Machine Learning with Elixir and Phoenix
summary
Claims a homegrown Elixir neural network can predict 20‑day returns on S&P 500 (SPY) data and be deployed in a Phoenix app. Uses Python to prep data, then trains an Elixir neural network on 2015 data with inputs SMA Bollinger Bands and RSI and tests on 2016. Portfolio simulation starts $100k, yields ~+3% in 2015 and ~+16% in 2016 versus SPY −0.7% and +10%, so the system works but needs more indicators and libraries.
AI-written summary · about
chapters (19)
- 0:00Intro
- 0:29Welcome
- 1:51Star Wars Day
- 3:42Overview
- 4:58Machine Learning is generalization
- 6:14Supervised Learning
- 8:46Data Sources
- 12:39Why Neural Networks
- 15:43Neural Network Visualization
- 19:59Disclaimer
- 21:02Finance jargon
- 23:49Bollinger Bands and RSI
- 26:55Phoenix
- 28:17Neural Network
- 31:21Live Demo
- 33:10Charts
- 34:43Portfolio Value
- 39:35Summary
- 40:27Outro
related talks
30mPredictive trading
25mApplied Machine Learning in Elixir
01h 26mImplementing a highly scalable Stock prediction system with R, GemFire and Spring XD
23mComplex AI Forecasting Methods for Investments
47mQuantitative Trading with Machine Learning
30mHow to build a stock market money printing machine with AI
32mLearn Elixir: Building a Neural Network from Scratch
30mApplying The Magic Of Neural Networks