Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

FreqAI: model not retraining itself. #10213

Closed
campera88 opened this issue May 15, 2024 · 3 comments
Closed

FreqAI: model not retraining itself. #10213

campera88 opened this issue May 15, 2024 · 3 comments
Assignees
Labels
freqAI Issues and PR's related to freqAI Question Questions - will be closed after some period of inactivity.

Comments

@campera88
Copy link

campera88 commented May 15, 2024

Describe your environment

  • Operating system: Ubuntu 20.04
  • Python Version: 3.9
  • CCXT version: 4.3.21
  • Freqtrade Version: 2024.5-dev-b1fd79d72

Your question

I developed a strategy for FreqAI and trained the model on a backtest to verify its functionality. After numerous attempts, it is now working properly.

Currently, the model is executing live trades as expected, but there is a small issue. The model is not retraining itself at all. The coin pairs' models in user_data/models/id are the same as those at the end of the backtest.

I even tried training a different model with a different ID and another configuration file, but nothing happened—it just performed a normal backtest. Although it created the folder user_data/models/newid, no further action occurred.

This is my config:

"freqai": {
        "enabled": true,
        "purge_old_models": 2,
        "train_period_days": 15,
        "backtest_period_days": 7,
        "live_retrain_hours": 0.5,
        "identifier": "id1",
        "save_backtest_models": true,
        "feature_parameters": {
            "include_timeframes": [
                "3m",
                "15m",
                "1h"
            ],
            "include_corr_pairlist": [
                "BTC/USDT",
                "ETH/USDT"
            ],
            "label_period_candles": 20,
            "include_shifted_candles": 2,
            "DI_threshold": 0.9,
            "weight_factor": 0.9,
            "principal_component_analysis": false,
            "use_SVM_to_remove_outliers": true,
            "indicator_periods_candles": [
                10,
                20
            ],
            "plot_feature_importances": 0
        },
        "data_split_parameters": {
            "test_size": 0.33,
            "random_state": 1
        },
        "model_training_parameters": {}
    },

Any idea? Thanks in advance.

@campera88 campera88 added the Question Questions - will be closed after some period of inactivity. label May 15, 2024
@xmatthias xmatthias added the freqAI Issues and PR's related to freqAI label May 16, 2024
@robcaulk
Copy link
Member

Please provide more information. Right now we are unable to assist since you are not providing any evidence for why you believe it’s not training or retraining.

Actually, your statements are just confusing, your final paragraph indicates that FreqAI is not training any models at all. But your title and first paragraph say it’s just not “retraining” but that it’s working.

Consider running the quick start command and reporting to us your observations plus logs. Otherwise, nothing we can do to help you.

https://www.freqtrade.io/en/latest/freqai/#quick-start

@campera88
Copy link
Author

Thank you very much for the response.

I did not explain myself well. The bot is operating and making trades in a live run for over 24 hours with the command: freqtrade trade --config user_data/myconfig.json --strategy Strategy --freqaimodel LightGBMRegressor

I know that FreqAI is not retraining any model because the bot is behaving like a normal bot and the files inside the directory of the id are from the last backtest I did, and it is set to train the models every 30 minutes.

Running it with the sample strategy and sample config works as expected.

I also tried my configuration with the FreqAI sample strategy, using a new ID, and the models are being trained. Could there be an issue with my strategy code that prevents FreqAI from retraining the models?

@robcaulk
Copy link
Member

Could there be an issue with my strategy code that prevents FreqAI from retraining the models?

There is no way to answer your question. You are asking us if there is a problem with your strategy, but not sharing your strategy.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
freqAI Issues and PR's related to freqAI Question Questions - will be closed after some period of inactivity.
Projects
None yet
Development

No branches or pull requests

3 participants