Version 0.1.15

Added signal_weight method to NaiveOptimization, signals and exposure attributes to PortfolioOptimization, added rolling_mean, rolling_median, rolling_sharpe, rolling_sortino methods to ReturnEstimators class, alpha_ewma to trend factor methods, alpha and beta returns to metrics

Version 0.1.14

Added signal arg to breakout method in Trend class, refactored triple_ewma_diff method to convert span (window_size) to half-life, set lags param to 1 in compute_signal_returns method in Signal class.

Version 0.1.13

Created more intuitive interface for Signal class and new methods (combine and power transformation), added SupervisedLearning class, new transformation methods in Transform class.

Version 0.1.12

Version 0.1.12 includes updated portfolio risk measures like correlation surprise and magnitude surprise, in addition to turbulence (Kritzman & Li, 1999)

Version 0.1.10

Fixed bug in PortfolioSort. Modified Signal class dual signal computation.

Version 0.1.9

Added dual signal signals and heatmap plotting.

Version 0.1.8

Modified compute_portfolio_returns method to allow for both single and multiindex dataframes.

Version 0.1.7

New version 0.1.7 includes new modules: size, value, trend and carry factors. Supervised and unsupervised learning modules. Signal generation modules. Portfolio sorts and optimization.

Version 0.1.6

Version 0.1.6, removed sklearn daal package which was causing issues with upgrade.

Version 0.1.5

Bug fixes to version 0.1.4.