The Precision Alpha datasets apply machine learning to historical stock prices to predict next-day movements, using seven proprietary metrics including probability indicators, market energy, and resistance.
Precision Alpha Price Dynamics and Price Predictor Datasets
Overview
These two datasets (NYSE and NASDAQ) use machine learning to process historical closing prices to predict stock price movements for the next trading day.
NYSE Precision Alpha Dataset - Daily data starting from March 16, 2016
NASDAQ Precision Alpha Dataset - Daily data starting from October 18, 2019Key Variables/Metrics
The datasets use non-equilibrium machine learning to calculate several predictive measurements, including:
- Next Day Probability Up: Probability that the stock price will increase the next trading day
- Market Energy: Measurement of potential price movement energy
- Market Power: Force driving price movement in dominant direction
- Market Resistance: Entropic force resisting change to the dominant price direction
- Market Noise: Diffusion that dissipates market energy (similar to viscosity)
- Market Temperature: Entropic temperature as defined in thermodynamics
- Market Free Energy (Helmholtz): Energy available to create price movement
Methodology
The technology uses non-equilibrium signal analysis to expose market patterns and structural breaks that typical traders might miss. According to Precision Alpha, their approach employs probabilistic mathematics, information theory and machine learning to reveal price moves before they occur
Delivery Timing
The data is delivered typically about 5 hours after market close.