Corrective AI in AlgoTrading

Description

My project facilitates application of Corrective AI to existing trading strategy. Strategy (Moving Average) sends signal to buy/sell and XGBoost predict the propability if this particular trade is profitable. Data is downloaded from Oanda via tpqoa liblary. Inspired by dr. Ernest Chan.

Results

XGBoost shows over 2% higher profit
Improved Sharpe Ratio
Improved Calmar

Languages and Utilities Used

• Python
• Tensorflow
• XGBoost
• Oanda

Program walk-through:

1. Set your access api in oanda.cfg
Fill you api credentials in oand.cfg file

[oanda]
account_id = 
access_token = 
account_type = practice

2. Download the data.

Specify wanted pairs and target that you want to trade.

symbols = ['AUD_CAD', 'BCO_USD'] # Used currency pair
env = prfenv.ProfundaEnv(symbols = symbols,target = 'AUD_CAD', start_date = '2021-12-20 00:00:00', end_date = '2023-12-20 00:00:00', freq='M5', api='oanda.cfg')

3. Results.