- Sample Model and Config running:
freqtrade trade --strategy EmaEngAI --strategy-path freqtrade/templates --config config_examples/config_freqai.example.json --freqaimodel GridRegressionDTMultiTarget --freqaimodel-path freqtrade/freqai/prediction_models/
- SVM:
- Don't use SVM.I've tested before ; RMSE was very high.
- Pytorch:
- DO NOT CREATE ANOTHER CLASS OR FILE AT freqai/torch PASS. IT'S ONLY ACCEPT ONE FILE AS A MODEL WITH THE EXACT NAME FOR CLASS AND FILE PyTorchMLPModel
- download the data:
freqtrade download-data --exchange binance --timerange 20230420-20230820 --timeframes 5m --trading-mode futures --config config_examples/config_freqai.example.json
- run backtesting:
freqtrade backtesting -s EmaEngAI --strategy-path freqtrade/templates --config config_examples/config_freqai.example.json --timerange 20230601-20230701 --freqaimodel GridRegressionDTMultiTarget --freqaimodel-path freqtrade/freqai/prediction_models/
- download data:
freqtrade download-data --exchange binance --timeframes 3m --timerange 20230801-20230904 --pairs BTC/USDT:USDT ETH/USDT:USDT XRP/USDT:USDT --erase --trading-mode futures
- run optimazer:
freqtrade hyperopt -s EmaEngAI --strategy-path freqtrade/templates --config config_examples/config_freqai.example.json --hyperopt-loss OnlyProfitHyperOptLoss -e 40 --timerange 20230501-20230601 --freqaimodel GridRegressionDTMultiTarget --freqaimodel-path freqtrade/freqai/prediction_models/
- download data:
freqtrade download-data --exchange binance --timeframes 1w --timerange 20230601-20230701 --trading-mode futures --config config_examples/config_freqai.example.json
- Run thic command:
freqtrade plot-dataframe --strategy BBvwap --strategy-path freqtrade/templates --userdir user_data/ -c config_examples/config_freqai.example.json --timerange 20230601-20230701 --pairs BTC/USDT:USDT
-
Be mindful of the
train_period_days
andbacktest_period_days
settings when running backtests and hyperparameter optimizations in FreqAI. The sum oftrain_period_days
andbacktest_period_days
must not exceed the specified time range. -
Remove all - pycache - file before backtesting and hyperopt:
find . -type d -name "__pycache__" -exec rm -r {} \; ; rm -rf user_data/* ; rm -rf trad*