = 0 # title + len(ev_millions_range)*len(evr_range)
for row in reader:
row.insert(0, evr_range[
row_index %
len(evr_range)]) # column 1 is evr repetitively cyclic
row.insert(
0, evr_range[int(row_index / len(evr_range))]
) # column 0 is the ev, increasing every len(evr_range)
ev_evr_rows_pm_cols.append(row)
row_index += 1
for index in range(len(ev_evr_rows_pm_cols_filenames_list)):
row_col_csv_filename = ev_evr_rows_pm_cols_filenames_list[
index].replace('.csv', '_evr_row_pm_col.csv')
os.makedirs(os.path.dirname(row_col_csv_filename), exist_ok=True)
with open(row_col_csv_filename, mode='w', newline='') as engine:
writer = csv.writer(engine)
writer.writerows(ev_evr_rows_pm_cols)
sorted_appearance_counter_dict_sss = {
k: v
for k, v in sorted(appearance_counter_dict_sss.items(),
key=lambda item: item[1],
reverse=True)
}
result_sorted_appearance_counter_dict_sss = {
k: v
for k, v in sorted_appearance_counter_dict_sss.items() if v > 0.0
}
sorted_appearance_counter_dict_ssss = {
k: v
for k, v in sorted(appearance_counter_dict_ssss.items(),
key=lambda item: item[1],
reverse=True)
}
result_sorted_appearance_counter_dict_ssss = {
k: v
for k, v in sorted_appearance_counter_dict_ssss.items() if v > 0.0
}
sorted_appearance_counter_dict_sssss = {
k: v
for k, v in sorted(appearance_counter_dict_sssss.items(),
key=lambda item: item[1],
reverse=True)
}
result_sorted_appearance_counter_dict_sssss = {
k: v
for k, v in sorted_appearance_counter_dict_sssss.items() if v > 0.0
}
recommendation_list_filename_sss = csv_db_path + '/rec_sss_' + results_filename.replace(
'results_', '')
recommendation_list_filename_ssss = csv_db_path + '/rec_ssss_' + results_filename.replace(
'results_', '')
recommendation_list_filename_sssss = csv_db_path + '/rec_sssss_' + results_filename.replace(
'results_', '')
with open(recommendation_list_filename_sss, 'w') as f:
f.write("Ticker,Name,Sector,sss_value,close,appearance_counter\n")
for key in result_sorted_appearance_counter_dict_sss.keys():
# Ticker, Name, Sector S Close appearance_counter
f.write(
"%s,%s,%s,%s,%s,%s\n" %
(key[0], str(key[1]).replace(',', ' '), key[2], key[3], key[4],
round(result_sorted_appearance_counter_dict_sss[key], 4)))
with open(recommendation_list_filename_ssss, 'w') as f:
f.write("Ticker,Name,Sector,ssss_value,close,appearance_counter\n")
for key in result_sorted_appearance_counter_dict_ssss.keys():
f.write(
"%s,%s,%s,%s,%s,%s\n" %
(key[0], str(key[1]).replace(',', ' '), key[2], key[3], key[4],
round(result_sorted_appearance_counter_dict_ssss[key], 4)))
with open(recommendation_list_filename_sssss, 'w') as f:
f.write("Ticker,Name,Sector,sssss_value,close,appearance_counter\n")
for key in result_sorted_appearance_counter_dict_sssss.keys():
f.write(
"%s,%s,%s,%s,%s,%s\n" %
(key[0], str(key[1]).replace(',', ' '), key[2], key[3], key[4],
round(result_sorted_appearance_counter_dict_sssss[key], 4)))
if older_path is not None:
diff_lists = sss_diff.run(
newer_path=newer_path,
older_path=older_path,
db_exists_in_both_folders=db_exists_in_both_folders,
diff_only_recommendation=diff_only_recommendation,
movement_threshold=movement_threshold,
newer_rec_ranges=newer_rec_ranges,
older_rec_ranges=older_rec_ranges,
rec_length=rec_length,
consider_as_new_from=PDF_NUM_ENTRIES_IN_REPORT)
# 0:15 is date and time
pdf_generator.csv_to_pdf(
csv_filename=recommendation_list_filename_sss,
csv_db_path=csv_db_path,
data_time_str=recommendation_list_filename_sss.replace(
'Results/', '')[0:15],
title=TITLES[scan_mode].replace('_', ' '),
limit_num_rows=PDF_NUM_ENTRIES_IN_REPORT,
diff_list=diff_lists[0],
tase_mode=tase_mode)
pdf_generator.csv_to_pdf(
csv_filename=recommendation_list_filename_ssss,
csv_db_path=None,
data_time_str=recommendation_list_filename_ssss.replace(
'Results/', '')[0:15],
title=TITLES[scan_mode].replace('_', ' ') + 'ssss',
limit_num_rows=PDF_NUM_ENTRIES_IN_REPORT,
diff_list=diff_lists[0],
tase_mode=tase_mode)
pdf_generator.csv_to_pdf(
csv_filename=recommendation_list_filename_sssss,
csv_db_path=None,
data_time_str=recommendation_list_filename_sssss.replace(
'Results/', '')[0:15],
title=TITLES[scan_mode].replace('_', ' ') + 'sssss',
limit_num_rows=PDF_NUM_ENTRIES_IN_REPORT,
diff_list=diff_lists[0],
tase_mode=tase_mode)
research_mode = True # Research Mode
if not research_mode: # Run Build DB Only:
if run_custom_tase:
sss.sss_run(sectors_list=[],
sectors_filter_out=0,
countries_list=[],
countries_filter_out=0,
build_csv_db_only=1,
build_csv_db=1,
csv_db_path='None',
read_united_states_input_symbols=0,
tase_mode=1,
num_threads=1,
market_cap_included=1,
use_investpy=0,
research_mode=0,
profit_margin_limit=0.0001,
ev_to_cfo_ratio_limit=20000.0,
debt_to_equity_limit=1000.0,
min_enterprise_value_millions_usd=5,
price_to_earnings_limit=9000,
enterprise_value_to_revenue_limit=1500,
favor_sectors=['Technology', 'Financial Services'],
favor_sectors_by=[4.0, 0.75],
generate_result_folders=1,
custom_portfolio=['IGLD-M.TA'])
if run_custom:
sss.sss_run(
sectors_list=[],
sectors_filter_out=0,