Neural Networks for Quant Finance
🚀 Master Quantitative Skills with Quant Guild
https://quantguild.com
📅 Meet with me 1:1
https://calendly.com/quantguild-support
📈 Interactive Brokers for Algorithmic Trading
https://www.interactivebrokers.com/mk...
👾 Join the Quant Guild Discord server here
/ discord
___________________________________________
🪐 Jupyter Notebook
https://github.com/romanmichaelpaoluc...
*TL;DW Executive Summary*
Data is the oil of the 21st century, models require data to learn anything
All models learn by minimizing the distance between some prediction and some observed target of interest, other features available before the event we are trying to predict are used to help inform the model's prediction
We can always overfit or underfit a model, we use training/validation/testing splits to ensure our model is learning in a robust manner while we can also emulate out-of-sample performance metrics on the unseen testing set
Our models are only as effective as the stability of the space, we are effectively learning expectation functions, in spaces that are random (fixed population distribution) or reasonably stable (think like facial recognition) we can train a model once and it can be sufficient for a good while, when population distributions are not stable and exhibit severe non-stationarity like stock prices or returns we can't apply these models here outright
Neural networks are a popular model to effectively learn these expectations as they are a universal function approximator
Neural networks are just functions that are parameterized by weight matrices, the optimal matrices are learned through the training process
There are a variety of models in this context from generative models to learn some target, even time variant, n-dimensional distribution, to non-linear dimensionality reduction, to image classification and language modeling
In quantitative finance we implement these models for a variety of regression and classification tasks, we saw a simple example above learning the Black-Scholes functional but we can apply this methodology to more complicated models to enhance calibration efficiency, manage portfolio risk, and even generate quant trading signals
I hope you enjoyed!
Roman
___________________________________________
📖 Chapters:
00:00 - Quants Build Models
03:25 - General Learning Model Framework
05:40 - Example: Max Throw Distance
08:56 - How Machines Learn
11:07 - Visualizing Machine Learning
14:55 - Underfitting, Overfitting, and Robust Models
18:16 - How to Train a Robust Model
21:21 - Why Predictions Don't Exist
25:21 - Why Models Can't Learn Stock Prices
29:33 - Introduction to Neural Networks
32:39 - Universal Approximation Theorem
35:28 - Moving from Theory to Practice
39:45 - Applications in Quant Finance
42:13 - Teaching a Neural Network Black-Scholes
45:38 - TL;DW Executive Summary
___________________________________________
🗣️ Shout Outs
A special thank you to my members on YouTube for supporting my channel and enabling me to continue to create videos just like this one!
⭐ Quant Guild Directors
Dr. Jason Pirozzolo
___________________________________________
▶️ Related Videos
Quant Builds 🔨
How to Build a Volatility Trading Dashboard in Python with Interactive Brokers
• How to Build a Volatility Trading Dashboar...
Statistics and Trading Profitability Over Time (Edge) 📈
Expected Stock Returns Don't Exist
• Expected Stock Returns Don't Exist
How to Trade
• How to Trade
How to Trade Option Implied Volatility
• How to Trade Option Implied Volatility
How to Trade with an Edge
• How to Trade with an Edge
Quant on Trading and Investing
• Quant on Trading and Investing
___________________________________________
🗂️ Resources
📚 Quant Guild Library:
https://github.com/romanmichaelpaoluc...
🌎 GitHub:
https://github.com/RomanMichaelPaolucci
https://github.com/Quant-Guild
📝 Medium (Blog):
/ quantguild
/ quant
___________________________________________
🛠️ Projects
The Gaussian Cookbook:
https://gaussiancookbook.com
Recipes for simulating stochastic processes:
https://papers.ssrn.com/sol3/papers.c...
___________________________________________
💬 Socials
TikTok: / quantguild
Instagram: / quantguild
X/Twitter: https://x.com/quantguild/
LinkedIn (personal): / rmp99
LinkedIn (company): / quant-guild
___________________________________________