issues
search
kkuette
/
TradzQAI
Trading environnement for RL agents, backtesting and training.
Apache License 2.0
164
stars
47
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Bump numpy from 1.14.5 to 1.22.0
#24
dependabot[bot]
opened
2 years ago
0
base_env.py check_time_before_closing() idx not defined
#23
AlconDivino
opened
5 years ago
2
Using other exchanges
#22
AlconDivino
closed
5 years ago
1
Error
#21
ghost
opened
5 years ago
3
Usage
#20
ghost
opened
5 years ago
2
live session error
#19
dgiunchi
opened
5 years ago
1
We cant afford any contract.
#18
dropcunt
closed
5 years ago
3
complex network "names" issue
#17
LucaKern
closed
5 years ago
1
Training on GPU
#16
dropcunt
closed
5 years ago
2
Update requirements
#15
QuentinDutot
closed
5 years ago
0
command 'wc -1 data/BTC_EUR_2018_07/BTC_EUR_2018_07_01.csv' returned a non-zero exit status 1
#14
greg2paris
closed
5 years ago
5
Possible bug in dataloader.py
#13
oscarberonius
opened
5 years ago
5
Daily trading
#12
gmomb
opened
5 years ago
2
Exception when running live version
#11
oscarberonius
closed
5 years ago
3
Choosing different agent than PPO produces faulty behavior.
#10
oscarberonius
closed
5 years ago
2
Performing training automatically evaluates the model.
#9
oscarberonius
closed
5 years ago
1
Cannot modify training data
#8
oscarberonius
closed
5 years ago
1
PPO implementation
#7
talvasconcelos
closed
5 years ago
1
How to change the agent type?
#6
lamhk
closed
5 years ago
6
How to specify which trained model to be used for "eval"
#5
lamhk
closed
5 years ago
1
We cant afford any contract. Please check wallet settings.
#4
lamhk
closed
5 years ago
9
Dev
#3
kkuette
closed
5 years ago
0
how to perform test using the trained model
#2
lamhk
closed
5 years ago
7
h_len is zero
#1
lamhk
closed
5 years ago
1