def market_pipeline(model, market_specs):
r"""AlphaPy MarketFlow Pipeline
Parameters
----------
model : alphapy.Model
The model object for AlphaPy.
market_specs : dict
The specifications for controlling the MarketFlow pipeline.
Returns
-------
model : alphapy.Model
The final results are stored in the model object.
Notes
-----
(1) Define a group.
(2) Get the market data.
(3) Apply system features.
(4) Create an analysis.
(5) Run the analysis, which calls AlphaPy.
"""
logger.info("Running MarketFlow Pipeline")
# Get model specifications
predict_mode = model.specs['predict_mode']
target = model.specs['target']
# Get market specifications
create_model = market_specs['create_model']
data_history = market_specs['data_history']
features = market_specs['features']
forecast_period = market_specs['forecast_period']
fractal = market_specs['fractal']
functions = market_specs['functions']
lag_period = market_specs['lag_period']
leaders = market_specs['leaders']
predict_history = market_specs['predict_history']
target_group = market_specs['target_group']
# Set the target group
group = Group.groups[target_group]
logger.info("All Symbols: %s", group.members)
# Determine whether or not this is an intraday analysis.
intraday = any(substring in fractal for substring in PD_INTRADAY_OFFSETS)
# Get stock data. If we can't get all the data, then
# predict_history resets to the actual history obtained.
lookback = predict_history if predict_mode else data_history
npoints = get_market_data(model, market_specs, group, lookback, intraday)
if npoints > 0:
logger.info("Number of Data Points: %d", npoints)
else:
raise ValueError("Could not get market data from source")
# Run an analysis to create the model
if create_model:
logger.info("Creating Model")
# apply features to all of the frames
vmapply(group, features, functions)
vmapply(group, [target], functions)
# run the analysis, including the model pipeline
a = Analysis(model, group)
run_analysis(a, lag_period, forecast_period, leaders, predict_history)
else:
logger.info("No Model (System Only)")
# Run a system
system_specs = market_specs['system']
if system_specs:
# get the system specs
system_name = system_specs['name']
longentry = system_specs['longentry']
shortentry = system_specs['shortentry']
longexit = system_specs['longexit']
shortexit = system_specs['shortexit']
holdperiod = system_specs['holdperiod']
scale = system_specs['scale']
logger.info("Running System %s", system_name)
logger.info("Long Entry : %s", longentry)
logger.info("Short Entry : %s", shortentry)
logger.info("Long Exit : %s", longexit)
logger.info("Short Exit : %s", shortexit)
logger.info("Hold Period : %d", holdperiod)
logger.info("Scale : %r", scale)
# create and run the system
system = System(system_name, longentry, shortentry, longexit,
shortexit, holdperiod, scale)
tfs = run_system(model, system, group, intraday)
# generate a portfolio
gen_portfolio(model, system_name, group, tfs)
# Return the completed model
return model
def market_pipeline(model, market_specs):
r"""AlphaPy MarketFlow Pipeline
Parameters
----------
model : alphapy.Model
The model object for AlphaPy.
market_specs : dict
The specifications for controlling the MarketFlow pipeline.
Returns
-------
model : alphapy.Model
The final results are stored in the model object.
Notes
-----
(1) Define a group.
(2) Get the market data.
(3) Apply system features.
(4) Create an analysis.
(5) Run the analysis, which calls AlphaPy.
"""
logger.info("Running MarketFlow Pipeline")
# Get any model specifications
predict_mode = model.specs['predict_mode']
target = model.specs['target']
# Get any market specifications
data_history = market_specs['data_history']
features = market_specs['features']
forecast_period = market_specs['forecast_period']
functions = market_specs['functions']
lag_period = market_specs['lag_period']
leaders = market_specs['leaders']
predict_history = market_specs['predict_history']
resample_data = market_specs['resample_data']
target_group = market_specs['target_group']
# Get the system specifications
system_specs = market_specs['system']
if system_specs:
system_name = system_specs['name']
try:
longshort = True
longentry = system_specs['longentry']
shortentry = system_specs['shortentry']
longexit = system_specs['longexit']
shortexit = system_specs['shortexit']
holdperiod = system_specs['holdperiod']
scale = system_specs['scale']
logger.info("Running Long/Short System %s", system_name)
except:
longshort = False
system_params = system_specs['params']
logger.info("Running System %s", system_name)
# Set the target group
group = Group.groups[target_group]
logger.info("All Members: %s", group.members)
# Get stock data. If we can't get all the data, then
# predict_history resets to the actual history obtained.
lookback = predict_history if predict_mode else data_history
new_history = get_market_data(model, group, lookback, resample_data)
if new_history < data_history:
logger.info("Maximum Data History is %d, not %d",
new_history, data_history)
if new_history == 0:
raise ValueError("Could not get market data from source")
# Apply the features to all of the frames
vmapply(group, features, functions)
vmapply(group, [target], functions)
# Run a system or an analysis
if system_specs:
# create and run the system
if longshort:
system_ls = System(system_name, longentry, shortentry,
longexit, shortexit, holdperiod, scale)
tfs = run_system(model, system_ls, group)
else:
tfs = run_system(model, system_name, group, system_params)
# generate a portfolio
gen_portfolio(model, system_name, group, tfs)
else:
# run the analysis, including the model pipeline
a = Analysis(model, group)
results = run_analysis(a, lag_period, forecast_period, leaders,
predict_history)
# Return the completed model
return model