AI Factor Tutorial: All Settings to Build a Machine Learning Stock Ranking Model | Portfolio123

In this tutorial, we walk through the full AI Factor setup process inside Portfolio123, including target selection, universe setup, normalization, predefined factors, custom formulas, data loading, validation, model training, lift charts, return buckets, predictor creation, and how to use AI Factor inside screens and strategies.

AI Factor is Portfolio123’s no-code machine learning tool for systematic investors who want to build stock ranking models using financial factors, custom formulas, and machine learning algorithms.

This video is especially useful for new Portfolio123 subscribers who want a practical, current walkthrough of the AI Factor workflow and a clearer understanding of what each major setting does.

You’ll learn how to:
- Set up a new AI Factor
- Choose a target for machine learning stock prediction
- Select a stock universe and benchmark
- Understand rank normalization and Z-score normalization
- Add predefined factors and custom formulas
- Load and review AI Factor data
- Use validation to reduce overfitting risk
- Compare machine learning model options
- Interpret lift charts and return buckets
- Create an AI Factor predictor
- Use AI Factor in Portfolio123 screens and strategies

The goal of this walkthrough is not just to show where to click, but to help new subscribers understand how the AI Factor workflow fits into systematic investing, factor research, stock screening, ranking systems, and live strategy development.

The video also introduces two new ready-to-use AI Factor models:
- AI Factor S&P 500
- AI Factor S&P 1500.
These public models allow subscribers to explore machine learning stock selection without having to build an AI Factor from scratch. The S&P 500 version focuses on large-cap stocks, while the S&P 1500 version expands the opportunity set across large-, mid-, and small-cap stocks. Both models show how AI Factor predictors can be applied directly inside live Portfolio123 strategies, giving users a practical starting point before building and testing their own machine learning stock ranking models.

Video Chapter Timestamps:
00:00 Introduction to AI Factor
00:30 Setting Targets and Training Universes
01:08 Target Normalization and Scaling Settings (Z-Score vs. Rank)
02:29 Period Settings and Frequency Configuration
03:14 Max Return Safety Limits
03:49 Adding Predefined Features and Custom Formulas
05:48 Importing Existing Ranking Systems
06:16 Loading Data and Two-Step Normalization
07:01 Feature Stats and Regression Audit
07:25 Validation Methods (Basic Holdout vs. Cross-Validation)
10:31 Choosing Machine Learning Algorithms and Grids
13:44 Launching and Running Model Validation
14:18 Results Analysis: Quantiles, Slippage, and Lift Charts
15:37 Portfolio Performance and Alpha Tracking
16:42 Running Out-of-Sample Final Testing
17:35 Creating Historical and Production Ready Predictors
18:29 Backtesting via Stock Screener & Feature Importance
19:40 Generating Live Stock Recommendations
20:00 Premade Models & Live Strategy Deployment (S&P 500 / 1500)

Portfolio123 helps investors build, test, and deploy quantitative stock strategies using screeners, ranking systems, simulations, backtests, live strategies, and AI-powered factor models.

Видео AI Factor Tutorial: All Settings to Build a Machine Learning Stock Ranking Model | Portfolio123 канала Portfolio123