Release 0.23

New:

• Settings moved as far as possible from settings.py to the Django admin panel
• Scan can be started from the home page
• Data download can be started from the home page

Release 0.22

This version is motivated by two changes:

• Upgrade of vbt and Keras

  • The function start in the background disappeared from telegram bot combined with vbt. As the bot uses telegram only to inform the user and not for taking commands as input, the code was simplified.

• YFfinance has become unusable because of error 429.

  • IB should be used as data source. This part of the code was improved.

• Ib_insync is obsolete and replaced by ib_async.

• Docker handling was removed as the bot is too heavy for it. The code needs anyhow to be customized on your end depending on your strategies. It is not a ready to use solution.

Note: a new version will quickly follow.

Release 0.21

vectorbtpro updated to v2023.12.23

Generally, further work on the ml function, to make more robust.

New features:

• reduce_memory_usage argument for the definition of LSTM (reduce the memory usage, but is significantly slower)
• by optimization introduction of proba_one to determine the initial proportion of 1 and 0 in the strat array.

Bug corrections:

• Slow performance of divergence
• Bug with cand_to_quantity relative to past quantity
• Handling of the ISLAND stock exchange which cause "pumping" by StockStatus actualization
• Number of trades filtered during optimization corrected
• Correction of the reverse attribute determination in the get_delta_size function
• Bug with editing of stocks attribute

Note: for some reason, that I don't understand, Bbands returns a different bandwidth than previously even though this function is not customized in the code.

Release 0.20

New features:

• Strategy array written in an human-readable way
• New patterns and signals added
• Removal of useless rank methods, replaced by pandas native one.
• Removal of VBTNATR as it is calculated directly by talib
• Dump size reduced
• Trades always routed to SMART
• Simplified telegram messages
• Implementation of MLP and LSTM as machine learning models using keras (beta)

Clean-up:

• Only IB and YF are supported as I have no meant to test with other platforms.

Bug correction:

• Underlying strategy could not be used directly.
• Issue trading indexes especially in ss_manager function
• Issue with ETF order direction
• By optimization an abs() was missing by filtering extremal values

Present backlog:

• Improve and put in production the machine learning models
• Introduce telegram commands

Release 0.10

Initial release