%OP	Percent of Over-Predictions
1 d, 1 h, 15 min	1 day, 1 h, 15 minutes
1st Qu., 3rd Qu.	1st quantile, 3rd quantile
3M, 6M	3 months, 6 months
AIS	algorithmic investment strategies
ARC	Annualized Return Compounded
ARIMA	Autoregressive Moving Average
ARIMAX	Autoregressive Moving Average with exogenous variables
ASD	Annualized Standard Deviation
AT	Attention based model
B&H	buy&hold strategy
bs040, bs080, bs160	batch size of 40, 80 and 160 observations, respectively
BTC	bitcoin
CEC	Constant Error Carousel
CET	Central European Time
dr001, dr002, dr004	dropout rate of 1%, 2% and 4%, respectively
GPU	Graphics Processing Unit
GRU	Gated Recurrent Unit
IR*	Information Ratio
IR**	Modified Information Ratio
IR***	Aggregated Information Ratio
KNN	K-Nearest Neighbours algorithm
Kurt.	kurtosis coefficient
LightGBM	Light Gradient Boosting algorithm
LS, LO	Long/Short strategy, Long only strategy
LSTM	Long-Short Term Memory
MADL	Mean Absolute Directional Loss
MAE	Mean Absolute Error
MAPE	Mean Absolute Percentage Error
MD	Maximum Drawdown
MLD	Maximum Loss Duration
MSE	Mean Square Error
nObs	number of observations
Norm.	Pearson chi-square normality test p-value
nTrades	number of trades
OHLC	Open High Low Close
RB	Rebalancing period
RH	research hypothesis
RMSE	Root Mean Square Error
RNN	Recurrent Neural Network
seq07, seq14, seq28	sequence length of 7, 14 and 28 observations, respectively
SGD	Stochastic Gradient Descent algorithm
Skew.	skewness coefficient
SVR	Support Vector Regression
TI	Technical indicators
tr0685, tr1371, tr2742	training set length (size) of 685, 1371 and 2742 observations, respectively
W10%, W20%	weight of 10%, weight of 20%
XGBoost	Extreme Gradient Boosting algorithm