Forecasting the spread
Budget: ₹750 – ₹1,250 INR
Requirements
Intuitive understanding of machine learning algorithms, supervised and unsupervised modelling techniques and their performance characteristics. Experience working with forex data sets, or price feeds. Experience with machine learning tools and libraries such as TensorFlow, Scikit-learn, R, Spark, PyTorch.Experience with ML in manufacturing and operations spaces are pluses. In particular, working with extensive and messy manufacturing data set, as well as deploying production ready models to the factory. Experience in Algo trading is an advantage.
Project description
Spread is the ratio between two price feeds, and prediction of the spread is crucial for pair trading. Prediction of the same can help in identifying the distance of spread to Moving average , 2 standard deviation and 3 standard deviation. Models to be developed and tested for the data are listed below.
Parametric model - Autoregressive moving average(ARMA)
Non parametric - LSTM and LSTM encoder or decoder
Recurrent Neural Networks (RNN) to generate a one-week ahead forecast
Intuitive understanding of machine learning algorithms, supervised and unsupervised modelling techniques and their performance characteristics. Experience working with forex data sets, or price feeds. Experience with machine learning tools and libraries such as TensorFlow, Scikit-learn, R, Spark, PyTorch.Experience with ML in manufacturing and operations spaces are pluses. In particular, working with extensive and messy manufacturing data set, as well as deploying production ready models to the factory. Experience in Algo trading is an advantage.
Project description
Spread is the ratio between two price feeds, and prediction of the spread is crucial for pair trading. Prediction of the same can help in identifying the distance of spread to Moving average , 2 standard deviation and 3 standard deviation. Models to be developed and tested for the data are listed below.
Parametric model - Autoregressive moving average(ARMA)
Non parametric - LSTM and LSTM encoder or decoder
Recurrent Neural Networks (RNN) to generate a one-week ahead forecast