[Machine Learning for Trading] {ud501} Lesson 21: 03-01 How Machine Learning is used at a hedge fund | Lesson 22: 03-02 Regression

a data-centric way to build predictive models

 

 

 

The ML problem

 

 

Supervised regression learning 

 

 

 

Robot navigation example 

 

 

 

 

 

 How it works with stock data

 

 

 

 

 

 Example at a fintech company

 

 

 

 

Price forecasting demo 

 

QuantDesk

factors we are using now <= choices of these factors are from another genetic algorithm 

  

 <= roll back time, and we look over all this last three months and look forward one month, see how accurate all those predictions were

 

 

https://lucenaresearch.com/#register

https://quantdesk.lucenaresearch.com/#login

 

 

Backtesting 

 

 

 

 

 ML tool in use

 

orange line => historical value of our portfolio

blue => benchmark (S&P500 here)

 

 

 

 

 

 Problems with regression

 

 

Problem we will focus on 

 

 

 

---

---

---

---

 

 

 

 

Parametric regression 

 

 

 

 K nearest neighbor

 

 

 

 

 

 

 How to predict

 

 

 

 

 

 Kernel regression

Kernel regression is different from KNN, because it uses kernel to weight the contribution of each nearest point

 

 

 

 

 Parametric vs non parametric

 

Yes, the cannon ball distance can be best estimated using a parametric model, as it follows a well-defined trajectory.

On the other hand, the behavior of honey bees can be hard to model mathematically. Therefore, a non-parametric approach would be more suitable.

 

 

 

 Training and testing

 

typically: train on older data; test on newer data

look ahead bias occurs if training reversely

 

 

 

Learning APIs 

 

 

 

 

 Example for linear regression

 

 Note: This is intended to be pseudo-code only, although some Python-specific syntax has been shown.