drakkar-software / octobot Goto Github PK

The Data update takes on more sec after each refresh_data of TimeFrameUpdateDataThread see log :

2018-04-05 19:26:46,013 TA THREAD - BTCUSDT - BinanceExchange - TimeFrames.ONE_MINUTE INFO     Notified by TimeFrameUpdateDataThread
2018-04-05 19:27:47,079 TA THREAD - BTCUSDT - BinanceExchange - TimeFrames.ONE_MINUTE INFO     Notified by TimeFrameUpdateDataThread
2018-04-05 19:28:48,505 TA THREAD - BTCUSDT - BinanceExchange - TimeFrames.ONE_MINUTE INFO     Notified by TimeFrameUpdateDataThread
2018-04-05 19:29:49,621 TA THREAD - BTCUSDT - BinanceExchange - TimeFrames.ONE_MINUTE INFO     Notified by TimeFrameUpdateDataThread
2018-04-05 19:30:51,038 TA THREAD - BTCUSDT - BinanceExchange - TimeFrames.ONE_MINUTE INFO     Notified by TimeFrameUpdateDataThread

• Force an indicator to not keep its initialized eval note
• Allow Strategies to check if the eval_note was updated --> if not don't use this indicator eval note

Ensure that self.is_evaluating end with False --> use try catch finally

Define parent exchange class architecture :

• Get historical data from exchange (multiple time frames)
• Get order book

Create one TA evaluation to provide :

• an example for other TA evaluation
• a start point for the Evaluation process architecture

Trader simulator --> same features as Trader with evaluators
--> register trade / balance / portfolio results

Define tool to measure CPU and Mem usage :

• When threads are active

Add TA implementations and evaluations :
RSI...

Portfolio trader --> manage a cryptocurrency portfolio with evaluators
Portfolio trader simulator --> simulate the management of a cryptocurrency portfolio with evaluators

In database:

• timetag
• previous evaluation
• data:
  --- new: all
  --- twitter: all(followed), filtered(#XXX)
  --- search results: all

Define CI architecture :

• Libs to test (python)
• Tools for CI
• Backtesting (Crypto-Bot)

Create strategies indicators :

• Evaluator Class
• SubClasses

Add delayed thread start for long time frames :

• Limit APIs call at startup

Github Wiki page
--> Installation

Add rules system skeleton with one concrete implementation (ex: twitter with TrendTAAnalysis)

--> Candle evaluator :

• Notify()

Implement a notification system to trigger TA when Social notifies global evaluator

Manage indicator pertinence

Sometimes at startup the "new state" appears 2 times with only TA indicators.

Normal : only one "new state" is possible because all TA indicators has received almost 1 time new data and has been re-evaluated.

by enum in cst.py ?

--> When an eval type is None return 0