Want to be able to easily log trades of any type of security and measure performance and analyze several factors, such as risk or return distribution? Then check out trade-stat-logger.
- log trades: long and short
- get a statistical summary:
- net profit: self-explanatory
- drawdown: vertical distance between peak and trough of net profits
- volatility of returns: measures deviation of returns distribution
- kurtosis: measures tails of returns distribution
- Kelly Criterion: estimate of optimal portfolio allocation
- And more
- visualize performance:
- visualize distribution of returns
- analyze movement of gains or losses after every trade or over a period of time
pip install trade-stat-logger
SimpleLogger(datetime_support=False)
Constructs a new SimpleLogger object. Set datetime_support=True if you want to log the time at which the trade executed. Ignore future dt key word arguments if you wish to not support datetime logging.
log(security, shares, share_price, dt=None)
Log trades with this method. Shares is the number of shares you wish to purchase, set to negative to short/sell, and share_price is the price of a share. If dt is left as None, it will log current time, else set dt to a datetime object to log a custom time.
log_cp(security, share_price, dt=None)
Clear a position of given security at given share_price and log it. If dt is left as None, it will log current time, else set dt to a datetime object to log a custom time.
get_position(security)
Returns a tuple in the format (# shares, position size), position size is (# shares) * (average share price). It is slightly different from the format of get_positions.
get_positions()
Returns a dict of current security holdings in form of {security: {'shares': # shares, 'position_size': position size}, another_security: {'shares': # shares, 'position_size': position size}, ...}, where the key corresponding to a security is the string passed through the log() method.
clear_all_positions(get_price_func, closure_date)
Clears all security holdings, which is advised before calling methods that measure performance of trades. get_price_func should be a function such that get_price_func(security, closure_date) == price of security. For example:
logger = SimpleLogger()
...log trades...
def get_price(security, date):
return api.get_price(security, date)
logger.clear_all_positions(get_price, some_date)
However, if you logged the tickers of stocks, you can use an easier alternative: clear_stock_positions(closure_date) will clear all stock positions
get_summary_statistics()
Returns a dictionary of various statistics in the form of {statistic1_name: value, statistics2_name: value,...}.
graph_statistics(time_axis=False, time_strformat='%m/%d/%Y', show_window=True)
Graphs the distribution of trade returns, net profit at the nth trade, and puts the summary statistics in a table, and shows it in a matplotlib popup window. If you would like to have the x-axis for the net profit graph be time, set time_axis=True, and optionally add a custom time format through time_strformat='%Y/%o/%u/%r format'. If you wish to get the plot, figure, and axes, the method will return these if you set show_window=False, and this will also mean the popup window will not appear.
Use within a simple Moving Average Crossover strategy here
Use within a Time Series Forecasting with Machine Learning strategy (Jupyter NB) here
from datetime import timedelta
from random import randint
logger = SimpleLogger(datetime_support=True)
day_counter = 1
for x in range(0, 100):
logger.log('fried chicken futures', randint(1, 100), randint(1, 100), dt=datetime.now()-timedelta(days=day_counter))
logger.log('fried chicken futures', randint(-100, 1), randint(1, 130), dt=datetime.now()-timedelta(days=day_counter))
day_counter += 1
# by setting show_window to false, the graphs aren't plotted, but instead returned
# axes is 2x2
plt, fig, axes = logger.graph_statistics(show_window=False)
# alter as you like, such as axes[0][0].set_xlabel('asdf')
# get current positions in securities (which we will then clear)
print(logger.get_positions())
# example of how to use clear_all_positions(),
# replace get_price() with a real api call or any other price getter
fake_api_method = lambda security, dt: len(security) + dt.timetuple().tm_yday
get_price = lambda security, dt: fake_api_method(security, dt)
logger.clear_all_positions(get_price, datetime.today())
# shows the statistics in a graphs in a pop up window. since it calls plt.show()
# when show_window is not set to false, all code beyond this point won't be executed
logger.graph_statistics()
If you like my work and have any inquiries (library support or job related), feel free to contact me at shilewen@uw.edu.
- more time support for SimpleLogger, including time analysis on returns
- I will add a ComplexLogger, which will be built upon SimpleLogger, and it will support initial portfolio size, ROI, and many more.
- KellyLogger, which will dynamically return Kelly Criterion scores based on trade logs