You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
very good curated list of notebooks showing deep learning + reinforcement learning models. Also contain topics on outlier detections/overbought oversold study/monte carlo simulartions/sentiment analysis from text (text storage/parsing is not detailed but it mentioned using BERT)
started by Columbia university engineering students and designed as an end to end deep reinforcement learning library for automated trading platform. Implementation of DQN DDQN DDPG etc using PyTorch and gym use pyfolio for showing backtesting stats. Big contributions on Proximal Policy Optimization (PPO) advantage actor critic (A2C) and Deep Deterministic Policy Gradient (DDPG) agents for trading
predecessor to tensortrade uses open api gym and neat way to render matplotlib plots in real time. Also explains LSTM/data stationarity/Bayesian optimization using Optuna etc.
implementation of deep reinforcement learning and supervised learnings covering areas: deep deterministic policy gradient (DDPG) and DDQN etc. Data are being pulled from rqalpha which is a python backtest engine and have a nice docker image to run training/testing
curated list of papers/repos on topics like CNN/LSTM/GAN/Reinforcement Learning etc. Categorized as deep learning for now but there are other topics here. Manually maintained by cbailes
Retrieve limit order book level data from coinbase pro and bitfinex -> record in arctic timeseries database then implemented trend following strategies (market orders) and market making (limit orders). Uses reinforcement learning (DQN) keras-rl to create agents and uses openai gym to implement POMDP (partially observable markov decision process)
repo for book hands-on-machine learning for algorithmic trading covering topic from data/unsupervised learning/NPL/RNN & CNN/reinforcement learning etc. Leverage zipline/alphalens/sklearn/openai-gym etc as well. Good references to have
docker based platfrom for developing algo trading strategies. Very interesting combinations of open source components were used including backtrader for backtest strategies / mlflow for managing the machine learning model life cycle (i.e. training and developing machine learning models) / airflow used as workflow management including schedule data download etc. / superset web data visualization tool similar to tableau / minio for fast object storage (i.e. storing saved models and model artifacts) / postgresql used to store security master and daily and minute data. Also contains some details on deployment on cloud
machine learning framework built on sklearn and pandas. Support pyfolio/xgboost/lightgmb/catboost(gradient boosting on decision tress) etc. Examples include financial market prediction/sports prediction/kaggle. Configurations are set though yaml file for all model process including feature selection/grid search on parameters and aggregate results for each model
accompanying materials for book Machine Learning and Data Science Blueprints for Finance on top of basic machine learning models i.e. nlp/reinforcement learning/supervised & unsupervised learning it covers wider topics including robo-advisors/fraud detection/loan default/derivative pricing/yield curve construction.
using fundamental and pricing data to predict future stock returns. Sklearn's randomforest classifier is trainded and author claimed positive live trading results. Not actively mainained
Research in investment finance for long term forecasts and a curated list of notebooks. Each topic contains a youtube video explaining in details. Interesting topics including using price per book ratio and other multiples for future return prediction and portfolio optimization. data sourced form simfin yahoo finance and s&p 500 earnings and estimate report etc.
lstm model using keras to predict msft prices. Data is from alphavantage which provides some free data through web services. Showing how to use concatenation layer to join timeseries data with TA data. Might be abit of overfitting on the model though
accompanying materials for book Machine Learning in Finance covering probabilistic modeling/sequence modeling/neural networks/reinforcement learning etc.
base framework trading bot for crypto. Stores data in local mongodb instance and supports backtest and live trading on poloniex and bittrex which are 12-15th ranked crypto exchanges by volume. Leverage talib for ta data and plotly for visualization
data processing platform which stream data from kafka. The example shows two incoming data stream stock vs tweets and two spark streams are created to consume the kafka data then end results are stored in cassandra. Older tech stacks were used and not actively maintained.
CryptoNets is a demonstration of the use of Neural-Networks over data encrypted with Homomorphic Encryption. Homomorphic Encryptions allow performing operations such as addition and multiplication over data while it is encrypted.
